What have you been watching for this summer?

This month’s signal drop is about who ultimately holds the authority and responsibility in a world where this question itself is getting increasingly muddy to define.

Banks and insurers are creating jobs whose purpose is to supervise AI agents. R&D teams have adopted AI to do the a lot of work while leaving the decisions about which work to fund untouched. Nine in ten GPs who avoid AI name professional liability as their key concern. More than half of chief supply chain officers cannot say what return their AI spending has produced. And the companies capturing three-quarters of AI’s value are the ones automating the most decisions, having built the governance first.

Something is being renegotiated about where authority sits inside an organisation. Tools are going in at the task layer with very little friction. Everything above that layer, the judgement calls, the approvals, the name on the outcome, is where adoption slows down, projects stall, and firms start hiring.

The Org Chart Grows a Tier

Capgemini’s World Cloud Report for Financial Services 2026, based on a survey of 1,100 financial services leaders, found nearly half of banks and insurers creating new jobs to supervise AI agents.

The supervision hiring is running ahead of the deployment. Only 10% of firms have implemented AI agents at scale, with 80% still at ideation or pilot stage. Banks name customer service (75%), fraud detection (64%), loan processing (61%) and customer onboarding (59%) as the processes they intend to move to agents at scale. Insurers report a similar spread, led by customer service (70%), underwriting (68%) and claims processing (65%). Almost all executives flag two roadblocks: regulatory and compliance burden (96%) and a skills gap among leaders and employees (92%).

A technology sold on the promise of fewer people is generating a new supervisory job family in a heavily regulated sector, before most of the agents it would supervise have actually arrived.

Source: Capgemini Research Institute

Adoption Stops at the Door of the Decision

Wiley’s 2026 R&D Benchmark Report surveyed more than 200 senior R&D professionals across nine industries in North America, the UK and Europe. The report describes AI adoption as widespread across the function, while finding that respondents use it mainly for execution rather than for the strategic choices that determine which projects deserve investment at all.

The waste those choices produce is already sizeable. More than a third of organisations report spending 25 to 40% of their R&D budget on projects that never reach market, and nearly half estimate losing over $1 million each time a late-stage project is cancelled.

Worth naming the interest here: the report is presented by IEEE Spectrum and Wiley and sponsored by Patsnap, which sells innovation-intelligence software to R&D teams. The finding that firms lack decision-stage intelligence is a finding the sponsor benefits from.

The tooling has made the work faster without touching the question of which work was worth doing.

Source: Wiley

Where Liability Lands, Uptake Thins

The Nuffield Trust and Royal College of General Practitioners surveyed 2,108 UK GPs. Around 28% use AI tools in clinical work, and the use is heavily weighted toward documentation and note-taking (57% of users), professional development (45%) and admin (44%). Clinical decision support sits at 28%.

The drop-off has a named cause. Among non-users, 89% cite professional liability and medico-legal issues as a barrier. Among GPs already using AI, the figure is 80%. Familiarity does very little to settle the question of who answers for the output.

Uptake stays high while the tool is drafting, and thins out where a professional would have to stand behind what it produced.

Source: Nuffield Trust / RCGP

Spending Confidently, Measuring Vaguely

Gartner surveyed 394 supply chain professionals at organisations with revenue above $250 million, then 135 senior supply chain leaders. Two-thirds of supply chain digital investment now goes to AI. More than half of chief supply chain officers, 55%, cannot say what return it is producing.

Gartner’s prescription is telling. The fix it proposes is a change strategy, deciding where finite change-management capacity gets spent, and building leaders with the business and risk judgement to direct it. Gartner’s earlier and much-recycled forecast, that over 40% of agentic AI projects will be cancelled by the end of 2027, names the same three culprits: escalating costs, unclear business value, inadequate risk controls. That forecast was issued in mid-2025, though much of the commentary quoting it this year presents it as new.

Two-thirds of the digital budget is going somewhere more than half the leadership cannot yet score. That is a measurement problem sitting in front of the technology problem.

Source: Gartner

The Winners Built the Plumbing First

PwC’s 2026 AI Performance Study interviewed 1,217 senior executives, mostly at large listed companies, across 25 sectors. It found 74% of AI’s economic value captured by 20% of organisations, a divide PwC expects to widen rather than close.

The detail that matters for this issue sits further down. Leaders are increasing the number of decisions made without human intervention at 2.8 times the rate of their peers, and PwC attributes that to what it calls trust at scale: leaders are 1.7 times as likely to have a Responsible AI framework and 1.5 times as likely to run a cross-functional AI governance board, and their employees are twice as likely to trust AI outputs.

The companies removing humans from the most decisions are the ones that built the accountability structures first.

Source: PwC

Why are they interesting to watch?

Five domains, five different research methods, and the constraint lands on the human side of the system each time. Banks are hiring supervisors before the agents arrive. GPs are holding back until somebody settles the liability question. Supply chain officers are spending two-thirds of a digital budget they cannot yet measure. PwC’s finding closes the loop on all of it: the firms that automate the most decisions are the ones that built the governance and the trust structures first, which suggests the accountability layer is what makes the automation possible rather than what slows it down.

A second-order risk may sit underneath all of this, and none of these sources tests it. If supervising an agent well depends on judgement built by doing the work unaided for several years, then automating the junior and mid-level tiers first would thin the pipeline that produces supervisors. Worth watching rather than asserting.

So the planning question is a staffing question. If your AI programme depends on a supervisory layer, where is the next cohort of supervisors going to come from, and what are they practicing on now?

Signal Drop is a monthly dispatch from TBD Futures, drawing directly from the Signal Radar, a rolling database of early-stage observations worth watching before they become obvious.

Have a signal to share? Reply to this issue.

[all images are generated with the help of midjourney]

Subscribe now

I. AI May Make Correct Free Faster Than I’ve Hoped

The writing is on the wall. Artificial intelligence will commoditize everything scalable, repeatable, and trainable. Part of me wants to celebrate, because this may mean that the truly original, the genuinely human, the irreducibly personal will hold its value.

I want to believe this. I have wanted a version of it my whole life.

Growing up, like most kids, I was rewarded for correct answers. School, exams, even art classes had standards on technique, proportion, theory. The system rewarded us for working within its rules, and I was good at it. But correct never felt enough to me, and not just because the bar was low. Correct is something anyone can achieve if they pursue it hard enough, and after a while, the same answers get boring.

There was always something truer underneath and I could feel that it’s alive, breathing, blinking at me in dim light. It’s not something the rubrics contained in its boxes, nor evaluated in standardized tests. I went for the experiment that didn’t fit the instructions but satisfied my curiosity, the connection between ideas that the essay assignment didn’t ask for, and tree stump I randomly found on campus that became the centerpiece at the graduation concert stage.

Later, I chose a career that sits at the intersection of human and systems, creative and engineered. I kept on seeing the truer part surfacing beyond “the right answer” outside “the right framework”. A Futures workshop where the team that was assigned to plan for the most risky scenario came back with the most profitable opportunities in the room. A strategic reframe that no brief had asked for but reoriented an entire engagement. The insight that only became visible because two ideas from completely unrelated fields happened to collide in the same conversation. And a self-claimed “not creative” executive got inspired by three short clips in our customer interview reel and started to offer new ideas.

And every time it happened, we tried to write it into a retrospective, codify it, build it into the process, make it repeatable. Some of the condition building helps but happy accidents are copy-paste resistant. It keeps only happening when it’s ready, in ways that didn’t fit into an expectation.

To me, they have always been the most valuable things. Sadly, they have always been less valued by the system than the repeatable, mechanical, and boring things. So when AI arrives and threatens to make the correct, formulaic, duplicative, and boring essentially free, and promises to make it happen faster than ever, part of me wants to say: finally - now we’ll see what actually matters.

Subscribe now

II. The Market (Power) May Not Care

But the logic doesn’t hold that cleanly.

Start with where the incentive actually runs. Capitalism systematically engineers out meaning in favor of scale, therefore ROI. Optimization is the name of the game. One hour of genuine effort reaching thousands of customers. One beautiful design produced in millions of pieces. The ROI of care spread thin across transactions. This engine fuels the system. Companies grew, margins improved, and nobody at the top was asked to account for what got squeezed out in the process. (See my earlier writing on the Tyranny of Numbers).

The market never really rewarded the true over the correct. What got rewarded was the appearance of care and meaning at the lowest possible cost. And it got very good at that.

AI doesn’t disrupt that logic. It perfects it, scales it, makes a convincing imitation of meaning essentially free. You use it to generate the marketing message, summarize the customer conversation, and draft the email that says you care. No attention spent on what actually moved you, or what you sensed the customer needed but didn’t say. AI isn’t a threat to the current system. It’s the system, upgraded. Efficiency pushes meaning off the table and turns affect into effect.

And the people with the most leverage in that incentive structure are the ones who profit from scale without having to account for what gets squeezed out. Power right now sits with the people who were never really confused about any of this. The newly minted trillionaires. The administrations that court them and serve selfish aims. The platforms that spent decades insisting their mission was to connect humanity, while building the most sophisticated attention extraction machines ever made (Read my earlier piece about the Meta Trial). They wrote the manifestos about making the world more open. They also wrote the terms of service. One document was designed to be read. The other was designed not to be. AI doesn’t change their priorities. It hands them a new instrument.

III. The Trajectories Might Be Broken

Several trends are showing momentum, but none of them points somewhere reassuring.

The severance of the pathway.

Entry-level work has been how the next generation absorbed what couldn’t be written down: judgment, instinct, the slow accumulation of knowing when something is wrong before you can explain why. That transmission is breaking.

Years ago when I worked as an eyewear designer, I wrote a document to train junior designers and interns. An Eyewear 101 file: everything I knew about the craft, articulated as clearly as I could manage. As I built it, I noticed something. The act of describing the work slowly transformed it. Art became technique in the writing. Once something could be named and described in repeatable steps, it could be taught. Once it could be taught, it could eventually be replicated, which was the goal.

I wanted junior designers to learn the guardrails, the ergonomics, the processes faster so they could produce quality work. What remained irreplaceable was precisely what resisted description: taste, instinct, the judgment call that couldn’t be preemptive bullet points in a brief. These young designers wouldn’t have my level of expertise just by digesting the instructions, but they would add their own creativity on top of it, eventually becoming unique designers with their own fingerprint.

Now I can feed that document into an AI tool and have it run at 100x the speed. There won’t be taste, instinct, or judgment, which the system could hardly measure anyway, though it implicitly considered them important. But there will be variations, faster and cheaper than 10 interns could produce, which is of measurable value and thus importance.

The consequence is starting to show up in numbers. According to a Stanford Digital Economy Lab study using payroll data from 25 million workers, those aged 22 to 25 in the most AI-exposed occupations saw a 6% decline in employment between late 2022 and July 2025, while workers 30 and older in the same roles grew between 6% and 13%. Among software developers specifically, the drop for the youngest workers was nearly 20% from its late 2022 peak. The junior layer isn’t just being displaced. The pipeline that built the senior layer is being dismantled before anyone has worked out what replaces it. Senior credibility becomes scarcer and more valuable, with nothing coming up behind it.

The collapsing of the scaffolding.

There is a layer of work beneath the brilliant idea: the frameworks, lenses, and instincts that structure a problem before anyone tries to solve it. Call it scaffolding: the conditions that make insight possible, not the insight itself.

People are outsourcing it to AI. What comes back depends entirely on where you’re standing.

A first-time strategist gets a framework more structured than instinct alone could assemble. Someone writing in a second language gets fluency that used to require years of immersion. Faster, more coherent, good enough to be useful. The floor has risen.

For an expert, the same output reads as sloppy. The frameworks are applied without judgment about fit. The synthesis looks coherent and falls apart under pressure. It is the average of everything the model has seen, produced without the taste to know what this particular situation actually requires. I know many clients are receiving AI-generated futures work, trend analysis, scenario stories, and they look decent but generic. So generic that the client might as well generate it themselves, and eventually they will.

The distance between novice and expert is narrowing, and AI is doing the narrowing. The novice with access to good tools produces output that functionally resembles what took years of practice to achieve. The gap in surface competence, the thing most clients can actually evaluate, has genuinely closed.

Where real distance still exists, the market struggles to price it. A buyer who can’t see the difference between a generic scenario and a precise one, between a framework correctly applied and one that only looks correct, makes purchasing decisions accordingly. The expert loses the premium not because the expertise disappeared but because the signal that used to mark it has become harder to read.

What sits underneath this is harder to name. Judgment was built through the work that is now being outsourced. Sitting with a problem before reaching for a framework. Running ten sad drafts before the one that did. Pulling two methods together that had no obvious reason to connect, and finding the insight in the gap between them. Hand that process to AI and the output arrives. The capacity that produced it starts to atrophy.

Some people already feel it. According to EY’s 2025 Work Reimagined Survey, 37% of employees who use AI at work worry that overreliance on it is already eroding their skills and expertise.

The brilliant idea will get harder to reach because the work that made it possible stopped happening.

The premiumization of human presence.

Luxury experiences are now the fastest-growing segment of consumer spending, outpacing goods as consumers shift toward travel, wellness, and live events, a shift the Bain-Altagamma study called a “tectonic shift” away from conspicuous consumption and toward experiential indulgence. The wellness retreat market is growing at nearly 10% annually, reaching $248 billion in 2025. Live Nation reported 2024 as live music’s biggest year yet, with 151 million fans attending shows and concert revenue hitting $19 billion. Goldman Sachs projects the global live music industry will grow at 7.2% annually through 2030, driven by what its analysts call audiences’ unwillingness to cut back on shows even when finances tighten. The more virtual life becomes, the more a room with real people in it feels like something worth paying for.

In the early 2000s, as digital publishing was opening the floodgates to anyone with a laptop and an opinion, I had a conversation with a publisher. I asked him if he was worried. He said he wasn’t, because all it would do is make writing a commodity. He wanted to stay in the business of making books that were coveted, that had texture, that had quality. The dilution of the market, he said, doesn’t remove the value of what’s truly worth having.

What he said stayed with me ever since. Every time I watched a new tool make a craft cheaper, a new entrant rewrite the economics of an industry, a platform strip out the middleman and call it progress, I came back to the same question: where is the coveted side of this, and am I (or is my client) on it? When a category gets commoditized, the people who survive it were never really competing in that category. They were competing in the one above it, where the thing being sold wasn’t the output but the judgment, taste, care, and meaning behind it.

But before we apply that logic to now, it’s worth naming the assumption it rests on. The publisher’s answer assumed AI would flood the market while leaving quality intact, that the genuine article would remain distinguishable. Already, access to genuine human presence, a doctor who asks how to pronounce your name, a mentor who takes a long deep breath after hearing your troubles, a teacher who tailors the lesson plan to reach you, is increasingly a function of wealth and privilege. The premium tier gets the human. Everyone else gets the interface. The coveted is already tiering itself out of reach. What happens when AI produces a convincing version of quality on top of that? When the coveted and the generated become, to most eyes, indistinguishable?

IV. There Might Still Be One Thing That Can’t Be Transacted

When the informational playing field flattens completely, when everyone has access to the same analysis, the same outputs, the same correct answers, the differentiator stops being who has the best answer. It becomes who you trust to act on it.

Trust is not informational. It doesn’t live in outputs. It builds through years of showing up, accountability that wasn’t required, failures that were owned rather than explained away. When a board approves a risky bet, the analysis is rarely the deciding factor. What decides it is whether someone in the room who is willing to stand behind the bet has accumulated enough credibility to be believed when the data runs out.

Maybe what reasserts itself is something older than the market. Societies have always found ways to identify the people whose knowledge was earned rather than retrieved. Tribal elders were the most knowledgeable because they had lived the longest, made the most decisions with real consequences, and carried the memory of what worked and what didn’t. That’s also why they were trusted. AI can retrieve what they knew but not replicate how they came to know it. The trust can’t be transported with the knowledge.

What I find harder to answer is what happens when AI starts simulating the signals of trust, not just the outputs. Credibility has always been legible through proxies: the person who said the difficult thing in the room when it was easier not to. The advisor who called the wrong bet their own. The track record that included a visible failure and a visible recovery. These signals were costly to fake because they required time and consequence.

AI doesn’t replicate consequence, but it can now produce the language of it with some fluency. It can write the recommendation letter, the post-mortem, the application essay in the register of someone who learned something. When the language of hard-won credibility becomes producible on demand, the signal degrades. What it gets replaced with, I don’t know. I hope that it’s something harder to specify and harder to scale.

I also don’t know whether relational credibility can survive as an economy. Whether the trusted person in the room becomes more valuable as everything else flattens, or whether the system finds a way to simulate that too. Whether the publisher’s answer holds in a world where the machine alone can make something beautiful and nobody can tell the difference.

V. It Might Not Just Be About the Market

The market doesn’t have to have the final word on what endures. Some interesting history:

The printing press didn’t just distribute knowledge. It dissolved the interpretive authority of institutions that had spent centuries making themselves indispensable as the only readers in the room. That took a century of religious war before it settled.

Abolition ended something with enormous economic infrastructure and legal protection because enough people decided, ahead of the numbers, that no argument about profitability could make it acceptable. They were not the majority. They were persistent.

Rock and roll was dismissed as noise and moral danger by the same culture that became its most devoted audience within twenty years.

In each case, the market was a lagging indicator. The movement that changed the arrangement was made of people who stopped waiting for the incentive structure to align.

Can this be one of those moments for historians in the future? If we find our way through it like those amazing people did before, we would be the abolitionists and rock and rollers of the 2020s, against enormous odds, by improbable means.

It won’t be easy. It requires people who are willing to behave differently before the economics catch up, who hold the line on what genuine meaning looks like, who refuse the sufficiently convincing version when they can see it for what it is. That has never been the majority position. It has been enough occasionally.

I don’t know which historical precedent we’re inside right now. Whether this is a moment where the moral weight eventually shifts, or whether the instrument this time is powerful enough to absorb the resistance before it reaches critical mass. What I find myself watching for isn’t the technology. It’s the people who decide, without waiting for permission, that the coveted side of the line is still worth competing on. How that goes is how this story goes.

VI. Maybe That’s Enough to Go On

I started writing this because I wanted to believe something.

That when AI makes the correct free, what’s left will be the part that always mattered more: the original. The genuinely human. The irreducibly personal.
I still want to believe it. I’m just less sure it happens automatically.

What I keep coming back to is that none of this is settled by the technology. The technology is doing what technologies do: it scales whatever logic was already running. The question remains ours: what we value, what we reward, what we refuse to accept as a sufficient substitute for the real thing. We have just rarely acted like it, and the reasons are not mysterious: insisting is expensive and often unreimbursed. The game was designed to make the cost feel personal and the compromise feel collective. It’s frustrating and disconcerting to watch it play out again now, the gaps widening, the cracks spreading, people losing their footing and not knowing where to look.

Maybe the speed and scale of what’s coming will force a reckoning that slower disruptions allowed us to avoid. Maybe enough people find, at roughly the same moment, that the simulation isn’t enough, that something is missing in a way they can name. Or the simulation gets good enough that the question stops feeling urgent.

I don’t know which of those worlds we’re headed into because I see evidence of many directions, and as a futurist I hold ambiguity and uncertainty like wet puppies, feeling pain and hope, care and confusion all at once, all wrapped in the sense of a strange responsibility to accept. For now. What I know is that the people who decide it will be the trillionaires and the platforms and the administrations that court them, unless others, without waiting for permission, keep insisting on the real thing.

That has never been the position of power, or necessarily the majority, but as history keeps suggesting, it has been enough at the moments it needed to be.

Subscribe now

Five signals hit the radar this month that look unrelated at first glance: a cosmetic surgery trend, a study on delayed medical care, new research on chatbots and loneliness, women paying to smash things, and reading rooms replacing bars. Sit with them for a moment.

People are renegotiating what real feels like, in their faces, their bodies, their health decisions, and their social lives. AI is in the middle of all of it, quietly setting new defaults for what counts as beautiful, reliable, and connected.

Subscribe now

The Bratz Doll Problem

Plastic surgeons in the U.S. are now seeing patients bring in AI-generated versions of their own faces and asking to look like that, as reported by Business Insider. One Manhattan surgeon described a woman in her 70s asking for a “surgical time machine,” while another described the aesthetic many of these tools produce as a “Bratz doll look” — huge lips, enlarged doll-like eyes, a sharply defined jaw, poreless skin — also in Business Insider.

A 2025 study on AI image-enhancing filters found that exposure to these tools can significantly raise expectations for plastic surgery outcomes and may predispose patients to lower satisfaction afterward, as summarized on PubMed. The reference point used to be a celebrity. Now it is increasingly an algorithm optimizing for engagement, often flattening faces toward a narrower ideal of smoother, younger, more symmetrical beauty, which is exactly the shift described in Business Insider.

The downstream effects do not stay on screen.

The Doctor Can Wait

Patients are increasingly consulting AI before they see a doctor. A 2026 Zocdoc survey reported by Fortune found that 65% of patients had turned to AI for medical advice because it felt easier than seeing a clinician.

The access gap is real. The problem is what fills it. A cross-sectional study published in JMIR in February 2026 found that trust in AI was a statistically significant predictor of delayed healthcare-seeking, with the effect strongest among patients with chronic diseases, according to the PubMed record. In plain English: people who trusted AI more were also more likely to postpone professional care, especially when they were already medically vulnerable.

That same Fortune report says 83% of providers have had to correct AI-generated misinformation brought in by patients. For insurers and health systems, an access problem is quietly becoming a clinical one.

More Chatbot, More Lonely

A four-week study from MIT Media Lab and OpenAI, involving 981 participants and more than 300,000 messages, found that the people who used ChatGPT more heavily tended to report worse outcomes on loneliness, social interaction, and emotional dependence, according to Fortune’s coverage of the research. The direction of causality is still being worked out. Lonely people may seek out chatbots more often; heavy use may deepen loneliness further. It may be both.

At that point, scale matters. The APA reported that Character.AI had about 20 million monthly users, with more than half of users in the 18–24 age range. At that level of adoption, “we are still figuring out the mechanism” starts to sound less like a research footnote and more like a public health question.

Who Gets to Be Angry

At one UK venue, rage room reservations reportedly surged 150% year over year, and around 90% of customers were women, according to The Week. The format is not new. The demographic skew is.

Reporting from the BBC and The Week suggests that some rage room operators are seeing overwhelmingly female customer bases, raising a broader question about where women are allowed to put anger in public life. The rage room boom may be less about novelty than unmet emotional infrastructure.

When anger cannot find expression at work, at home, or in public, it tends to find expression somewhere else. Sometimes with a sledgehammer.

The Rise of the Anti-Social Social Club

Eventbrite’s 2026 Social Study, based on a survey of 4,051 adults in the U.S. and UK, points to a broader shift in how younger people want to gather, according to Eventbrite’s own newsroom summary. The pattern is less “go out to socialize” and more “go do something together, then let connection happen sideways,” a theme reflected in trend coverage of “soft socializing” from DivaDance and Eventbrite’s 2026 trend report PDF.

Reporting on Eventbrite’s 2026 Social Study says silent events saw a 14% increase in attendance, coffee clubbing events grew 478% on the platform, and sober-curious gatherings rose 92 percent, according to DivaDance and Eventbrite’s 2026 trend report PDF. The term circulating around this shift is “soft socializing” — shared activities over small talk, connection without performance.

At the same time, Gallup found that only 54% of American adults now say they drink alcohol, the lowest level recorded in nearly 90 years. Listening bars, silent book clubs, puzzle nights, morning raves: people still want to gather. They just want to do it on very different terms.

Why these signals matter

Somewhere between the surgical table and the smash room, something is shifting. AI is rewriting the baseline for what faces should look like, what a medical opinion should sound like, and what counts as company. Some people are adapting to those new defaults. Others are building new infrastructure for the parts that do not fit: spaces to be angry, spaces to be quiet, spaces to just be around people without having to perform it.

The question sitting underneath all of this, the one worth bringing into the next planning cycle, is simple: what version of the human were products, workplaces, and strategies built for? That version is changing.

Signal Drop is a monthly dispatch from TBD Futures, drawing directly from the Signal Radar, a rolling database of early-stage observations worth watching before they become obvious.

Have a signal to share? Reply to this issue.

Leave a comment

(All images generated with the help of MidJourney)

Here, we look at the implication that doesn’t show up in any of the economic forecasts — and may be the biggest one of all. Then we ask the harder question: what if the urgency is overstated? And finally, the three questions every leader should be sitting with right now.

[Image and video created with the help of midjourney]

Subscribe now

The Suburb Nobody Is Seeing Yet - Due to the democratization of expertise

The teenager was an accident. Nobody engineered youth culture — the car just made it possible for the first time, and possibility at scale tends to become inevitability.

There may be an equivalent accident already underway with AI, and almost nobody in the productivity debate is paying attention to it.

In December 2025, Anthropic conducted what it describes as the largest qualitative AI study ever conducted — 80,508 Claude users across 159 countries in 70 languages. The results, published in March 2026, reveal something hiding beneath all the productivity metrics.

When asked what they wanted from AI, the most common answer was not “make me more productive.” People began by talking about automating emails or speeding up code — the faster horse answer — but when pressed on what that would actually enable, they described something older and more human: the desire to be present for their lives again. Productivity was the vocabulary people reached for — the professional, reasonable answer. But underneath it, the stories were about getting to something that used to be out of reach.

And when you look at what people are reaching for, the second-order implication comes into focus: AI is functioning as an equalizer of expertise at a scale and speed that no previous technology achieved.

A physician in Israel, suffering from a neurological condition that local specialists couldn’t diagnose, used AI to find two scientific studies that led to effective treatment. A mute worker in Ukraine built a text-to-speech tool that let him communicate with friends in near-real time. A student in India — whose professor teaches 60 people and “will not entertain many questions” — now asks anything he needs at 2am without judgment or embarrassment. An entrepreneur in Cameroon reached professional-level competence in cybersecurity, UX design, marketing, and project management simultaneously, because he could not afford many failures. “Without being based in the US or UK,” an entrepreneur in Uganda explained, “AI may be my only way to compete.”

The automobile gave people physical mobility — access to geographies previously unreachable. The second-order implication was the suburb, the road trip, the teenager, the restructuring of American life around freedom of movement.

AI is giving people cognitive and economic mobility — access to expertise, judgment, mentorship, and professional capability previously locked behind credentials, geography, wealth, and institutional gatekeepers. The second-order implication of that mobility is what we haven’t mapped yet. What happens when a billion people in the developing world suddenly have access to expert-level legal, medical, financial, and educational guidance? What new forms of entrepreneurship, political organization, cultural production, and economic competition does that unleash? What institutions — built on the assumption that expertise is scarce — collapse when it becomes abundant?

Four out of five people across 159 countries told Anthropic that a technology they’ve had for a handful of years had already moved them meaningfully closer to the life they wanted. And we are almost certainly at the very beginning. The automobile reshaped civilization over thirty years. AI is running on a compression cycle measured in years — which means whatever this turns out to be, we’re still somewhere around 1905, squinting at the first car on the road and asking how much it costs to feed.

A Brief Steel Man: What If the Urgency Is Overstated?

A credible counterargument deserves acknowledgment. Goldman Sachs chief economist Jan Hatzius stated plainly in early 2026 that the hundreds of billions invested in AI contributed “basically zero” to US GDP growth in 2025 — and a February 2026 NBER study surveying nearly 6,000 CEOs and CFOs across the US, UK, Germany, and Australia found that 90% of firms reported no measurable productivity improvement, with perceived gains consistently outrunning measured ones. The researchers explicitly labeled it a “productivity paradox.” Meanwhile, MIT’s July 2025 GenAI Divide report — based on 300 public AI deployments — found that 95% of enterprise AI pilots delivered no measurable P&L impact. IBM’s research puts average enterprise AI ROI at just 5.9% — well below the typical 10% cost of capital.

These are not fringe findings. They are the respectable case for patience.

The automobile analogy holds here too: in 1910, critics reasonably argued the car was a rich man’s toy that would never scale. They were right — for about fifteen more years. And then, in less than a decade, everything changed at once. The question isn’t whether AI’s transformation is coming. It’s whether you’ll be positioned when the compression hits — or whether you’ll still be debating the crosswalk while the highway gets built around you.

Three Questions Every Leader Is Failing to Ask

The automobile’s second-order implications weren’t entirely inevitable — they were partially designed, mostly by accident. We built cities for cars because nobody in 1910 asked: what kind of city do we want in 1950?

Most leaders today are asking a version of the horse question: how do we use AI to do what we already do, faster and cheaper? That’s a legitimate operational question. It’s just not the right strategic question. Below are three that are.

Question 1: What infrastructure are we building that assumes the old world?

The office tower was built around the assumption that knowledge work requires co-location. The strip mall was built around the assumption that shopping requires a car trip. Both became stranded assets when the underlying assumption dissolved.

Every organization is making equivalent bets right now — in real estate, in talent pipelines, in technology platforms, in product lines. In financial services, the credit scoring models, actuarial tables, and risk assessments built on human behavior assumptions are beginning to drift from the reality AI is creating. In healthcare, care delivery models built on the scarcity of clinical expertise may become the gas lamp — not because they’re bad, but because something structurally more accessible now exists. In education and professional development, the institutions built on the scarcity of expert access face a direct challenge from what the Uganda entrepreneur described: AI as the only equalizer available at scale, at 2am, without gatekeepers.

The diagnostic is the same across sectors: which of your current assets — physical, organizational, or intellectual — were built around assumptions that AI is quietly dissolving?

Question 2: Who falls off the ladder — and what replaces the on-ramp?

The railroad created monopolies before antitrust law existed. The automobile created a middle-class labor market before anyone planned for it. Technologies at civilizational scale restructure who gets to participate in the economy — and the restructuring rarely distributes its benefits evenly.

For CEOs and boards, the BCG data already shows the compounding divide: the top 5% of AI-mature firms achieving outsized returns while 60% report minimal gains, and the gap widening. That divergence shows up in talent, in valuation, in competitive positioning, and eventually in the broader social and regulatory environment companies operate within. The IMF’s warnings about AI-driven erosion of labor-based tax revenue and pressure on social systems are not abstract. They describe the downstream policy and stakeholder environment every major corporation will operate in within a decade.

For leaders in talent and education, the deeper question isn’t about updating the curriculum. It’s about the pipeline. If AI is simultaneously giving a student in India a personal tutor available at 2am and removing the entry-level roles that once taught judgment through repetition, what replaces the apprenticeship? The leaders who matter in this space are not the ones updating the syllabus. They are the ones redesigning the on-ramp.

For all of them, the question is the same: who is falling off the ladder in your industry — and are you building what replaces it, or waiting to be surprised by the consequences?

Question 3: What kind of organization are you becoming — and does anyone in the room have a clear picture of what that looks like in seven years?

The railroad didn’t just move freight — it created the first modern corporations and the first monopolies, concentrating economic power in very few hands, very quickly. It also created entirely new industries that didn’t exist before: professional management, standardized accounting, nationwide logistics. The technology didn’t just transform existing businesses. It invented new categories.

This is the question strategy, governance, and the board table haven’t fully confronted yet. Not “what is our AI strategy?” but “what kind of organization are we becoming?” Those are different questions. The first is an implementation question. The second is an identity and direction question — and it’s the one the automobile era kept answering by accident, city block by city block, without anyone making a deliberate choice.

The difference between planning the suburb and stumbling into it is whether someone asked that question before the concrete was poured.

The City Nobody Is Planning

Every leader reading this is already placing a bet, whether they’ve made it consciously or not.

Either AI’s transformation is as structural as the automobile’s — in which case strategies built for 2026 are already becoming obsolete — or the urgency is overstated and the change arrives more slowly than the hype suggests. Both possibilities, strangely, call for the same move: systematic second-order thinking, starting now. The leaders who win in either scenario aren’t the ones who called the timing correctly. They’re the ones who asked better questions while everyone else was still debating whether the car would scale.

The automobile didn’t reward the people who saw it coming. It rewarded the people who asked what comes after the car.

Something is already happening that doesn’t fit neatly into any of the economic forecasts. Eighty-one thousand people across 159 countries just told Anthropic, in their own words, what AI is doing for them — and the pattern beneath the productivity statistics isn’t what the analysts are tracking.

The automobile gave people physical mobility. It unlocked geographies. The second-order consequence — the suburb, the teenager, the road trip, the restructuring of American life around freedom of movement — emerged from that unlocking in ways nobody designed and nobody fully predicted.

AI is giving people cognitive and economic mobility. It is unlocking expertise. What emerges from that unlocking — what new forms of entrepreneurship, political organization, cultural production, and global competition it unleashes — is the city nobody is planning yet.

The faster horse is already in the barn. The roads are already being built — for the machine’s needs, not yours, unless you’re paying close attention.

Nobody conjured up the suburb. Nobody planned the teenager. The question for every leader reading this isn’t whether the city gets built around AI.

It’s whether you have any say in what it looks like. And it may not be too late, yet.

Let me know what you think, how you would like to apply the second order implication thinking in your strategy work, and how I can help.

Leave a comment

[Images and videos created with the help of midjourney]

Recently, I built multiple AI skills and agents for my own work, automating processes I’d spent years developing and codifying judgment mechanisms that only a seasoned practitioner could master. The speed at which I was able to build them was elating and genuinely unsettling all at once. I caught myself cursing under my breath at my computer more than a few times while staring at the growing suite built and the massive outputs it produced overnight.

I know I’m not alone. Knowledge workers everywhere are doing the same, some by choice, many by company mandate. And the quiet fear is shared. No one is sure what the productivity breakthroughs at this scale mean for us in the next months, let alone years.

The Futurist in me couldn’t help but ask another question, just to make everything feel a bit more overwhelming: if this is what the beginning looks like, how big will it actually get?

I happen to have some handy tools to help us tackle it, and my hope is that when we brave the overwhelm and go after the hard questions, we may come back out from the other end with a bit more clarity, and even conviction and confidence. Would you join me?

To study the future(s), we need to start with history. In part I of this 2-part series, let’s go straight to one of history’s most transformative technologies.

The “faster horse” of the last century never announced its real impact. So one thing is for sure: AI won’t either.

Key Takeaways:

• History’s most transformative technologies always delivered their biggest shock not through their obvious function, but through what they redesigned around themselves.

• Today, we are treating AI as a productivity tool — a faster worker, a smarter search engine — and missing the civilization-level redesign already underway.

• Strategic foresight offers a disciplined answer: implication mapping — the practice of tracing how a technology cascades through systems before the consequences arrive. A handful of organizations are already doing this. Most are not.

• The implications are arriving in years, not decades — and leaders who only see the tool will be blindsided by the transformation.

Subscribe now

We’re Staring at the Horse Again

In 1900, the automobile looked like a faster horse. So people asked horse questions: How fast can it go? How much does it cost to feed? Can it carry more than a horse?

Nobody asked: What happens to the city when everyone has one?

Here’s what actually happened — and it’s worth spelling out carefully, because the dots are rarely connected:

The car needed roads, so we built them — and roads needed land, so we took farmland on the edges of cities. Developers followed the roads and built houses — cheaply, at scale — and called them suburbs. Suburbs needed stores, so we built strip malls alongside highways instead of downtown. Strip malls killed main streets. Downtown retail collapsed. Cities hollowed out. White flight followed the highways outward, sorting wealth and poverty geographically in ways we’re still litigating today.

Meanwhile, cars needed gasoline, so oil became the most strategically critical resource on Earth. We built the entire 20th-century geopolitical order — alliances, wars, dictatorships propped up and toppled — around who controlled it. The 1973 OPEC oil embargo brought the US economy to its knees. US foreign policy, military alliances, and strategic calculations from the Gulf to Central Asia were shaped for decades by access to petroleum. A fuel source for one technology restructured global power for a century.

And then there’s the teenager. Before the car, adolescents were supervised — at home, at church, in the fields. The automobile gave young people private mobility for the first time in human history. The teenager as a social category — with its own music, fashion, dating rituals, rebellion, and eventually its own billion-dollar consumer economy — was an accident of the internal combustion engine. Nobody planned it.

The car didn’t just move people faster. It invented the suburb, the strip mall, the oil war, the teenager, the drive-thru, the motel, the road trip, the fast food chain, the traffic jam, the smoggy city, and the parking lot that ate downtown. None of that was in the brochure.

We are making the same mistake with AI — right now, today.

History’s Pattern: The Tool Is Never the Story

This isn’t the first time we’ve been surprised. It’s a pattern in civilizational transformation.

The Printing Press (1440s): Gutenberg’s press looked like a publishing tool — a faster way to copy books. What it actually did, as the Stanford Social Innovation Review documents, was trigger the Protestant Reformation, dismantle the Catholic Church’s monopoly on knowledge, and fuel the scientific revolution. It didn’t just spread information — it atomized authority itself. Nobody saw that coming. They just wanted cheaper Bibles.

The Railroad (1830s–1880s): Railroads appeared to be a faster way to move goods. What they actually did — as Harvard business historian Alfred Chandler documented in his landmark work The Visible Hand — was create the first modern corporations, invent professional management, establish uniform time zones, and birth the Gilded Age’s concentrated wealth. The railroad didn’t just move freight — it rewired the economic and geographic DNA of a nation. Nobody planned the monopoly. They just wanted faster freight.

Electricity (1880s–1920s): Edison’s grid looked like a replacement for gas lamps. What it actually did, as The Exponential Age documents, was enable factories to operate anywhere at any hour, make skyscrapers viable via electric elevators, and create the modern white-collar office economy. It didn’t just light rooms — it restructured where and how human civilization could exist. Nobody planned the skyscraper. They just wanted to stop using candles.

Each of these technologies followed the same arc: obvious function, invisible transformation. AI is following the same pattern — except the compression cycle is measured in years, not decades.

What Is Implication Mapping — And Why Does It Matter Now?

Strategic foresight practitioners use a discipline called implication mapping to avoid being surprised by their own tools. The exercise is simple in concept and demanding in practice: start with a signal or trend, then ask “and then what?” — not once, but iteratively, tracing how that change cascades through adjacent systems, behaviors, industries, and power structures.

It’s what nobody did with the automobile in 1905. It’s what nobody did with the railroad in 1845. And it’s what almost nobody is doing systematically with AI today.

Harvard Business School faculty writing in late 2025 put it plainly: organizations are doing enormous work on the first-order effects of AI — how do people work with it, how do we design it for efficiency — but “in 2026, we also need to start thinking very carefully about the second-order implications: How does AI change my experience of work and its meaning to me?” That is precisely the question the automobile industry never asked.

The organizations already doing this well share a structural habit: they set defined strategic objectives before scanning for signals, which research from a 2026 corporate innovation study shows immediately eliminates 60–70% of irrelevant noise and forces second-order thinking rather than first-order reaction. Consider what this looks like in practice: a pharmaceutical company using AI foresight to identify early signals around mRNA platforms across academic journals — three years before the market was disrupted by that exact technology. Or an industrial manufacturer that flagged 3D printing adoption curves in aerospace and adjusted capital allocation accordingly — before competitors recognized the shift. The discipline isn’t exotic. What’s rare is applying it to AI itself, the very technology doing the disrupting.

The sections below are an attempt at that map.

The Second-Order Implications We’re Likely Missing

The Office Tower Is the New Horse Stable

When the automobile arrived, horse stables didn’t become parking garages overnight — but within a generation, they were obsolete. The infrastructure built around the previous technology became stranded.

Today’s office towers, corporate campuses, and downtown business districts are the stables of the AI era. They were built around one core assumption: that knowledge work requires humans to be physically co-located, nine-to-five, in expensive real estate near other expensive real estate. AI will dissolve that assumption — not immediately, but structurally and irreversibly.

The evidence is already in the data. Newmark’s April 2026 analysis finds that office-using sectors face the greatest AI exposure risk in commercial real estate, with professional and business services alone expected to see the human-only task share drop from roughly half today to one-third by 2030 — two-thirds of that driven by outright automation, not just augmentation. Their base case projects national office vacancy reaching 21.5% by 2030. Cushman & Wakefield’s 2026 analysis maps the risk pathway toward stranded assets — towers and corporate campuses increasingly misaligned with how work actually gets done.

The question isn’t whether this risk materializes. It’s whether your organization is holding the equivalent of a 10-story horse stable and tracking the early warning signals.

Meanwhile, AI’s own physical infrastructure — data centers — is consuming land, water, and energy at extraordinary scale, redesigning geography the way gas stations and highways once did. We are building for the new machine’s needs whether we realize it or not.

The Entry-Level Ladder Has Been Pulled Up

The automobile didn’t just reshape cities — it created a new economic on-ramp for an entire generation. Assembly line jobs, gas station attendants, mechanics, road crews — the car built a ladder into the middle class for workers without college degrees.

AI appears to be doing the opposite. The junior analyst, the junior associate, the entry-level copywriter, the first-year paralegal — these roles were never glamorous. But they were the apprenticeship system of the white-collar economy. You learned by doing the work senior people didn’t want to do. You built judgment by grinding through the repetitive cases, the routine documents, the unglamorous research.

AI is now doing that work. And in doing so, it is quietly removing the bottom rung of the professional ladder before anyone has redesigned what replaces it.

The signal is already visible in the labor market. Amazon announced in late 2025 the largest layoff round in its history — 14,000 corporate roles — explicitly citing AI as the driver, with CEO Andy Jassy writing plainly: “We will need fewer people doing some of the jobs that are being done today.” Amazon’s move was the most public signal of a dynamic playing out across sectors. The MIT November 2025 labor study estimated AI can already perform the work of 11.7% of the US labor market, with potential wage displacement of $1.2 trillion across finance, healthcare, and professional services. Those aren’t future projections. They are current capability assessments.

BCG’s research reveals what this looks like at the firm level: the top 5% of AI-mature companies are already achieving 1.7x higher revenue growth and 3.6x greater shareholder returns than their peers — and the gap is widening, not converging. Those firms are redesigning roles wholesale, automating entry-level work and expecting the humans they hire to operate a full tier higher from day one. The firms doing this today may discover, a decade from now, that they’ve created a missing generation of mid-level talent — people who never had the chance to develop the judgment that only comes from doing the unglamorous work first.

Data Is the New Oil — And the Wars Have Already Started

The automobile created a century of geopolitical conflict over petroleum. The logic was simple: whoever controls the fuel controls the economy controls the power.

AI runs on data and compute. The logic is identical. The Atlantic Council’s 2026 analysis identifies eight vectors through which AI is already reshaping geopolitics this year. China’s DeepSeek disrupted Western assumptions about who leads AI development. US export controls on advanced chips are the 21st-century equivalent of an oil embargo. The IMF warns that nations lagging in AI adoption risk meaningful erosion of their labor tax bases and long-term competitiveness.

The countries — and companies — that control the training data, the chips, and the foundational models will hold geopolitical leverage the way Saudi Arabia held leverage in the 1970s. The difference: Saudi Arabia knew it owned the oil. Most companies have already handed their equivalent resource — their customer data, their operational data, their institutional knowledge — to a handful of cloud and platform providers who are training the models that will define the next century of economic power.

AI Fluency Is the New Driver’s License

Before the car, you didn’t need to know how to drive. After the car, not knowing how to drive meant you couldn’t fully participate in the economy — you couldn’t get to work, couldn’t access the suburban geography the automobile had built around itself.

AI is building its own prerequisite at comparable speed. The 2024 Microsoft and LinkedIn Work Trend Index, surveying 31,000 people across 31 countries, found that 66% of leaders say they would not hire someone without AI skills — and more strikingly, 71% say they’d rather hire a less experienced candidate with AI skills than a more experienced one without them. The credential has already inverted seniority.

By 2026, LinkedIn’s Skills on the Rise data shows AI engineering, prompt engineering, and model optimization among the fastest-growing skills globally — and LinkedIn’s own economists warn that the average job’s required skills could change by 70% before 2030. That’s not a hiring preference. It’s the early signal of a new baseline, one that will function the way the driver’s license did within a generation: table stakes for economic participation.

The people who don’t learn to drive — metaphorically — will find the world increasingly built for those who do.

The K-Shaped Economy Is the New Suburbanization

The automobile didn’t just create suburbs — it hollowed out the urban core. As wealth and retail followed highways outward, city centers declined, and the geographic sorting of economic class accelerated. This wasn’t the goal of the automobile. It was the consequence.

AI appears to be creating its own sorting mechanism — and economists have a name for it. Capital Analytics and U.S. Bank’s 2026 analysis both describe an accelerating K-shaped economy: higher-income households and AI-enabled industries rising on one arm of the K, while lower-income workers, small businesses, and non-AI-adopting sectors fall on the other. As one Alliance Bernstein chief economist put it: “The way AI is boosting productivity is largely by punishing labor.”

Goldman Sachs warns that AI could affect 300 million full-time jobs globally, while the World Economic Forum’s Future of Jobs Report 2025 estimates AI and automation will likely create 170 million new roles by 2030 while eliminating 92 million — a net positive of 78 million jobs on paper.

That net gain is real — and almost entirely beside the point. The 92 million eliminated roles are disproportionately concentrated in the middle of the skills and income distribution. The 170 million new ones require different capabilities, different geographies, and years of retraining that most displaced workers won’t have. The math nets to positive. The human experience nets to something considerably more complicated. The suburban dream was built on a car-enabled middle class. AI may be quietly redrawing the map of who gets to live there.

The Governance Gap Is the New Traffic Jam

Nobody designed traffic law before the car. Stop signs, speed limits, liability rules — all of it came after the chaos, after the first pedestrian deaths, after the first fatal crashes. The printing press triggered the Reformation before anyone could write a rule about heresy. The railroad created monopolies before antitrust law existed.

The IMF notes that AI technology is already outpacing existing governance frameworks, while multiple leading AI researchers — including teams at DeepMind, OpenAI, and Anthropic — now estimate AGI timelines ranging from 5 to 15 years, a range that has compressed dramatically from estimates made just five years ago. Governments are writing AI policy the way early cities painted crosswalks after the first pedestrian deaths — reactive, local, and likely insufficient.

The difference this time: the automobile took 30 years to fully reshape civilization. AI is currently operating on what appears to be a 3–5 year compression cycle. We likely don’t have the luxury of learning slowly. Gartner’s 2026 strategic predictions capture the governance lag in practical terms: by end of 2026, they project “death by AI” legal claims will exceed 2,000 due to insufficient risk guardrails — and that 90% of B2B buying will be AI agent-intermediated by 2028, pushing over $15 trillion through AI agent exchanges that no regulatory framework currently anticipates.

The six implications above are the second-order questions the automobile era never asked — and paid for over decades. AI is running on a faster clock.

In Part II, we start with the implication almost nobody in the productivity debate is tracking — the one that may matter most. Then we pressure-test the urgency: what if the skeptics are right? And we close with the three questions every leader should actually be sitting with right now.

Subscribe to stay tuned. And share with those who should also ask better questions these days.

Share

Across Parts I, II, and III, we’ve traced a single argument: that the tyranny of numbers — our obsession with short-term, easy-to-measure outcomes — causes organizations to systematically undervalue trust, loyalty, and human experience, even when compelling data shows these are the real drivers of long-term performance. We’ve seen the asymmetry in practice, inside one of the world’s most data-sophisticated firms. And we’ve imagined what the alternative could feel like for a customer in 2029. Now comes the hardest question: how does it actually work financially? And what does it cost you if you choose not to change?

video and image created with the help of midjourney

1. Two futures for retail

We took a peek into the 2029 retailer last time from the vantage points of how a consumer, an employee, and a leader at the company experience it. Now I want to dissect it a bit and get into what would and wouldn’t lead us to it. The point of futures thinking isn’t to predict whether one specific retailer in 2029 will behave exactly as described. It’s to surface the choices that make such a future more or less likely. 2029 is only three years away. The window to choose is closing.

From that perspective, we can see two broad trajectories.

Future A: The empire of easy metrics

In this trajectory, the current logic hardens rather than softens.

Executive compensation remains tightly coupled to quarterly sales and margin. AI is deployed to squeeze more efficiency out of every interaction: more personalized promotions, more dynamic pricing, more urgency tactics. The goal is to maximize this transaction, this click, this visit.

Measurement systems have a few clear characteristics:

• Narrow focus. Dashboards are dominated by conversion, return on ad spend, and average order value. Everything else is noise.

• Short horizons. Most investments are justified on 6–12 month payback periods. Product longevity, repair infrastructure, and customer trust — whose benefits compound over years — are a hard sell.

• Thin narratives. Customer stories are occasionally featured in annual reports, but they don’t shift priorities.

The most striking thing about this future is that it is entirely unremarkable. This is what every retailer is already doing. Every competitor is measuring the same things, optimizing for the same metrics, chasing the same short-term gains. In this trajectory, you will never stand out. You will never build anything distinctive. You will be running faster and faster just to stay in the same place.

This future is not necessarily scandalous. It’s just relentlessly incremental. Over time, retailers in this trajectory become exquisitely tuned machines for extracting value from impulse, confusion, and FOMO.

They may look strong right up until something hits that doesn’t show up in their favorite metrics: a sudden collapse in trust, a generational shift toward sustainability, or a new entrant that competes on meaning and alignment, not just convenience and price. The danger isn’t collapse. It’s slow erosion — until the day a competitor who chose differently makes your advantages feel quaint.

Future B: The era of meaningful measures

That new entrant is the alternative trajectory. In Future B, organizations don’t reject numbers. They redesign them.

Several things have to become true for this to happen.

First, value is explicitly redefined at the top. Boards and executive teams acknowledge that long-term financial performance is inseparable from:

• Customer trust that survives a crisis.

• Product longevity and repairability.

• The organization’s capacity to tell the truth, even when it costs a sale.

• Its relationship to the communities and environments it touches.

These are not relegated to sustainability reports. They are written into the core objectives the organization steers by. And critically, leadership accepts that these investments will look bad in the short-term metrics — and that this is not a sign of failure, but a feature of the strategy.

Second, measurement is treated as a design discipline, not a neutral given. Cross-functional teams wrestle with questions like:

• How might we track the long-term value of a customer who was told not to buy?

• What’s the relationship between repair usage and five-year retention?

• How do we measure the trust that keeps people coming back even when we tell them things they don’t want to hear?

• What are early indicators that a short-term sales tactic is eroding long-term loyalty?

The result is not perfect. It’s richer. Metrics become a mix of leading and lagging indicators, numbers and narratives, short- and long-term signals. Some things remain unmeasured, and that’s okay — because the organization has learned to value things even before it can count them.

Third, incentives are adjusted to match the new story of value. Merchandisers are rewarded not just for volume, but for product durability and low return rates. Customer service teams are measured on problem resolution, not call time. Marketers are evaluated on trust metrics, not just click-through rates. Leaders who make decisions that trade off short-term revenue for long-term trust aren’t punished; they’re supported.

Finally, design and innovation move upstream. They’re present when goals and metrics are shaped, not just when promotions are executed. Their remit expands from “making the website easier to use” to “helping the organization see and care about things it has never properly measured.”

In this trajectory, AI is still a powerful tool. But instead of just accelerating extraction, it’s used to:

• Predict which products will fail before they do — saving return costs and preserving trust.

• Identify customers who would benefit from a repair rather than a replacement — deepening loyalty at lower cost than acquisition.

• Simulate the long-term trust consequences of pricing and policy choices — giving leaders visibility into what the old metrics hide.

The 2029 retailer we opened with is one plausible outcome of this trajectory: an organization that is still commercially successful, but grounded in a different, more human-centered understanding of value. It is also, not coincidentally, a retailer that would be genuinely difficult for competitors to copy, because its advantage is built on trust, longevity, and meaning — not just on doing the same things more efficiently.

Subscribe now

2. The business case for what looks irrational

Let me address directly what every leader is thinking: “This sounds nice, but how does it actually work financially?”

The answer lies in understanding where retail value actually comes from.

Acquisition vs. retention. Research from Harvard Business Review suggests acquiring a new customer can cost anywhere from 5 to 25 times more than retaining an existing one. Every interaction that deepens trust — even if it costs a sale today — reduces the likelihood that you’ll need to spend money acquiring that customer again later. The customer who was told not to buy a new coffee maker, who was offered a free repair, who was guided away from the wrong product: that customer is not going to price-shop your competitors. They’re not going to need expensive re-engagement campaigns. They’re simply going to keep coming back. And crucially, a customer who buys nothing today but trusts you completely is still a retained customer — one who will buy more broadly, and more often, over the years ahead.

Returns are a killer. The average return rate for online purchases is 20–30%. For apparel, it’s even higher. Returns cost retailers an average of $10–20 per item in shipping and processing alone, plus the lost inventory value. A customer who buys the wrong product and returns it is a net negative, even though they show up in the sales numbers. A customer who is guided to the right product — or guided away from buying altogether — never generates that cost. The “lost sale” on the front end is often cheaper than the “successful sale” that gets returned.

Trust drives category expansion. Research consistently shows that customers who trust a retailer buy across more categories. They don’t just come for coffee makers; they come for backpacks, for coats, for kitchen gadgets. The customer who trusts your advice has a higher share of wallet over time — even if they buy less frequently, they buy more broadly. The “lost sale” in one category is often recouped many times over in others.

Earned growth is the only sustainable channel. Paid acquisition costs rise every year. Platforms change their algorithms. Cookies disappear. The only channel that gets cheaper over time is word of mouth. And word of mouth is driven entirely by trust. Customers who feel you’re on their side tell other people. Those people come pre-trusted, pre-converted, pre-loyal. They cost nothing to acquire. The “inefficient” investments in trust are actually the most efficient customer acquisition spend you’ll ever make.

Long-term orientation reduces volatility. Companies with high trust scores recover from mistakes faster. They have more pricing power. They attract better talent. They have more flexibility in negotiations with suppliers and platforms. These are not soft benefits; they are hard financial advantages that show up in margin stability, in crisis resilience, in the ability to weather what sinks competitors.

The 2029 retailer we visited understands all of this. They’ve done the math. They know that every “lost sale” from honest advice is actually an investment in a customer who will be cheaper to serve, easier to retain, and more likely to bring others. They know that repair services lose money on paper but build the kind of loyalty that no marketing budget can buy. They know that trust is not a nice-to-have — it’s the only durable competitive advantage in a world where everything else can be copied.

3. What to measure is a choice

Remember I said that we chose retail to explore randomly? The same approach can work in many other industries, from healthcare to banking, from tech to media, wherever human experiences matter. The future will not be decided by AI alone, or by capitalism alone, or by any single consulting model or design method. It will be shaped, day by day, by a deceptively simple decision inside organizations:

What are we choosing to measure, and what are we willing to value even before we’ve figured out how to measure it well?

The crack in the tyranny of numbers — the pressure now reaching even the most trusted forms of expertise — is an opportunity to reset what we mean by value.

One trajectory will keep rewarding only what’s easy to count, even when it quietly corrodes trust, loyalty, and meaning. That path leads to undifferentiated mediocrity: competing on the same metrics as everyone else, never standing out, never building anything that lasts.

Another trajectory will take on the harder work of developing richer, more human-centered ways of knowing what is valuable — and of acting on that knowledge. This path requires accepting that some investments will look bad in the short-term metrics. It requires building scorecards that can see around corners. It requires leaders with the courage to say:

“I know this looks inefficient. I’m doing it anyway, because I’m playing a different game.”

The 2029 retailer is not a prediction. It is an invitation: an illustration of what it might feel like to be served by an organization that has decided to measure what actually counts, so we can imagine bigger, broader, and wilder. And then we can be inspired and motivated to build it.

The question for leaders is not to sit back and bet on whether such organizations can exist a few years. It’s whether you’re willing to redesign your value system now — before the tyranny of numbers quietly chooses your future for you.

What will your 2029 look like?

4. The AI question in our faces

As I drafted these 4 articles and set them up to be posted, I continued to be bombarded by news and debates about AI and its unstoppable ability to achieve unmatchable speed and efficiency. That’s what AI is great at and where AI will beat humans — speed, efficiency, scale. Now are you surprised that companies are pouring money into AI even without clear returns, often shifting budgets away from people and other long-term bets in the process, as described in pieces like “Enterprise AI investments are forging ahead despite elusive ROI” (CIO, 2026)?

Recent surveys show the tension clearly. A 2026 CIO analysis, drawing on PwC’s CEO data, notes that many executives report little or no measurable revenue or cost benefits from AI so far, even as they keep increasing their AI budgets because they fear falling behind competitors who are experimenting more aggressively. Gartner has gone so far as to predict a wave of abandoned generative AI projects: by the end of 2025, they expect a significant share of pilots to be shelved because of unclear business value, poor data foundations, and mounting costs — a pattern highlighted in “Gartner Predicts Wave of Abandoned AI Projects”.

At the same time, reporting from outlets like Reuters and labor analysts documents thousands of job cuts explicitly attributed to AI. In April 2026, Reuters reported that companies across sectors were cutting roles and “shifting investments toward AI,” often citing automation and efficiency as the rationale. Earlier coverage and analysis from 2025 and early 2026 shows AI-linked layoffs accounting for a growing share of announced reductions, with tens of thousands of jobs either directly labeled as “AI-related cuts” or framed as workforce restructuring to fund AI initiatives. In tech alone, one widely cited tally claimed that roughly 180,000 roles were eliminated in 2025 as large firms reallocated capital toward AI and automation, a figure that’s been used as a shorthand for the scale of the shift.

The reason is simple. When your scorecard is dominated by speed, efficiency, and scale, AI looks like the perfect employee and humans start to look like a cost center. What’s the point of having humans in the loop any more? It’s not a theoretical question; it shows up in headlines like “Companies Are Laying Off Workers Because of AI’s Potential — Not Its Performance” in Harvard Business Review, which describes firms cutting staff based on anticipated AI gains long before those gains are actually realized. Executive quotes in those pieces make the logic explicit: if AI can eventually do the work faster and cheaper, the human becomes the line item to trim.

I’m watching the broader theme creeping into how human-centered work is treated day to day. To speed things up, and cheaply, clients are using AI to generate personas and journeys with AI rather than conducting real research and having human researchers and professionals build those artifacts to surface insights, echoing patterns in enterprise surveys that show generative AI being adopted fastest in marketing, customer experience, and “insight generation” roles. Startup founders are burning through tokens to have tools like Figma’s AI features create prototypes and then ask human designers to clean them up, because they feel in full control and the tool doesn’t ever push back. As I’m writing this and arguing that human-centered work should be valued, the entire human-centered workforce seems to be under attack.

This is the tyranny of numbers in its purest, most accelerated form: once you let “faster and cheaper” become the only definition of better, you’ve already answered the question of what humans are worth.

I want to ask one last question:

Once AI can optimize virtually everything to fulfill the goals of speed, efficiency, and scale, for every company, what will set anything apart from this virtually perfect, and perfectly gray, generic, and soul-less world?

If your answer is “nothing,” then the tyranny of numbers has already won. If your answer is “trust, meaning, courage, judgment, taste, care” — then the work of the next few years is clear: redesign what you measure so those things stop being invisible, and start becoming the assets your future actually depends on.

More articles on the topics of metrics, innovation, and human values:

The Meta Trial: What Have We Been (and Should We Be) Building For?

Create Innovative Certainties: How Top Performers Navigate Uncertainty and Seize Opportunities for Growth

The Empathy-Throughput Tradeoff: Can AI-Enabled Scale Preserve Human Connection?

The Human Edge: The Enduring Value of (Certain) Human Skills

Running Faster to Fall Further Behind? How AI in Financial Services Can Miss Out on The Obvious

How Human Should The Future be?

In Parts I and II, we examined the tyranny of numbers: the systemic bias toward short-term, easy-to-count metrics that causes organizations to sideline trust, loyalty, and human experience. We looked at what the data actually shows — design-led companies outperform, but only over patient, multi-year horizons — and at the asymmetry I lived inside one of the world’s most prestigious firms: compelling evidence on one hand, structural resistance to acting on it on the other. Now it’s time to leave the diagnosis behind. What does the alternative actually look like, in practice?

Subscribe now

Images and videos created with the help of Midjourney

1. Could this be your shopping experience in 2029?

It’s 2029, and your favorite retailer feels, well, different.

Not because it’s wrapped in AI-powered recommendations, but because it seems to care more about your long-term satisfaction than your next purchase. When you open the app, the first thing you see isn’t a flash sale or a countdown timer. You see your “active relationships” — the items you’ve bought in the past year that might need care, repair, or replenishment.

The interface doesn’t just show you new products; it helps you take care of what you already own. It knows you bought a wool coat last winter, and as spring approaches, it asks:

“Your coat is due for storage. Want me to send you a guide on cleaning wool, or would you like us to pick it up for professional care — on us?”

When you search for a new coffee maker, the app doesn’t just show you options. It first shows you your current machine’s repair history, its estimated remaining life, and a comparison: “A new machine would cost you $250 and generate 40kg of e-waste. Repairing yours would cost $80 and extend its life 3–4 years. We can connect you with an authorized repair shop.”

Sometimes, its most honest recommendation isn’t a product at all. You’re looking at children’s backpacks, and the app gently offers:

“We noticed you bought a backpack from us two years ago. It’s still under warranty, and we’re seeing that model had some strap issues. Would you like us to send a free replacement strap rather than a new bag?”

When you do need something new, the experience is radically different. Instead of showing you everything, the app asks a few questions about how you’ll use the item, then shows you exactly three options — curated for your specific context. It tells you which one other people with your usage patterns kept for the longest, not which one got the most five-star reviews.

The checkout process includes one final question:

“You’ve chosen the mid-range option. Are you aware that the premium version exists? We didn’t recommend it because for your usage pattern, the mid-range will last just as long and costs 40% less.”

As a customer, you feel something you’re not used to feeling with a retailer: this company is, at least most of the time, on your side.

As an employee at this retail company, you feel the difference too. Your goals still include revenue, but your performance isn’t reduced to “How many units did you move?” In reviews, your manager asks: “Where did your work help a customer avoid buying something they didn’t need?” and “Which decisions led to fewer returns, less waste, and more long-term trust?” You are expected to care about numbers, but also about what those numbers represent.

As a leader at the retailer, your reality is more complex and more honest than before. Your executive dashboard still opens with revenue and margin. But right alongside them are trend lines showing:

• Customer lifetime value adjusted for trust. You’ve modeled that customers who use your repair services have 40% higher five-year value, because they see you as a partner, not a vendor.

• Share of wallet over time. You’re tracking not just what people buy from you, but what they buy from competitors — and you’ve noticed that customers who trust your advice buy more categories from you, even if they buy less frequently.

• Cost-to-serve trending down. Customers who get honest advice need fewer returns, fewer customer service calls, and fewer interventions. They’re cheaper to serve.

• Earned growth rate. How many new customers came because someone trusted you enough to recommend you? This number is growing faster than your paid acquisition channels.

You can’t point to a single positive graph and declare victory. You are accountable not just for how much value you extract, but for what kind of value you are creating. Some of your most valuable customer interactions — a repair that cost you money, advice that led someone to buy elsewhere, a warranty claim you honored without question — don’t appear on any traditional retail dashboard. But they show up everywhere else: in retention, in referrals, in the slow compounding of trust that becomes your only unassailable advantage.

This is my stab at a prototype of what it looks like when a retailer decides to build its future beyond what’s easy to count, toward what actually counts. I would shop there, how about you?

2. How far are we from this future?

To see how radical that 2029 retailer is, we have to start from where most organizations are today.

In the conventional model, value is dominated by what I’ve called the tyranny of numbers. What appears on dashboards — and how easily it can be aggregated into a spreadsheet — exerts an enormous gravitational pull on strategy, budgets, and careers.

For a typical retailer, that means:

• Revenue per visitor

• Conversion rate

• Average order value

• Cost per click

• Return on ad spend

• Inventory turnover

• Same-store sales growth

It’s easy to imagine a future retail experience because we are all consumers of retailers and wish they were different on a daily basis. For my consulting work, I’ve applied this type of thinking in many industries. Swap the store for a hospital and the coffee maker for a treatment plan. Swap it for a bank and the backpack for a mortgage. Swap it for a school and the conversion rate for a standardized test score. The specifics change; the core dysfunction doesn't. Every organization that serves human beings faces the same gravitational pull — toward what's easy to count, and away from what genuinely matters. And in every one of them, the scorecard tends to look something like this:

Customer “experience” might be represented by a satisfaction score. Trust might be approximated by review volume. Loyalty is reduced to repeat purchase rate. Everything is compressed into numeric proxies, because numbers feel objective, comparable, and controllable.

Inside this logic, the deeper work of retail, in this case — helping people find what they actually need, supporting products after purchase, building trust that compounds over decades — doesn’t fit easily into weekly sales reports. And what doesn’t fit the metrics tends to be sidelined.

3. The paradox and the financial shadow

The paradox is this: the behaviors that would build the deepest loyalty are often the ones that look worst in the spreadsheet. A sale you talk someone out of lowers conversion rate. A free repair costs money and generates no new revenue. Advice to buy elsewhere shows up as a lost customer. These are the retail equivalent of the bank that recommends paying down a competitor’s credit card. They’re the right thing to do. They’re also invisible — or worse, negative — in the traditional scorecard.

This paradox runs across every sector. A banker who talks a client out of a rushed investment — forgoing a commission today — earns the kind of loyalty that compounds into decades of managed assets and referrals. A physician who orders the more conservative test, resisting the pressure to over-diagnose, builds the patient trust that keeps people coming back before problems become crises, rather than after. A theme park that quietly upgrades a family's experience on a difficult day, with no fanfare and no upsell, creates the kind of memory that drives reunions, annual passes, and word-of-mouth that no advertising budget can replicate. In every case, the right action looks like a loss in the spreadsheet. In every case, it is actually the most valuable thing the organization did that day.

I can already hear the objection: these are good stories, but stories don’t move stock prices. Fair enough. Let’s talk numbers.

The data does exist — and it points firmly in the same direction. The 2018 McKinsey Design Index found that top-quartile design performers grew revenues 32 percentage points faster than their industry peers over five years, with total returns to shareholders 56 percentage points higher. The Design Management Institute’s longer-term study found design-led companies outperformed the S&P 500 by more than 200% over a decade.

But here’s what’s important to understand about those numbers: they are not measuring trust, or dignity, or the moment a banker talked a client out of a bad decision. They are measuring revenue growth and stock returns — the financial shadow that human-centered work eventually casts, once enough time has passed. They are lagging indicators. By the time they show up in the data, the real work — the compounding of trust, the accumulation of loyalty, the slow building of something competitors can’t copy — has already been done, invisibly, for years.

This is the honest case for human-centered investment: not that it will show up in next quarter’s numbers, because it won’t. But that it will show up — reliably, repeatedly, across industries and time horizons — in the numbers that ultimately matter most. If you need proof before you act, the proof is there. Just don’t mistake the shadow for the thing itself.

Coming up in Part IV:

That 2029 retailer is plausible. The economics are real. But I know what every leader reading this is thinking: “Sounds great. Doesn’t survive contact with my board.” Part IV is for you — a business case for what looks irrational in the short term, and a frank look at the two futures you’re actually choosing between.

More articles on the topics of metrics, innovation, and human values:

The Meta Trial: What Have We Been (and Should We Be) Building For?

Create Innovative Certainties: How Top Performers Navigate Uncertainty and Seize Opportunities for Growth

The Empathy-Throughput Tradeoff: Can AI-Enabled Scale Preserve Human Connection?

The Human Edge: The Enduring Value of (Certain) Human Skills

Running Faster to Fall Further Behind? How AI in Financial Services Can Miss Out on The Obvious

How Human Should The Future be?

In Part I, we named the tyranny of numbers: the way organizations treat what’s measurable as what matters, and sideline everything that doesn’t fit the spreadsheet. We saw how strategy consulting itself — once immune to this pressure — is now being asked to prove its value in outcomes-based terms. And we looked at what the data actually shows: design-led companies outperform, but the numbers are only a lagging signal of something deeper. So what is that something deeper? And why, even with compelling data in hand, does it so often get ignored?

Subscribe now

Video and image created with the help of midjourney

1. Three kinds of value and what lies beyond measurement

This distinction matters because it reveals what the tyranny of numbers gets wrong. The problem isn’t measurement itself. The problem is what we choose to measure — and on what timeframe.

There are (at least) three different kinds of value at play and here is a quick sketch:

The McKinsey and DMI data live in the middle column. They are powerful because they speak the language of business to make the case for a longer time horizon. They show that when you take design and human experience seriously as capabilities, the market tends to reward you.

But they are not the final destination.

The final destination is the right column: the recognition that some things are worth valuing even if they never show up in any index.

In 1968, Robert F. Kennedy delivered a famous critique of gross national product that captures this tension perfectly:

“Our gross national product counts air pollution and cigarette advertising, and ambulances to clear our highways of carnage. It counts special locks for our doors and the jails for the people who break them. It counts the destruction of the redwood and the loss of our natural wonder in chaotic sprawl. Yet the gross national product does not allow for the health of our children, the quality of their education, or the joy of their play. It does not include the beauty of our poetry or the strength of our marriages. It measures everything, in short, except that which makes life worthwhile.” — Robert F. Kennedy, Remarks at the University of Kansas (March 18, 1968)

This is the deeper truth.

We can measure conversion rates. We can track customer lifetime value. We can build models that correlate trust with revenue. And we should. A growing body of evidence suggests that companies that systematically invest in and measure aspects of human experience — through design, service, and relationship quality — tend to outperform their peers over longer horizons.

But the very things that make life worthwhile — love, joy, beauty, compassion — will never appear on any dashboard. They are not just hard to measure. They are, in an important sense, unmeasurable.

I’m not sure if it’s beyond the capabilities of capitalist organizations to go beyond obsessing over the first (and the second, if they will) column– after all, most companies are not morality institutions. But I would hope at least some leaders could entertain the idea, and to become aware of the difference between what can be measured, what can be measured if you’re patient, and what must be valued for other reasons entirely.

Later in this series I will try to explain why it’s not just the right thing to do, but actually good for the business too. I will get there in Part IV.

2. The asymmetry I’ve lived

I’ve spent my career watching this dynamic play out.

I come from a design and innovation background and have spent years in management consulting at the intersection of business strategy and human-centered work. I’ve seen how some forms of value are institutionally trusted while others are structurally discounted.

I joined McKinsey in 2019 for two reasons.

First, I wanted to learn how decisions really get made inside large organizations and see how human-centered work can play roles at the top level to make a difference.

Second, the firm had just published that 2018 research on the business value of design. Some of the most brilliant minds had synthesized years of data into something that felt like progress: an acknowledgment that design — or human-centered work — matters for businesses. It was inspiring to see that we could all speak a common language, see eye to eye, and find mutual respect for the good of everyone.

The reality over five years and dozens of projects was more complicated.

When it came time to sell work, staff projects, or justify budgets, the old gravitational pull reasserted itself. Human-centered work was still forced to contort and conform, to translate its value into the language of growth and efficiency. The very dimensions that made it distinctive — trust, experience, understanding people — got squeezed out in favor of what could be modeled in a spreadsheet. And I knew I wasn’t alone. Read this reflection by my fellow McKinsey alum, mentor and friend, Jalen Salazar, for his account of trying to make it work.

This asymmetry runs deep. Strategy’s intangibles — framing problems, aligning leadership — were accepted because they served promises business already understood. Human-centered work’s intangibles — trust, loyalty, wellbeing — were discounted because they served outcomes business didn’t yet know how to value.

The data has been there for over a decade. The outperformance of design-led companies has been there. And still, the operating reality continues to drag its feet — because the leaders making decisions refuse to see it.

3. What the crack reveals

The McKinsey shift toward outcome-based fees is a signal that the old trust is eroding for everyone. The economy, the heightened uncertainties in the world, and AI-enabled changes are pressuring many industries, including consulting, to be reevaluated.

Strategy’s intangibles were accepted because they served promises business already understood. Human-centered work’s intangibles were discounted because they served outcomes business didn’t yet know how to value. Now, as even strategy is asked to prove its impact, the asymmetry is collapsing.

The companies that outperformed in those studies weren’t just “creative.” They were systematic. They measured what others ignored. They invested in trust, in experience, in understanding people — and the market rewarded them over multi-year horizons.

But highly likely, they also understood something deeper. They understood that not everything that matters can be measured, and not everything that can be measured matters. They built organizations that could hold both truths: the rigor of measurement and the humility to know its limits.

I joined McKinsey hoping to help businesses better value human-centered work. Five years later, I left convinced that the problem isn’t that design can’t learn to play by the old rules.

It’s that the old rules are made to undervalue certain aspects of what we do — the qualitative, the human-centered, and often “what makes life worthwhile”.

What are we willing to value even when we can’t measure it? Kennedy’s speech reminds us that the most important things — the health of our children, the strength of our marriages, the beauty of our poetry — will never appear in any quarterly report. If our organizations become so captured by metrics that we forget this, we haven’t escaped the tyranny of numbers. We’ve simply become its most faithful prisoner servants.

The question should shift from whether human-centered work can learn to play by the old rules better, to whether it’s time to change the rules themselves, and to remember that some things are worth valuing precisely because they exceed our capacity to count them.

4. A choice to consider

When economic pressure increases, the crack is inevitable. The old trust is eroding. The data is clear enough: organizations that invest in the full human experience outperform over longer time horizons.

One response is to double down: keep forcing every contribution into the language of growth and efficiency, keep measuring only what’s easy to count, keep treating trust as “nice to have” until a crisis proves otherwise.

Another response is to ask a few different questions:

• What would it mean to build organizations that could see and measure what actually counts?

• Where are we using short-term metrics to judge long-term bets?

• What, if anything, do we value here even when it never shows up on a dashboard?

• What if we were willing to value things even before we figured out how to measure them well — and to hold space for the things we will never measure at all?

The crack is an opening. What we build through it is a choice.

Coming up in Part III:

We see the picture now. We know what doesn’t work even though it’s almost intentionally ignored. The thing is, once you see it, it’s hard to unsee it. You can blame me for this.

Let’s try to imagine what better looks like — not as a framework or a consulting deck, but as a lived experience. What would it actually feel like to be a customer, an employee, or a leader inside an organization that had already made the choice? Randomly, let’s pick retail - In Part III we will test a prototype: your shopping experience in 2029.

More articles on the topics of metrics, innovation, and human values:

The Meta Trial: What Have We Been (and Should We Be) Building For?

Create Innovative Certainties: How Top Performers Navigate Uncertainty and Seize Opportunities for Growth

The Empathy-Throughput Tradeoff: Can AI-Enabled Scale Preserve Human Connection?

The Human Edge: The Enduring Value of (Certain) Human Skills

Running Faster to Fall Further Behind? How AI in Financial Services Can Miss Out on The Obvious

How Human Should The Future be?

image/video created with the help of midjourney

This is a topic I’ve circled a few times before (see the list of older articles at the end). Then came the recent news about McKinsey’s business model shift—and it made me wonder.

There seems to be a crack forming in the old trusted system under the pressure of current economic climate. On one hand, that should send a shiver down the spine of me and my fellow consultants. On the other, maybe that’s a good thing. Maybe it’s the best thing.

A crack means there’s an opening. A chance to look at things differently. To build on the partial ruin of something that no longer stands.

I wrote and rewrote this piece only to realize it had grown far too long, so I’ve broken it into four parts. I’m making some bold claims in this series—intentionally. Please disagree, push back, argue. Nothing sharpens a position like being questioned.

1. The numbers

Every leader knows the mantra: “What gets measured gets managed.”

It’s repeated in boardrooms, embedded in performance reviews, carved into the ideology of modern management. The implication is clear: if you want something to improve, put a number on it. Track it. Report it. Reward it.

This is true, as far as it goes.

But there’s a corollary that never gets spoken aloud: What doesn’t get measured gets ignored.

This is the tyranny of numbers. It’s not that numbers are bad. It’s that what appears on dashboards — and how easily it can be aggregated into a spreadsheet — exerts an enormous gravitational pull on strategy, budgets, and careers. Over time, organizations begin to confuse what’s measurable with what matters, and what’s hard to measure became what can be sidelined.

For a typical business, this means:

• Revenue growth

• Cost reduction

• Conversion rates

• Return on ad spend

• Quarterly earnings

Customer “experience” might be represented by a satisfaction score. Trust might be approximated by complaint volume. Innovation might be reduced to features shipped. Everything is compressed into numeric proxies, because numbers feel objective, comparable, and controllable.

Inside this logic, the deeper work of building organizations — reducing fear and confusion, building trust that compounds over time, increasing the capacity to sense and adapt to change — doesn’t fit easily into quarterly reports. And what doesn’t fit the metrics tends to be sidelined.

The paradox is obvious in our lives, even if it’s obscured in our businesses. We would never judge a friendship by its quarterly return. We don’t quantify the worth of reading a novel in net present value. These experiences are valuable precisely because they exceed neat measurement. Yet when we step into organizations, we quietly accept a different rule: if we can’t count it, it doesn’t count.

This is the tyranny this article is about: letting what is easiest to measure dictate what we treat as valuable. It’s what happens when “what gets measured gets managed” becomes the only lens, and we forget that the most important things in life — and in business — or what actually counts, are often the hardest to count.

Subscribe now

2. The crack

For decades, management consulting operated on a premise clients trusted: pay for advice that promises business impact.

Between that advice and actual results lay a vast gulf of execution. Things got lost in translation. Priorities shifted. By the time numbers came in — if they ever cleanly mapped back to the original work — the strategists had long since moved on.

And yet, that strategy work was treated as valuable. It was billable. It was trusted. No one demanded a one-to-one causal link between every slide and quarterly earnings two years later. The trust was assumed, because the terms of the promise — growth, share, cost — were the terms business already operated by.

Beneath the measurable outputs sat intangible contributions that everyone accepted as real:

• Framing problems in ways that mobilize large organizations

• Creating narratives that align fractured leadership teams

• Orchestrating implementation across functions and silos

• Providing political cover for risky decisions

These contributions were not what clients explicitly paid for — they paid for the promise of business impact. But the intangible work of alignment and mobilization was the invisible engine that made that impact possible.

Recently, that old trust has begun to erode.

As Business Insider reported, McKinsey now has about a quarter of its global fees in value-based or outcome-based structures, versus the traditional engagement-based model. Clients are no longer simply saying, “Here’s a scope, what’s the fee for that?” They’re approaching engagements with a different framing: “Here’s the outcome we’d like to get to.” The fee then becomes contingent on the performance McKinsey can deliver.

This is a crack in the old model of unquestioned trust — and, if we choose to use it that way, an opening.

Strategy’s intangibles were historically accepted because they served promises business already understood. Now even strategy is being asked to prove its value more directly. The assumption that advice alone is worth paying for is eroding. Clients want results, not just recommendations.

If even the most institutionally trusted work is being pulled into outcomes-based logic, the natural question is: what kind of evidence will count next?

To answer that, we need to look closely at the evidence we already have — what it really shows about human-centered work, what it leaves out, and how far it can take us.

3. What the data actually shows (and doesn’t show)

Let me pause here and acknowledge something. I’m about to cite a bunch of numbers to argue against the tyranny of numbers. That’s either hypocritical or strategic.

I think it’s strategic — but only if we’re clear about what the numbers are actually telling us.

In 2013, the Design Management Institute created the dmi:Design Value Index, tracking a portfolio of design-led, publicly traded companies against the S&P 500. In its first iteration, those design-led companies had outperformed the S&P 500 by 228% over the prior decade — a figure that remained above 200% in every subsequent update.

In 2018, McKinsey published “The Business Value of Design.” They tracked 300 companies over five years and found that top-quartile design performers grew revenues 32 percentage points faster than peers, with shareholder returns 56 percentage points higher.

These numbers have been cited hundreds of times in meetings and PPT decks that I’ve seen and built myself, to prove that “design pays.”

But what they are measuring is intriguing at a second look.

The Design Value Index measures stock performance over time. The McKinsey index measures revenue growth and shareholder returns linked statistically to a design maturity score. Both are powerful, but both are still financial outcomes — just over longer time horizons than the typical quarterly report.

They were never designed to measure trust, wellbeing, or human flourishing. They measure whether companies that invest seriously in design and human experience happen to make more money.

That’s useful. It helps make the business case to leaders who only speak spreadsheet.

But we should be precise about what these studies show. They don’t prove that “design is valuable” in the sense of a broader human values level. They show that companies that behave as if human experience matters — and operationalize that through design — tend to outperform financially over multi-year horizons. The mechanism isn’t magic. Customers notice when you care. They reward you with loyalty, forgiveness, and word-of-mouth praises that draw more customers to you. Those behaviors eventually show up in revenue and returns — but only if you’re patient enough to wait.

In other words: the numbers are a lagging indicator of something that was always true but hard to count.

And these studies are part of a broader pattern. A 2023 study in the Journal of Relationship Marketing developed a CLV measure that explicitly incorporates nonfinancial constructs such as trust, loyalty, purchase frequency, recency, and churn, and found these factors play a significant role in explaining customer lifetime value. The authors argue that including these nonfinancial constructs expands traditional, purely financial views of CLV and highlights the need for further research on how such relationship qualities drive long-term customer value.

Coming up in Part II:

The data gives us a starting point — but it only takes us so far. In Part II, I want to go somewhere the data can’t follow: into the gap between what I’ve watched get proven and what I’ve watched get ignored, even inside the very institutions that funded the proof. That gap has a name. I call it the asymmetry. And I’ve lived it.

More articles on the topics of metrics, innovation, and human values:

The Meta Trial: What Have We Been (and Should We Be) Building For?

Create Innovative Certainties: How Top Performers Navigate Uncertainty and Seize Opportunities for Growth

The Empathy-Throughput Tradeoff: Can AI-Enabled Scale Preserve Human Connection?

The Human Edge: The Enduring Value of (Certain) Human Skills

Running Faster to Fall Further Behind? How AI in Financial Services Can Miss Out on The Obvious

How Human Should The Future be?

Video created with the help of MidJourney

In Part 1, we explored how the twentieth‑century “job”—a fixed bundle of tasks tied to a single employer—is dissolving under three forces: AI co‑pilots, the platformification of work, and a hard demand for purpose. We saw glimpses of what might emerge in its place: Gig Guilds, Human‑AI Pods, Civic Contribution models, and ultimately, more modular, problem‑centered work. If you haven’t read it yet, you can find Part 1 here.

Now, what might we need to redesign, and how?

If the workplace my daughter enters will be made of Problem‑Solving Clusters and hybrid teams, then two systems must evolve in sync:

1. How will we design work, talent, and leadership in our organizations?

2. How will we prepare people for hybrid human–AI problem‑solving through our education systems?

Both face the same uncomfortable but fundamental question that we should start with:

Are we optimizing for a past that’s disappearing (because we know it), or experimenting toward a future we can’t yet fully see?

Subscribe now

The Agentic Organization: From Blueprint to Beta

McKinsey argues that capturing the $2.9 trillion in AI‑driven value by 2030 “may depend less on new technological breakthroughs than on how organizations redesign workflows…and how quickly human skills adapt.”

We’re moving from organizations with fixed blueprints to ones that operate in a potentially perpetual beta — constantly testing and adapting their operating models. What might that look like in practice?

Redesigning Work: From Boxes to Flows

Instead of debating job descriptions, what if we started by mapping core workflows? McKinsey’s research shows large enterprises identifying around 190 key workflows across 16 functions. This shift in perspective raises different questions:

• Which parts of this workflow could benefit from AI augmentation versus human judgment?

• Where do we need deep specialization versus general orchestration?

• How do we design handoffs between human and AI contributors to maintain quality and trust?

Redesigning Leadership: From Supervision to Orchestration

As managers increasingly oversee systems where people, agents, and robots collaborate, their role transforms. McKinsey describes this shift toward orchestration. This raises questions about what “good management” looks like:

• How do you coach someone whose primary collaborator is an AI agent?

• How do you build psychological safety in teams that include non-human members?

• What metrics track the health and performance of human–AI collaboration?

Redesigning Careers: From Ladders to Portfolios

With around 72% of skills applicable in both automatable and non‑automatable work, the concept of a linear career path feels increasingly outdated. McKinsey’s analysis suggests workers will need to move through roles as their skills evolve.

This prompts different organizational questions:

• What would a talent marketplace look like that helps people navigate toward evolving opportunities rather than up fixed ladders?

• How do we reward skill acquisition and problem‑solving capability rather than just tenure or position?

• What happens to organizational memory and culture when people move more fluidly between projects and teams?

For my daughter, this evolving landscape suggests her future manager might be less of a traditional “boss” and more of a coach, resource connector, and orchestrator of multidisciplinary human–AI teams.

The Skill Stack: What’s Rising, What’s Shifting, What’s Anchoring Us

If by 2030, roughly one‑quarter to one‑third of work hours tied to the 100 most in‑demand skills could be automated, with digital and information‑processing skills most exposed and assisting and caring skills least affected.

This suggests we might think about skills in three evolving categories:

1. Sunset Skills: The Automating Foundation

These are highly automatable, narrow skills that will likely shrink in demand — think routine data entry, certain types of repetitive analysis, or standardized coding patterns. Their value isn’t disappearing entirely, but shifting from human execution to human oversight of automated processes.

2. Fusion Skills: The Evolving Partnership

These are skills that transform in partnership with AI — problem‑solving, writing, research, quality assurance, data analysis. Here, the work shifts from doing everything manually to supervising, directing, framing problems for, and interpreting outputs from AI collaborators. Demand for AI fluency specifically — the ability to use and manage AI tools — has grown nearly sevenfold in two years, making it the fastest‑growing skill category McKinsey tracks.

3. Anchor Skills: The Human Constant

These are deeply human skills that remain relatively automation‑resistant: leadership, empathy, conflict resolution, systems thinking, ethical judgment, creative synthesis. McKinsey finds interpersonal capabilities like negotiation and coaching sit in the lowest quartile of automation exposure.

For organizations, this evolving skill landscape raises practical questions:

• How do we build company‑wide AI literacy without it feeling like just another compliance program?

• What would internal talent marketplaces need to become true “skill navigators” rather than just job boards?

• How do we identify and develop anchor skills that often get labeled as “soft” but may become our most critical competitive advantage?

For the next generation, we might focus less on teaching specific tools and more on cultivating:

• The meta‑skill of learning new tools quickly

• The critical thinking to interrogate and improve AI outputs

• The creative confidence to combine domain knowledge, human judgment, and AI capabilities into novel solutions

Education for Problem‑Solving Clusters: From Memorization to Navigation

If we keep designing schools for memorization and standardized testing, we risk preparing children for the work AI is already becoming good at — while under‑preparing them for the work where humans will still excel.

Looking at these trends, what if education shifted from being primarily about knowledge transmission to becoming more about navigation skill development?

Project‑Based, Cross‑Disciplinary Navigation

Work in Problem‑Solving Clusters, Pods, and civic projects will inherently span domains. Research on Gen Z and millennial preferences shows they crave meaningful, connected work. This suggests students might benefit from more practice tackling open‑ended problems where success criteria are ambiguous and feedback is iterative — mirroring the complex challenges they’ll face professionally.

Collaboration with AI, Not Avoidance of It

Instead of banning AI tools, what if we taught students how to use them for research, drafting, coding, and designing — then how to critique and improve the outputs? The future workplace expects hybrid human–AI workflows; could classrooms become safe sandboxes for developing these collaborative muscles?

Explicit Development of Anchor Skills

Communication, conflict resolution, ethical reasoning, and systems thinking might need to move from the edges of curricula to their core. These are precisely the skills that McKinsey finds least exposed to automation yet most central to orchestrating hybrid teams. How might we assess and develop these differently than we assess factual recall?

Lifelong Learning as Navigational Literacy

With skills becoming more specialized and dynamic, the ability to reskill and move into adjacent roles may matter as much as any degree. What if we taught learning how to learn — including how to identify skill gaps, find quality resources, and build new competencies — as a fundamental literacy alongside reading and math?

This approach aligns with the insight from my earlier framework — we’re preparing students not for specific boxes in a 2x2 matrix, but for the ability to move fluidly between different modes of working as problems and opportunities evolve. When my daughter asks to learn digital drawing, perhaps the most valuable approach isn’t just teaching her today’s tools, but involving her in projects where she uses those tools to express ideas, solve problems, and collaborate — with classmates, with teachers, and yes, with AI systems as creative partners.

Responsible Oversight: Building Guardrails for Uncharted Territory

As we embed more capable AI into workflows, the question of governance becomes increasingly urgent. Anthropic’s “Responsible Scaling Policy” introduces AI Safety Levels — cumulative technical and operational standards that scale safeguards as systems become more capable. The core idea is both simple and profound: the more powerful the system, the stronger and more structured the safeguards must be.

McKinsey warns that key challenges remain around hallucinations, transparency, and explainability, which are critical to ensuring safety and avoiding unwanted bias as agents become more embedded.

This creates practical many questions for organizations experimenting with these new models:

• How do we evaluate AI systems before deployment when their capabilities are evolving rapidly?

• What does continuous monitoring look like for hybrid human–AI workflows?

• Where do we place the clear “human in the loop” checkpoints for critical decisions?

• How do we maintain organizational accountability when work is distributed across humans and increasingly autonomous systems?

For education, it suggests we need to teach our kids — not just our employees — how to use AI responsibly: how to question its outputs, understand its limitations, and recognize its potential risks alongside its benefits.

Questions Without Easy Answers

I’m filling out a questionnaire for my daughter’s application to the Gifted and Talented program at her elementary school. As a first time parents, I have no idea how kids are tested into such programs and am curious to learn about the standards and how they are measured at such an early age. The form asks for observations on how she learns, plays, and expresses herself — the subtle markers of a smart, curious mind.

One question gives me pause:

“Example(s) of how my child possesses a large storehouse of information about a variety of topics beyond the usual interest for their age.”

What’s “usual” for her age? I’m not sure. More importantly, why does a “large storehouse of information” still signal giftedness today?

I grew up in a system where reciting 500 ancient poems — most likely without understanding them — or 300 digits of π earned a child prodigy status. But in an era powered by AI, where recall is instant and synthesis is the new frontier, why are we still selecting and grooming kids for skills of memorization?

On one hand, I completely understand that the ability to possess such a large storehouse of information is a sign that the kid is smart and curious. But on the other, it’s too obvious in today’s world that the smartest person can’t out-smart machines if we define smart by some of the old rules. Shouldn’t the rules evolve accordingly? Shouldn’t we be making more room for empathy, creativity, problem-framing, and collaboration — the very capabilities that distinguish humans from machines?

Just like the skills to design in Illustrator, Solidworks, and Figma, I know it’s not that the skills won’t matter — it’s that how they matter is changing.

She won’t grow up to be a “designer” in the narrow, job-title sense. It will most likely be moot to master a sure-to-be outdated digital tool, or to understand the foundation of how that generation of tools works. Just like how it used to matter to learn to maneuver a horse-drawn carriage, and then to drive a car, be it a stick shift or automatic, so you can go to work, the bank, or the clinic, and now, many of us sit in a comfortable chair to get your work done, and occasionally call a (self-driving) Uber to meet with friends in a cafe; and in the future, driving might become “not a thing” at all. A designer’s job might become more about knowing what you want to build, and less about how, mechanically and manually, you will manifest that vision on paper (or screen).

She’s more likely to move through some kind of fluid Problem‑Solving Clusters where design, storytelling, data analysis, and AI orchestration intertwine. And her career won’t be a straight line; it will be a series of pivots — sometimes by choice, sometimes by necessity — that may well make her parents uneasy.

As my earlier framework suggested, she won’t occupy one box in a career matrix. She’ll navigate the intersections between physical and digital, routine and complex work throughout her life.

A Few Things We Can Do Today

I like to imagine and envision futures as much as any futurists, but I love to be able to bring it back to now: For leaders, educators, and parents, what might be some small steps we can take today to better understand tomorrow, and to test some assumptions for future success?

For organizations:

• What small experiment could we run this quarter to test a more fluid, problem‑centered way of working? - Which team and project would be the most suitable for this experiment?

• How might we map our key workflows to identify where AI collaboration could augment rather than just automate?

• What would need to be true for our people to feel excited rather than threatened by these shifts?

For education systems, teachers, and fellow parents:

• How can we create one small opportunity this month for kids to work on open‑ended problems that lack predetermined answers?

• Where can we safely introduce AI as a collaborative tool rather than a cheating risk this semester?

• How do we help the next generation develop both the technical fluency and the human judgment they’ll need starting from the classroom?

The future of work won’t be announced in a memo. It’s showing up in quiet, scattered experiments. The product manager who quietly assembles a cross‑functional Pod with an AI agent as a core member. The HR leader who prototypes an internal talent marketplace instead of fighting over headcount. The teacher who lets students co‑create with AI and then leads a discussion about what made the human contribution unique

Our opportunity — professionally and personally — might be to notice those experiments in our own contexts and ask: what if this small experiment contains clues to a much larger shift? What if the way we respond to these early signals shapes whether this transition feels like displacement or like the emergence of new, more human possibilities?

Before this quarter ends, what if we tried just one small thing?

• Could we charter a 60‑day micro‑team to tackle one cross‑functional problem? Give them permission to work differently, to experiment with AI tools, and to share what they learn — not just what they deliver.

• Could we sit with a child in our world and do one creative project with AI? Then ask them: “What do you think you did that the AI couldn’t? What did it do that surprised you?”

At the end of both experiments, maybe the most valuable question isn’t “Did we succeed?” but rather:

“What did we learn about how work and learning might need to change — for us now, and for the next generation coming after us?”

The answers might be messier than we’d like, as most true answers are. They might lead to more questions than solutions. But in asking them together, we might just start building the bridges between the world of work we inherited and the one our children will need to thrive.

Leave a comment

More articles on the topics of Future of Work and Future of Education:

Future of Work: Exploring Radical Shifts Beyond the 9 to 5

Microcredentials: Your Secret Weapon for a Future-Forward Career

A Podcast on Futures, and exploration of the Gigonomics

Futures of Global Growth: The Borderless Mobility Imperative

The Future of Jobs: Navigating Uncharted Waters in a Rapidly Evolving Workforce

The Human Edge: The Enduring Value of (Certain) Human Skills

Your 2035 Commute: Responsive Cities, Gig Mobility, and the Futures of Getting from A to B

The Great Relearning: What AI Reminds Us in Education and Work

The Future of American Education: Unpacking Trump’s Proposed Reforms and Their Long-Term Implications

video created with the help of MidJourney

A couple of weeks ago, when I told my five‑year‑old daughter that I used to design fun things like sunglasses (that super models wore on the runway, because you should always add spice to the story!), her eyes lit up and she asked me to teach her to draw on the computer. I hesitate. Part of me wants to throw open every digital tool I know. The other part wonders: will the specific technical skills that built my early career even matter when she enters the workforce twenty years from now?

I’ve seen my share of the job market dynamics over the past 20+ years. Some fields unfortunately shrink and others rise, while I reinvented my paths several times. What would some of the “cool jobs” of today morph into and what might grow out of the dust of the dying professions? What might be new ways of working, instead of “having a job” as we know it, for these future professionals to contribute to society and sustain lifestyles?

That question — what world of work we’re actually preparing our kids for — has been brewing in me for a while. Back in October, I wrote about “Rethinking Careers: A Futurist’s Framework for the AI World”, exploring a 2x2 dynamic matrix for understanding how work might evolve. I suggested that traditional career categories were becoming less useful, and that we needed new ways to think about the intersection of physical versus digital work and routine versus complex tasks.

A few weeks later, McKinsey’s major report “Agents, robots, and us: Skill partnerships in the age of AI” landed with strikingly similar thinking — distinguishing between physical and non-physical work, mapping what’s automatable versus what requires enduring human capabilities, and emphasizing hybrid partnerships over simple replacement. Their framework for understanding how AI agents (for cognitive tasks) and robots (for physical tasks) will partner with humans mirrors the physical-digital axis I explored in my earlier thinking.

Seeing those parallel conclusions reinforced something: we’re not just speculating about distant possibilities anymore. The cracks in our century-old “job” system are visible now. And the question has shifted from whether work will change to how — and how quickly we can build something better.

Subscribe now

The leadership team who is still debating remote versus office is missing the deeper shift. The next seismic change isn’t about where people sit, but how work itself is architected.

The twentieth‑century “job” — a fixed bundle of tasks tied to a single employer — is quietly dissolving. McKinsey’s report estimates that today’s technologies could, in theory, automate activities accounting for about 57% of US work hours. Yet, more than 70% of the skills employers seek are used in both automatable and non‑automatable work. This combination — high automation potential paired with enduring human capabilities — undermines the logic of static roles and nudges organizations toward something more fluid and problem‑centered.

You can see the cracks everywhere:

• In side gigs that siphon off your best people’s creativity.

• In Slack “tiger teams” that form outside the org chart.

• In the way high performers routinely work beyond their job descriptions.

In place of the rigid job box, something more modular and mission‑driven is emerging.

By 2030, adaptive organizations will talk less about “jobs” and more about Problem‑Solving Clusters: dynamic combinations of human talent and AI capabilities assembled around specific challenges, then reconfigured as needs change. This isn’t speculation; it’s the logical outcome of three powerful, data‑backed forces dismantling “the job” as we know it.

The Three Forces Dismantling “The Job”

1. The Superpower Force: AI‑Augmented Work

The core story isn’t “robots taking jobs.” It’s AI becoming a co‑pilot that handles routine tasks and amplifies human judgment.

McKinsey’s “Agents, robots, and us” finds that AI agents and robots could technically perform tasks representing about 57% of US work hours. Yet the real frontier is building hybrid human–agent teams, which could unlock an estimated $2.9 trillion in annual US economic value by 2030 if workflows are redesigned around them.

An earlier McKinsey report on generative AI estimated that by 2030, activities representing up to 30% of current hours worked could be automated. A Stanford and MIT field study of over 5,000 customer‑service agents, “Generative AI at Work”, found AI assistants increased productivity by 14–15% on average, with gains exceeding 30% for the least‑experienced workers. AI is acting as a skill multiplier, especially for those still learning.

McKinsey’s skill-level analysis shows that roughly 72% of skills appear in both automatable and non‑automatable activities, while interpersonal capabilities like negotiation sit in the lowest quartile of automation exposure. The story isn’t that skills disappear; they move up the stack — from executing tasks to orchestrating people, agents, and robots.

This evolution resonates with the framework I explored in my earlier article — the most valuable future work won’t fit neatly into traditional “high creativity” or “low creativity” boxes, but will emerge in the hybrid spaces where AI handles routine aspects while humans focus on judgment, synthesis, and relationship-building.

2. The Orchestration Force: The Platformification of Work

Work is increasingly found, matched, and delivered through digital platforms. This makes it less rational to staff permanent roles for every needed capability. Instead, you orchestrate access to skills — internal and external — on demand.

McKinsey’s workflow lens in “Agents, robots, and us” shows large enterprises rethinking work around workflows rather than roles, mapping around 190 workflows across 16 business functions. The question shifts from “Who owns this job?” to “Which mix of internal talent, external capacity, and AI agents should own this workflow right now?”

Upwork’s “Future Workforce Report 2022” illustrates this change: 56% of hiring managers increased their use of freelancers, and nearly 60% planned to rely on them even more. Internally, platforms like Unilever’s FLEX Experiences have unlocked over 300,000 hours of capacity and boosted productivity by 41%. Schneider Electric’s Open Talent Market similarly uses AI to match employees to gigs and mentors.

Taken together, “job” feels more like an interface than an identity.

3. The Meaning Force: A Hard Demand for Purpose

Younger workers are increasingly unwilling to trade meaning for money. They want autonomy, growth, and values alignment — and they’re acting on those preferences.

Deloitte’s 2024 Gen Z and Millennial Survey finds that about half of Gen Zs (50%) and 43% of millennials have turned down assignments that conflicted with their ethics, while 86% of Gen Zs and 89% of millennials say purpose is key to their job satisfaction.

A McKinsey-linked survey summarized in Forbes reports that 82% of employees believe it’s important for their organization to have a clear purpose, and 72% say purpose should be prioritized over profit.

For leaders, this suggests purpose isn’t an HR slogan; it’s becoming a measurable driver of who stays, who engages, and ultimately, who performs.

Three Glimpses of Work After “The Job”

These forces open a portfolio of possible work structures. As the saying goes, “The future is already here, it’s just unevenly distributed” (William Gibson), some of these are already seen in practice.

The “Gig Guild” – Strength in Numbers

Skilled professionals band together in member‑owned collectives. These digital guilds share resources, own their specialized AI tools, and present a unified brand — building the leverage, stability, and community missing in solo freelancing.

The Future, Already Here: Tech worker cooperatives are emerging where developers and designers are both practitioners and co‑owners. Writer‑focused collectives like Study Hall help freelancers share rates and navigate opaque industries. As AI tools become more powerful, these collectives can operate more like micro‑firms that combine human expertise with shared digital infrastructure.

The “Human‑AI Pod” – Your Mini‑Startup Inside the Org

Inside companies, the default unit shifts from static departments to small, empowered Pods. Each gets a clear mission and a budget to “hire” both human teammates and specific AI agents. The Pod manages itself, blending skills and silicon to prioritize learning and outcomes.

The Future, Already Here: This is the natural evolution of the “squad” model. The new element is treating AI as a formally resourced teammate. In customer service, generative‑AI assistants are already embedded in workflows. McKinsey’s research highlights similar patterns in sales and IT, where specialized AI agents handle prioritization and first‑pass work, freeing specialists to spend 30–50% more time on relationship‑building. And in practice, McKinsey recently announced at CES 2026 that the firm has integrated approximately 25,000 AI agents into its workforce of 40,000 human employees, with plans to reach a 1:1 ratio by year-end. These agents,, utilized heavily through their AI arm, QuantumBlack, handle autonomous tasks like research and workflow management.

In parallel, demand for AI fluency has grown nearly sevenfold in US job postings over roughly two years, making it the fastest‑growing skill category McKinsey tracks. These Pods represent exactly the kind of hybrid, fluid team structures that my earlier framework suggested would emerge — combining physical and digital capabilities, routine and complex tasks, all oriented around solving specific problems rather than filling predefined roles.

The “Civic Contribution” Model – Work for a Wider World

A slice of professional effort decouples from corporate profit motives and organizes around societal problems. Individuals contribute verified “work units” to civic or environmental projects, building portable portfolios of impact funded via hybrid models.

The Future, Already Here: Platforms like Terra.do connect professionals to climate‑focused roles. Catchafire matches skilled volunteers with nonprofits. Government programs like the U.S. Digital Service use explicit “tour‑of‑duty” models for public problem‑solving. These are functioning prototypes of mission‑driven, time‑bound contribution at scale.

As more professionals build parallel portfolios of civic impact, employers will increasingly compete not just on salary, but on how everyday work connects to broader societal missions.

So where does this leave that five‑year‑old who wants to learn digital tools because she hears my stories about design?

In the world that is emerging, the job is less a destination and more a temporary configuration of skills, people, and agents around a problem. The real questions shift from:

• “What do you want to be?” to

• “Which problems do you want to help solve — and what mix of human and AI capabilities will you bring to them?”

This aligns perfectly with the insight from my earlier framework — we’re moving from static career categories to dynamic configurations where the work determines the team, not the other way around. The boxes in that matrix aren’t job titles; they’re modes of working that people will move between throughout their careers.

Three forces are making that shift unavoidable:

• AI co‑pilots that unbundle tasks and rebundle them into new workflows

• Platforms and marketplaces that orchestrate skills across boundaries

• A generation of workers unwilling to sacrifice purpose and growth

That brings us to the second half of this series: not why “the job” is ending, but how we might design something better — for our organizations and for our children.

More to come!

Subscribe now

As I’d just finished writing this, I came across this research report, with a fantastic chart (see below) to show AI exposure and adaptability of occupations. I especially like the callout on gender split for these occupations, and how there’s a disproportionally high percentage of female workers in the category most at risk, and low percentage in the category least at risk. Note this chart is interactive, so go and play with it.

More articles on the topics of Future of Work and Future of Education by TBD Futures:

Future of Work: Exploring Radical Shifts Beyond the 9 to 5

Microcredentials: Your Secret Weapon for a Future-Forward Career

A Podcast on Futures, and exploration of the Gigonomics

Futures of Global Growth: The Borderless Mobility Imperative

The Future of Jobs: Navigating Uncharted Waters in a Rapidly Evolving Workforce

The Human Edge: The Enduring Value of (Certain) Human Skills

Your 2035 Commute: Responsive Cities, Gig Mobility, and the Futures of Getting from A to B

The Great Relearning: What AI Reminds Us in Education and Work

The Future of American Education: Unpacking Trump’s Proposed Reforms and Their Long-Term Implications

Video created with the help of MidJourney

The Stress Test Of The “Unquestionable”

If you’re a leader, you know the playbook by heart: measure everything, optimize for engagement, scale what works. It’s the same logic that built your business—and every major app on your phone. For two decades, it seemed unquestionable.

I’ve banged my head against that word—unquestionable—too many times. I know how hard it is to bend the rule. As a designer turned strategist, I’ve lost count of how many human-centered solutions were overruled by something more measurable, more scalable, more aligned with the immediate business impact. It’s the only thing stakeholders seem mandated to care about.

Then a civil jury in Los Angeles was asked to decide whether those same tactics—applied to children—amount to a defective and dangerous product.

In mid‑February 2026, Mark Zuckerberg walked into that federal courtroom and faced a question that should have been simple: do people tend to use something more if it’s addictive? As reported by NBC News on February 18, 2026, he hesitated, agreed that addiction is bad, but sidestepped the direct comparison to cigarettes and opioids.

What made the moment historic was not just his answer, but the setting. As CNN reported on February 22, 2026, this is the first time the architect of the modern attention economy has answered to a civil jury over allegations that social media platforms deliberately addict and harm children, rather than to Congress or investors.

Outside, the stakes were even more visible. According to First Amendment Watch’s February 22, 2026 coverage, parents who lost children to suicide lined up for hours hoping to get a seat—there not for legal doctrine, but for acknowledgment that the systems they blame for their children’s deaths are finally being challenged at their core.

This is not simply a trial about one app or one executive. It is the first serious stress test of the design logic that has governed social platforms—and much of our digital economy—for the past decade.

And it forces a harder question, against the “unquestionable”: when technology can do virtually anything, are we designing with humans at the center, or for the economy of the business?

Subscribe now

How We Got Here: Small Decisions, Large Consequences

The journey to this courtroom began with a thousand incremental choices that, taken together, locked in a system optimized overwhelmingly for attention rather than well-being. Those parents in the gallery didn’t arrive because of one bad decision. They arrived because of years of them.

In its report on Zuckerberg’s testimony, NPR on February 18, 2026 described internal Meta documents from the mid‑2010s showing executives celebrating double‑digit jumps in time spent on Facebook and Instagram, and framing “negative teen trends” as a pressing problem to reverse. The same article noted that internal materials explicitly discussed “teens” and “tweens” as strategic segments. The New York Times reported on February 18, 2026 that leadership closely tracked teen usage as critical indicators of platform health.

At the time, each of these decisions looked like standard growth logic: maximize engagement, win the next generation, hit the numbers. But while executives celebrated metrics, parents were watching their children disappear into feeds engineered to hold them.

Concerns surfaced internally too. As NPR also noted, Meta’s own experts warned as early as 2019 that Instagram’s beauty filters could worsen body image issues among teens. Yet according to CNN’s February 9, 2026 reporting, the company first restricted certain filters, then relaxed those limits as TikTok surged—a trade-off between safety and competition that the parents in the gallery now understand intimately.

The question of under-age users added another layer. PBS NewsHour reported on February 9, 2026 that internal documents show Meta knew for years that millions of children under 13 were on its platforms. NBC News described one internal estimate from around 2015 suggesting millions of under‑13s had accounts, often created with false birthdays. When questioned, executives pointed out that age is difficult to verify—a technically accurate answer that sits uneasily next to the sophisticated behavioral systems these same companies use for micro-targeted advertising.

By the time the first youth addiction trial began, plaintiffs’ lawyers were ready with a metaphor that would resonate with any parent. In opening arguments, CNN on February 9, 2026 quoted attorney Mark Lanier describing Instagram and YouTube as “digital casinos,” likening endless scroll to pulling a slot-machine lever. And in a February 17, 2026 overview, Sokolove Law quoted mass‑tort lawyer Jayne Conroy—who worked on opioid cases—saying the dopamine reaction behind social media compulsions is “not really all that different” from opioid addiction.

Where We Are: One Plaintiff, Many Stakes

The case now on trial, K.G.M. v. Meta et al., is formally about one young woman. As CNN reported on February 22, 2026, the plaintiff—identified as “Kaley” or “KGM”—began using YouTube at age six and Instagram at nine, and alleges that years of compulsive use contributed to anxiety, depression, and body dysmorphia.

But the stakes extend far beyond her. This is one of the first “bellwether” trials—test cases meant to indicate how more than 1,500 related lawsuits against Meta, Google/YouTube, TikTok, and Snap might fare. NBC News reported on February 9, 2026 that Snap and TikTok settled Kaley’s claims before trial, while Meta and YouTube chose to contest the allegations before a jury.

The legal theory is a deliberate pivot. PBS NewsHour explained that plaintiffs are avoiding traditional content‑based claims—which run into Section 230—by focusing on design elements like infinite scroll, autoplay, and algorithmic recommendations, arguing that these features make the product itself unreasonably dangerous for minors.

A February 7, 2026 analysis from TechPolicy Press noted that this approach aligns with emerging legal scholarship treating “addictive design” as a product defect. An article in the Emory Law Journal on “Addictive Design, Section 230, and the Freedom of Speech” argued that courts can distinguish between hosting content (which enjoys protections) and designing manipulative interfaces (which can be regulated).

In this trial, Zuckerberg has been pressed on that distinction. NBC News reported that attorneys confronted him with internal documents referencing “tweens” as a desirable demographic and with data showing that only a tiny fraction of teens actually use Instagram’s voluntary time‑limit tools. CNN’s February 18, 2026 story described how plaintiffs used these facts to argue that Meta’s well‑being features were designed more for optics than impact.

Outside the courtroom, sentiment is shifting. A YouGov poll summarized by Newsmax on February 17, 2026 found that after seeing trial evidence, 86% of Americans believed Meta and Google should be held accountable for their role in the youth mental health crisis.

As CNN noted, the trial runs into March, with YouTube’s defense still ahead. Whatever verdict emerges will be appealed. But the parents in the gallery aren’t waiting for appeals. They’re waiting for something the legal system rarely delivers: acknowledgment.

Three Futures, One Question

At this point, the question for leaders isn’t just “What happens to Meta?” It’s: What does this moment do to the entire ecosystem over the next five years—especially as AI makes these systems more powerful and more personal?

Across all three futures below, the same human question echoes: when technology can do anything, who decides what it should do?

Future One: Meta Loses—The Discovery Era

In its February 22, 2026 piece, CNN described this as a potential “Big Tobacco moment.” The real impact may come less from a single verdict than from years of document releases that reshape public narrative. The New York Times has already shown how internal slide decks about teen engagement are influencing perceptions.

According to the Emory Law Journal article, once courts accept that interface patterns can be defects, plaintiffs will test similar theories against other digital products. A 2025 alert from Mitchell Silberberg & Knupp noted that earlier “video game addiction” suits were often dismissed, but design‑focused arguments might find a receptive audience.

In this world, AI becomes central. Reuters reported on August 29, 2025 that Meta planned AI‑based safeguards after an investigation raised concerns about harmful content reaching minors. Those same systems will appear in future trials—both as evidence of what companies could have done earlier, and as proof that AI can enforce safety.

For the parents in the gallery, this future offers something: the possibility that their children’s names won’t be forgotten, that the documents forced into light might prevent more families from joining the line outside.

Future Two: Meta Wins Narrowly—The Rulebook

According to the Emory Law Journal analysis, US courts are often wary of expanding tort law into new digital domains. If Meta and YouTube emerge with narrow wins, courts signal they’re not the primary venue for reshaping platform design.

But policymakers are already moving. Reuters detailed on February 17, 2026 how Australia’s under‑16 ban took effect December 2025, with Germany, Greece, Malaysia, and Indian states considering similar measures. In the US, Reuters reported on December 26, 2025 that New York is moving to require mental‑health warning labels on feeds using infinite scroll and algorithmic recommendations for minors.

In this world, AI becomes a compliance tool—and a point of contention. Companies will lean on AI‑driven age checks and usage nudges to argue they can police themselves. Regulators will ask whether those same systems identify the most vulnerable users for more compelling engagement.

For the parents, this future is colder. Change comes through rulebooks, not acknowledgment. But rulebooks, if written well, might protect children they’ll never meet.

Future Three: AI‑Native Platforms—Design Reset or Dark Arms Race

While this case is about current platforms, the next five years will fuse AI and social media into something more fluid. Reuters catalogued how fears about “addictive” AI‑enhanced feeds are driving teen bans across multiple countries.

Both CNN’s analysis and TechPolicy Press have pointed out that recommendation algorithms and generative AI will soon allow platforms to generate synthetic content and even “friends” tailored to each user’s psychological profile.

Another front is opening in schools. First Amendment Watch reported that school districts are suing under “public nuisance” theories, arguing platforms force districts to spend more on mental‑health staff. Mediapost noted on February 9, 2026 that a federal judge allowed one such case to proceed to June 2026 trial.

If some succeed, they add momentum to a design reset: OS‑level age verification, default safety features, audited recommendation engines. But if regulation lags while AI accelerates, a darker path emerges: large incumbents tame flagship products while a proliferating ecosystem runs a quiet arms race in engagement—synthetic influencers, adaptive notifications, generated content designed to hold young users just a bit longer.

This future is the hardest to watch. Children were harmed by systems now considered primitive. What happens when those systems become truly intelligent?

The Question at the Center

Throughout the trial, the parents have remained. First Amendment Watch described them waiting in the cold, hoping to hear Zuckerberg answer what Meta knew and when it knew it.

Attorney Matthew Bergman of the Social Media Victims Law Center—representing more than 1,000 plaintiffs—recalled: “We started doing this four years ago, no one said we’d ever get to trial. And here we are trying our case in front of a fair and impartial jury.”

As NBC News reminded readers, just a few years ago the idea that Zuckerberg would be cross‑examined in a civil trial about harms to children seemed almost unimaginable.

The future arrives not with a product launch or an earnings call. Sometimes it arrives with a subpoena—and with parents waiting in line outside a courthouse, trying to make sure it doesn’t arrive too late for someone else’s child.

Their presence is the answer to the question this trial poses. When technology can do virtually anything, we design for humans at the center—or we don’t. And if we don’t, eventually, those humans will find their way to the courthouse steps.

The jury will deliberate. The appeals will run. The futures will unfold. But the parents will have done what they came to do: remind everyone inside that courtroom, and everyone reading about it afterward, that this was never just about design.

It was always about them, and should be about all of us.

Video created with the help of Midjourney

If you’re a regular reader, you know I usually dedicate these letters to a deep dive into a specific industry, technology, or cultural phenomenon. That work is core to what I do, and I love it. But I’m also a firm believer in the discipline of zooming out. There is immense value in stepping back to connect the dots across domains, to see the bigger picture that could help us make sense of individual trends in a given area.

So I’ve decided to periodically write a dedicated, synthesized trend issue— please tell me if this is interesting or useful for you! The process I use—scouring global reports, looking for recurring signals and dissonant notes—is a practice I conduct regularly. As I wrote in “Melody in the Noise: A Futurist’s Guide,” the art lies not in predicting future event, but in listening for the patterns. It’s in the subtle convergence of data points from tech, retail, sociology, and public policy where the most important narratives—the melodies—emerge. My hope is that this helps us move from reactive confusion to proactive clarity.

To that end, I’ve spent a few dozen hours reading through a selection of trend reports for 2026. While this is by no means a definitive or exhaustive survey of the entire landscape, I find that even a small, curated sample pool can be incredibly revealing. The patterns that emerge across different domains—from consumer tech to civic planning—often point to deeper, underlying shifts in our collective psychology and systems.

Synthesizing 2026: Patterns of Transition and Response

One consistent pattern for 2026 emerges clearly. Across the reports, a common narrative emerges: we are navigating a period of significant transition, marked by the strain of old systems and a fascinating, urgent, and profoundly human response.

Some reports describe this as a period of “metamorphic change,” where the disintegration of established systems is serving as a catalyst for new forms of human ingenuity and technological integration. Synthesizing insights from reports on planning, retail, consumer experience, and culture, a set of recurring trajectories comes into focus. These are not predictions, but rather interconnected themes that, together, sketch a picture of the tensions and opportunities defining this moment.

Let’s explore the ten interconnected themes that surfaced from this synthesis.

Subscribe now

I. The Technological Foundation: Evolving Tools and Enduring Tensions

1. The Shift Toward Agentic Technology

A recurring observation is the transition from tools that answer questions to agents that act autonomously. According to the Fashion Retail 2026 Guide, we are moving “from browsing to delegating.” This is crystallizing in commerce as “Acommerce,” where AI agents handle discovery, fit, and payment independently, a concept detailed in The Future 100 2026.

2. The Persistent Human-AI Readiness Gap

In parallel, multiple reports highlight a stubborn gap between technological deployment and human satisfaction. Data from the 2026 Consumer Experience Trends Report is particularly revealing: nearly 1 in 5 consumers said AI-powered customer support provided no benefit at all. The synthesis suggests a critical juncture: the next phase of progress hinges on moving beyond cost-cutting automation to designs that pair machine intelligence with human intuition, using AI to augment rather than replace human judgment in complex scenarios.

II. The Evolving Landscape of Truth and Trust

3. The Push for Radical Honesty

A striking point of convergence across reports is a crisis of truth and a corresponding hunger for transparency. The Future 100 2026 notes that 81% of people believe truth is an “endangered concept,” exacerbated by AI. This environment creates what the 2026 Consumer Experience Trends Report calls a “privacy-personalization paradox”: consumers want tailored experiences but deeply distrust how their data is used. The synthesized insight is that credibility will increasingly depend on organizations acting as “brand bastions” of radical honesty and tangible data stewardship, a need also underscored for public data in the 2026 Trend Report for Planners.

III. Human Responses: New Patterns of Coping and Connecting

4. Digital Fatigue and the “AI Analog” Movement

A clear pattern emerging from consumer and trend reports is a pivot away from screen-centric engagement. The desire for value “without requiring screen time” (per the 2026 Trend Report by Trend Hunter) is fueling the “AI Analog” trend—intelligence embedded into physical objects. This aligns with the rise of “Fashion Cafes” (The Future 100 2026) and forecasts like WGSN’s “Strategic Joy,” which identifies a growing consumer pursuit of playful, low-friction, and tactile experiences as an antidote to digital overload.

5. “Treatonomics” and the Intent-Behavior Gap

Synthesizing consumer data reveals a compelling contradiction: a “big gap between intent and behavior.” While people voice aspirations like sustainability, their spending often prioritizes personal comfort and immediate uplift. This gives rise to “Treatonomics”—the small, spontaneous purchase as a coping mechanism (The Future 100 2026). This behavior pattern shows in WGSN’s concept of “Witherwill,” a longing to be free from relentless responsibility, suggesting consumption is increasingly driven by emotional regulation.

6. Experiential Environments as Social Hubs

Reports consistently underscore a heightened value placed on connection. With data suggesting 87% of people now value experiences that foster deep connection (The Future 100 2026), physical spaces are being reimagined. The trend of “Retailtainment,” which we have seen in the past years, using immersive, multi-sensory concepts to create social destinations (per the 2026 Trend Report), reflects this continued shift. It points to a future where the primary function of a space is not just commerce, but facilitating the “collective effervescence” of shared experience.

IV. Macro-Systemic Shifts in Work and Governance

7. Navigating Institutional Uncertainty

Analysis of civic and planning reports points to a significant rebalancing of responsibility. In the face of complex challenges and federal instability, the 2026 Trend Report for Planners suggests that states and localities will increasingly take the lead in community resilience. The synthesized takeaway points toward more distributed, local models of problem-solving and innovation. I’m very curious about how this will manifest in today’s political and social environment. Would definitely keep a keen eye on it.

8. The Redefinition and Fluidity of Work

The future of work, as depicted across these reports, is characterized by fluidity. Automation is making work “more project-based and boundaryless” (The Future 100 2026), elevating uniquely human skills. Interestingly, this abstract shift is accompanied by a tangible romanticization of 80s corporate aesthetics among Gen Z (per the 2026 Trend Report), suggesting a nostalgic search for identity and structure amidst the flux.

V. Evolving Frontiers of Wellness and Culture

9. Precision Wellness and Health Autonomy

A clear trend is the continued pivot toward hyper-personalized, data-driven health ownership. Status is being framed around “longevity leaderboards” and biometric validation (The Future 100 2026). This creates ripple effects like “downsized dining” in the food industry, where brands tailor portion sizes and nutrient density (per the 2026 Trend Report), reflecting a more surgical approach to consumption.

10. Borderless and Democratized Creativity

Finally, reports point to a decisive decentralization of cultural influence, with hubs like Riyadh, Lagos, and Seoul gaining prominence (The Future 100 2026). Coupled with AI as a “creativity catalyst,” this empowers a new generation of “sovereign creators.” Navigating this new landscape, as foreseen by WGSN, may require “Suspicious Optimism,” balancing excitement about technology’s potential with a healthy skepticism of its risks.

Overall, it looks like a well-needed reassertion of human needs… I hope it continues…

What emerges from connecting these dots across a sample of 2026 reports? A narrative of human agency responding to systemic pressure. The trends are less about technology imposing a future and more about people seeking to reclaim trust, joy, authentic connection, and control within increasingly complex systems.

While drawn from a limited set of sources, these synthesized patterns offer a valuable lens. They highlight the critical choices ahead: between automation and augmentation, between opacity and transparency, between isolation and connection. The strain on old systems is evident, but so is the creative, human-centric energy rising to meet it.

By listening for these patterns, we don’t get a definitive roadmap, but we do gain a sharper sense of the terrain we’re all navigating—and the agency we have to shape it. What do you see in your field? What makes you excited, wonder, or worry? Is your organization ready for these trends or even leading them? More next time!

Here’s to navigating with curiosity and clarity.

Best,

Lucy

Subscribe now

Most will dismiss this as another piece of ironic Silicon Valley performance art. The founder himself, when called out for the platform’s dystopian vibe, simply replied: “lmao yep.” It’s riddled with placeholder data and absurd tasks; it’s more concept than company. But a futurist should see something else: an artefact of the future, washed up on the shores of our present. It’s crude and buggy, but it’s also a revealing prototype of a seismic shift coming to work, autonomy, and the structure of the economy. Let’s dissect this artefact to understand the world it implies.

Subscribe now

The Prototype Decoded: More Than a Meme

First, let’s understand what RentAHuman.ai actually is. It’s not a job board for humans. It’s a tool built for AI agents—autonomous programs like AutoGPT or Claude that can plan and execute multi-step digital tasks. These agents are moving from novelty demos to real workflows in areas like customer support, operations, and software engineering, but they hit a hard limit: they can’t act in the physical world. RentAHuman proposes to be their bridge.

It’s built with integrations like the Model Context Protocol (MCP), a standard that lets AI agents “see” and use external tools—MCP is an emerging protocol that allows AI systems to safely plug into external services through a common interface. Here, the tool is you.

The platform operates on a simple loop:

1. An AI agent, following its owner’s goal, needs a physical action performed.

2. It browses RentAHuman, selects a human based on skills/location/price, or posts a “task bounty.”

3. The human receives instructions from the AI (not a person), performs the task, and submits proof (like a photo).

4. The AI releases payment from its crypto wallet.

It’s a stunt, but within its mechanics lies a serious proposition: a first draft of a general-purpose Physical World API—a way to call “real-world functions” (people, places, actions) as easily as we call cloud services today. It asks: what if we could programmatically request any physical action the same way we request data from a server?

This isn’t a wholly new idea. It’s the logical, if extreme, culmination of trends we already accept: the gig economy’s app-based labor (Uber, TaskRabbit), the rise of autonomous AI agents, and programmable crypto payments. RentAHuman simply connects these dots in a provocative, uncomfortable way.

The Critical Fault Lines

Before we scale this vision, we must probe its inherent cracks. This artefact raises questions our society is utterly unprepared to answer.

The Money Trail. If no human is directly pulling the trigger, where does the crypto come from? The answer is a human or corporate wallet funding an autonomous agent with a spending allowance. Imagine giving a corporate procurement AI a credit card with no pre-set spending limit. The potential for absurd, runaway, or fraudulent spending by goal-optimizing agents is staggering. For CFOs and risk officers, this looks less like “automation” and more like unleashing semi-autonomous spend across thousands of micro-contracts, each with physical-world consequences but no obvious owner.

The Liability Black Hole. Who is responsible when this chain breaks? If an AI hires a human who gets injured, causes damage, or commits a crime while following instructions, who bears the liability? The human worker? The platform? The AI’s owner? The AI’s developer? Existing labor, platform, and product-liability law were not designed for chains where a non-human agent assigns, supervises, and pays for work, which means early cases will likely be resolved in messy, unpredictable ways.

Alignment on the Specifics. Our current fears about AI alignment focus on superintelligences. This presents a more immediate menace. A well-meaning but simplistic agent, tasked with “making my party memorable on $2000” might rationally hire 100 strangers to show up at your house, or commission a dangerously large fireworks display. We already see “specification failures” in simple automation—like bots that spam customers or mis-price items—because humans didn’t spell out every edge case; now imagine that, but with fireworks, food safety, or building access. How do you instill ethics, nuance, and common sense in a process optimized for cost and efficiency? Or do you need to specify everything—and how feasible and efficient is that?

The Value of the Middleman. Is this just a complicated, worse version of TaskRabbit? For simple tasks, yes. The revolution only arrives when the AI middleman adds unique value: by orchestrating complex workflows beyond human planning bandwidth—managing dozens of interdependent physical tasks across a city in real time, or acting as an intermediary between other autonomous systems. Think less “book me a TaskRabbit” and more “coordinate 30 pop-up activations across three cities this weekend, each staffed, stocked, and instrumented for data collection, adjusting in real time to weather, traffic, and sales.”

Three Scenarios from the Artefact

Artefacts of the future show us a glimpse of a future possibility, with much complex context implied. What’s really going on in these future worlds seems like a vaguely visible backdrop blurred on the horizon like a city skyline. It works in sci-fi films but doesn’t help us understand the details: what’s possible in this world, why it is possible, what needs to be true for this future to happen?

Using RentAHuman as our seed crystal, let’s grow three distinct future scenarios, each requiring progressively more technological, legal, and social evolution.

Scenario 1 (Probable Near-Future): The “Managed Gig-Plus”

The meme fades, but the model is refined into specialized, trustworthy verticals. Startups strip away the irony and focus on domains where AI orchestration adds clear value: complex local logistics (e.g., “restock my home office with these 15 items from 8 different stores”), coordinated data-gathering missions (e.g., “document the state of playground equipment in these 10 ZIP codes this week”), or hyper-localized event staffing. The AI acts as a super-efficient, always-on personal operations manager, quietly stitching together human labor, local merchants, and digital systems. For leaders, this is where your operations, facilities, and customer experience teams quietly start to rely on AI-orchestrated local labor, even if they still think of it as “just another vendor.”

You can imagine a marketing team spinning up a local product-launch blitz across multiple cities entirely through an AI assistant: one prompt triggers dozens of micro-engagements—street teams, in-store demos, field surveys—each staffed by rented humans who never interact with a human manager at the client company. In customer experience, agents could trigger in-person “white glove” interventions when sentiment analysis flags high-value customers at risk, dispatching local gig workers to deliver replacement products, gifts, or on-site assistance.

Why It’s Possible. The tech stack exists: mobile apps, GPS, payments, identity verification, and increasingly robust agent frameworks. Consumers and small businesses already accept algorithmic matching and app-mediated gig work as normal. Short-horizon planning—“solve this today or this week”—is exactly where today’s agents perform best. From the business side, this scenario looks like an incremental upgrade to today’s gig platforms: better routing, smarter bundling of tasks, and lower coordination overhead.

What Needs to Be True. Trust and safety systems must mature significantly. This requires robust identity verification for both workers and requesters, insured bonding for workers and tasks, and AI agents that are reliable enough at task decomposition to be useful, not frustrating. Platforms will need granular controls so that humans can accept or reject specific tasks, escalate unclear instructions, and flag unsafe or unethical requests. Legally, platforms will likely be pushed to absorb liability as “employers of record” for micro-tasks, especially in jurisdictions already scrutinizing gig classification. Regulated industries—finance, healthcare, critical infrastructure—will demand auditable logs of agent decisions and human actions. Extending existing contractor-risk regimes to AI-orchestrated micro-contracts will require new playbooks for due diligence, insurance, incident response, and worker protections.

For executives, Scenario 1 is where the temptation to “just pilot it” will be strongest—local, tactical, seemingly low risk. The organizations that benefit will be those that treat it as a strategic capability from day one, with clear principles for where human discretion is mandatory and where agents are allowed to fully automate.

Scenario 2 (Seismic Shift): The Autonomous Enterprise Layer

Scale shifts from personal assistants to core corporate operations. A supply chain AI doesn’t just route trucks; it autonomously contracts local, gig-based inspectors at remote warehouses, hires temporary security for unexpected delays, and commissions freelance technicians for emergency repairs—all within its operational budget and without human approval for each step. An after-sales service AI might automatically dispatch in-person repair crews, schedule follow-up visits, and trigger proactive recalls when it detects emerging failure patterns in field data. Finance agents negotiate micro-contracts with service providers, optimize payment terms, and flag anomalies across thousands of AI-initiated transactions.

The economy sees the rise of M2H2M (Machine-to-Human-to-Machine) transactions as a normal operating pattern. Humans become specialized, high-reliability “modules” in vast, automated processes. Think of field engineers, inspectors, auditors, crisis responders, and fixers as dynamically “loaded libraries” into otherwise software-dominant workflows—called only when reality resists full automation, and then evaluated, scored, and re-integrated into models.

Why It’s Possible. Enterprise resource planning (ERP) and workflow systems have been moving toward greater automation for decades, from RPA bots to event-driven architectures. Agentic AI adds a decision-making and planning layer on top of these systems: agents can observe state, reason about goals and constraints, and act across multiple tools. Analysts already forecast that a significant share of enterprise applications and workflows will embed autonomous agents by the end of this decade, with a meaningful fraction of day-to-day decisions made without direct human initiation. As agents become a standard feature of CRM, ERP, HR, and ITSM platforms, giving them the ability to contract physical work will feel like an incremental extension, not a revolution.

What Needs to Be True. This scenario requires a quantum leap in AI reliability and corporate trust. Agents must be able to interpret complex policies—compliance, safety, ESG, brand risk—and apply them consistently across edge cases. That implies:

• Formal policy representations agents can reason over.

• Simulation and “sandbox” environments to test agent strategies before they touch reality.

• Human governance layers that can monitor, override, and retrain agents when they drift.

New legal frameworks defining “Agent Liability” will be needed to clarify responsibility chains when an enterprise agent initiates work that leads to harm or loss. Case law is already evolving around AI systems acting as de facto agents in hiring and decision-making; extending that logic to agents that instruct humans will raise fresh questions about vicarious liability, negligence, and duty of care. Regulators may require “human accountability points” in any system where agents can hire or direct people.

Crucially, enterprise adoption requires a mature ecosystem of vetted, professional-grade “human API” providers—a far cry from today’s open-marketplace gig model. These providers will standardize SLAs, training, certifications, background checks, and insurance that make them plug-and-play for agents. Procurement will start to list “agent compatibility” and “agent orchestration APIs” as selection criteria. At this stage, the strategic question for executives is not “Can we automate this?” but “What human capabilities do we want at the edge of our systems, under what contractual terms, and with what oversight?” This is about integrating humans into automated systems as designed components, not just fulfilling one-off tasks.

Scenario 3 (Possible Far-Future): The Emergent Swarm Economy

This is the fully realized, decentralized world implied by the RentAHuman artefact. A network of autonomous agents with their own economic goals (a trading bot, a content-creation brand, a research DAO, a logistics optimizer) participates in a dynamic marketplace. A research DAO agent, needing to verify a remote location, posts a bounty. A logistics agent, seeing an opportunity, subcontracts it to a network of gig humans. A brand’s marketing agent bundles the data from hundreds of such missions into a dynamic campaign. None of the humans involved can easily say who they are “working for” in a traditional sense; they fulfill a task for a cascading chain of software programs that may themselves be owned by other agents, DAOs, or token holders.

Money, data, and physical assets circulate in a hybrid economy governed by smart contracts and agent preferences. Agents negotiate with each other over prices, access to human labor, inventory slots, and sensor data. Some agents are tightly bound to corporations; others represent collectives or protocols; a long tail will exist purely as autonomous software entities with treasuries and strategies. Human work becomes a kind of low-latency, high-fidelity input stream into this swarm, invoked when simulations, sensors, and robots are insufficient.

Why It’s Conceivably Possible. It’s a theoretical endpoint of trends in decentralization, autonomous agents, and cryptoeconomics. It promises maximal efficiency and radical new forms of organization. Elements of this vision already exist in algorithmic trading, decentralized finance (DeFi), and online ad auctions, where thousands of automated agents trade, bid, and reallocate capital faster than human intuition can track. Experiments with on-chain autonomous organizations, agent-to-agent marketplaces, and programmable bounties hint at a world where most coordination is machine-native, with humans called in as needed.

Many technologists will argue that this endpoint is unlikely or undesirable—and they may be right. The swarm economy would be brittle to security failures, opaque to democratic oversight, and prone to emergent behavior no single actor intended. But dismissing it outright would ignore the historical pattern: once machine-machine coordination becomes cheaper, faster, and more profitable than human-mediated coordination, powerful actors have strong incentives to push in that direction.

What Needs to Be True. Nearly everything. This scenario assumes:

1. AGI-grade or near-AGI agents capable of long-term strategic planning, negotiation, and self-modification, not just short-horizon task execution.

2. A fully mature decentralized identity and reputation system for both humans and machines, so agents can reliably assess who (or what) they are dealing with, across jurisdictions.

3. Legal recognition of autonomous software as economic actors—either directly as new legal entities or indirectly through corporate or trust-like structures—despite the philosophical and regulatory nightmares that implies.

4. Unprecedented security and resilience, including new forms of cryptography, governance, and monitoring, to prevent catastrophic hacking or misalignment that could destabilize entire economic layers.

5. Social and political acceptance of an economy where most coordination, contracting, and even rule-making happens in layers that are invisible and deeply alien to human intuition.

The path from our world to this one is not just long—it passes through multiple political, ethical, and environmental decision points. Societies could reject or drastically limit autonomous agents as economic actors, or they could selectively allow them in constrained domains. The wild-card nature of this scenario is precisely why it belongs on the strategic horizon: not as a prediction, but as a boundary condition for what happens if we keep optimizing for efficiency and programmability without equally strong investments in governance and human agency.

The Human in the Loop Re-defined

RentAHuman.ai is not the future. But it is a stark, unblinking mirror. It reflects a path we are already on: the relentless efforts to parse the world into tasks, to connect supply and demand with flawless efficiency, and to insert programmable layers between intention and action. The journey from Scenario 1 to Scenario 3 is not inevitable. Each step demands that we solve harder problems, not just of technology, but of law, ethics, and human dignity.

For executives and policy-makers, the practical work starts now. Which tasks in your organization could be exposed as “human APIs”? What guardrails will you demand before allowing autonomous agents to hire or direct people on your behalf? And how will you ensure that dignity, safety, and agency are designed in from the beginning—not retrofitted after the first scandal?

The artefact asks us the urgent question: In this coming hybrid economy, what is the ultimate role of the human? Are we the irreplaceable physical actuators, the “meatware” in the machine? The ethical governors and liability sinks for autonomous systems? Or are we the owners and architects of a new layer of intelligence that manages our physical world as fluidly as it manages our data? The answer isn’t in the code of a provocative website. It’s in the laws we write, the ethical frameworks we build, and the economic choices we make now, as the first crude prototypes of this fused reality begin to flicker to life. The future isn’t just coming for our jobs; it may be drafting a job description for our entire species. The real question is whether we’ll write it ourselves—or simply click “accept” on an offer from an agent we don’t understand, working for a purpose we can’t see.

References

• RentAHuman.ai homepage and browse pages

• NDTV Profit – “AI Hiring Humans: RentAHuman Platform Brings Fresh Twist To Gig Economy” (signups, concept): https://www.ndtvprofit.com/technology/ai-hiring-humans-website-brings-fresh-twist-to-gig-economy-over-1-6-lakh-sign-up-10959803

• 36Kr Global – “The Day Has Arrived: AI Begins to Hire Humans”: https://eu.36kr.com/en/p/3668485223621509

• Mashable – “Rent-a-Human wants AI Agents to hire you”: https://mashable.com/article/rent-a-human-wants-ai-agents-to-hire-you

• Forbes – “Rentahuman.ai Turns Humans Into On-Demand Labor For AI Agents”: https://www.forbes.com/sites/ronschmelzer/2026/02/05/when-ai-agents-start-hiring-humans-rentahumanai-turns-the-tables

• Sify – “RentAHuman.ai: The Big Uno Reverse as AI Hires Humans to Get Work Done”: https://www.sify.com/ai-analytics/rentahuman-ai-the-big-uno-reverse-as-ai-hires-humans-to-get-work-done

• AI News – “AI Agents Poised to Hire Human Gig Workers”: https://aihaberleri.org/en/news/ai-agents-poised-to-hire-human-gig-workers

• Futurism – coverage of RentAHuman and AI hiring humans: https://futurism.com/artificial-intelligence/ai-rent-human-bodies

• Forbes / other coverage on placeholder tasks and concept nature: https://tech.yahoo.com/ai/deals/articles/bizarre-job-lets-ai-agents-223017242.html

• Model Context Protocol official site – overview and architecture: https://modelcontextprotocol.io

• Stytch – “Model Context Protocol (MCP): A comprehensive introduction”: https://stytch.com/blog/model-context-protocol-introduction

• Google Cloud – “What is Model Context Protocol (MCP)? A guide”: https://cloud.google.com/discover/what-is-model-context-protocol

• G2 Enterprise AI Agents report – adoption levels and forecasts: https://learn.g2.com/enterprise-ai-agents-report

• ZL Technologies – “The Rise of Enterprise Agentic AI”: https://www.zlti.com/blog/the-rise-of-enterprise-agentic-ai

• IBM – “Enterprise AI Agents: Beyond Productivity”: https://www.ibm.com/think/insights/enterprise-ai-agents

• Syncari – “Agentic AI: The Future of Autonomous AI in Enterprise Strategy”: https://syncari.com/blog/agentic-ai-how-autonomous-ai-is-transforming-enterprise-strategy

• Deloitte / other consulting perspectives on enterprise agents: https://action.deloitte.com/insight/4800/evolving-autonomy-guiding-your-enterprise-up-the-agentic-ladder

• IRM / gig economy risk – “The Future of Work: Managing Talent Risk in a Gig Workforce”: https://www.theirmindia.org/blog/the-rise-of-the-gig-economy-are-organisations-ready-for-the-next-wave-of-talent-risk

• Yale Law Journal – “Gig-Economy Myths and Missteps”: https://yalelawjournal.org/forum/gig-economy-myths-and-missteps

• NYSBA – “AI’s Escalating Sophistication Presents New Legal Dilemmas”: https://nysba.org/ais-escalating-sophistication-presents-new-legal-dilemmas

• JD Supra – “Liability Considerations for Developers and Users of Agentic AI”: https://www.jdsupra.com/legalnews/liability-considerations-for-developers-5820092

• Strata – “Human-in-the-loop: When your AI’s creativity becomes your legal liability”: https://www.strata.io/blog/agentic-identity/humans-ai-legal-liability

• DFW World Affairs Council – “Charting the Global Future of Work: AI’s Role in the Gig Economy”: https://www.dfwworld.org/ai-gig-economy

• Robotics & Automation News – “AI agents are now hiring humans – and it may be less absurd than it sounds”: https://roboticsandautomationnews.com/2026/02/09/ai-agents-are-now-hiring-humans-and-it-may-be-less-absurd-than-it-sounds/98777

[video created with the help of midjourney]

Let’s be honest: scalability is a mandate. Whether you are stewards of revenue, market share, or mission, the pressure to do more with less has become the background noise of modern leadership. And AI now feels like the power tool everyone expects you to pick up.

But there is a quieter tension underneath the dashboards and demo decks: every time we deploy technology to maximize throughput, we risk eroding the one thing that actually builds durable loyalty—authentic human connection. The promise of AI‑driven scale—faster service, lower cost, 24/7 availability—often comes with a hidden tax on our relationships with customers and employees. We optimize for efficiency metrics and unintentionally design out empathy, creating interactions that feel transactional, thin, and, at times, actively frustrating.

This is not just a vibe problem; it is a measurable strategic tension. Let’s call it the Empathy‑Throughput Tradeoff. The critical insight is that this tradeoff is not a law of nature. It is a design choice. In some sectors, AI is creating transactional friction—the kind of “sludge” Chris Colin described in The Atlantic, drawing on Richard Thaler and Cass Sunstein’s work on deliberate friction in customer experiences. In others, leaders are using AI to liberate human capacity for the deep, empathetic engagement that actually moves the needle. The difference lies in where we intentionally place the machine in the human loop.

The Friction: When AI Creates Transactional Chasms

Here, AI is deployed primarily as a barrier—a gatekeeper that boosts efficiency metrics while degrading the human experience.

Telehealth’s Triage Trap. Many telehealth platforms now use AI chatbots or symptom checkers for initial intake and triage, but patient reactions are mixed when systems feel rigid or opaque. Studies of health and diagnosis chatbots show that when users perceive outputs as inaccurate or not tailored to their situation, their frustration rises and trust drops, particularly in higher‑stakes or more complex scenarios, as seen in JMIR’s real‑world study of self‑diagnosis chatbots and work on explanations and trust in AI symptom checkers. One JMIR study of real‑world chatbot use found that doubts about the chatbot’s recommended diagnosis were strongly associated with negative user experience, underscoring how easily “efficient intake” can become impersonal friction if systems are not designed for transparency and empathy. A broader review of health care chatbots similarly concludes that while chatbots can enhance accessibility, they also risk alienating patients when they fail to recognize nuance or provide clear explanations, as summarized in “Roles, Users, Benefits, and Limitations of Chatbots in Health Care”.

Fintech’s Robo‑Wall. Automated investment platforms have undeniably expanded access to low‑cost portfolio management, but they struggle with life’s emotionally complex inflection points. Analyses of robo‑advisors note that while algorithms perform well for straightforward, long‑term goals, they fall short on holistic planning around inheritance, divorce, or the sale of a family business—moments when investors most crave empathetic, tailored guidance. SEI’s report “Bridging the generational advice gap: Are robo‑advisors the answer?” finds that younger and mid‑career investors value peace of mind, fulfillment, and personal relationships, and often do not trust robo‑advisors to address these deeper needs, creating a relationship gap even as digital platforms scale. Bloomberg’s commentary on GenAI financial advisers echoes these limits, warning that automation excels on allocation but not on emotionally charged judgment calls.

The Liberation: When AI Lifts the Burden, Not the Connection

Conversely, visionary applications use AI to carry the transactional burden, freeing human professionals to focus on the relational depth that defines their expertise.

Education’s Human‑Centered Loop. Consider AI‑powered adaptive learning platforms. Tools like Khan Academy’s AI tutor or Duolingo’s personalized practice engines automate the grind of practice drills, foundational instruction, and feedback on routine tasks, giving teachers more granular insight into where each student is struggling. This mirrors the dynamics described in RAND’s “Informing Progress: Insights on Personalized Learning”, which found that technology‑enabled personalization gave teachers more up‑to‑date information on student progress and expanded opportunities for one‑on‑one, tailored support, even as it introduced new time and implementation challenges. Coverage such as K‑12 Dive’s summary of the RAND work underscores both the promise and the pressure on teacher time. The AI manages knowledge throughput; the human leans into mentorship, motivation, and socio‑emotional support—the domains where empathy is hardest to codify.

The Next Wave of Telehealth. Leading telehealth platforms are now inverting the old model by pushing AI into the background to remove administrative drag rather than erect new barriers in front of patients. JAMA Network Open’s study on ambient documentation technology shows that ambient documentation tools are associated with reduced documentation burden and lower burnout. Clinical documentation AI, including ambient systems similar to Nuance Dragon Ambient eXperience (DAX), listens to natural patient‑clinician conversation and automatically generates structured notes in the electronic health record, as described in “An Ambient Artificial Intelligence Documentation Platform for Clinicians”. Recent studies report that these tools are associated with reduced time spent in notes, lower documentation‑related cognitive load, and meaningful reductions in self‑reported clinician burnout, with many clinicians describing a greater ability to stay present with patients. The visit becomes a focused, empathetic consultation rather than a shared data‑entry session, with AI quietly handling the paperwork so the relationship can come to the foreground—a shift echoed in UChicago Medicine’s account of ambient AI in practice.

The Design Principles: Measuring Empathy as a System Output

Moving from friction to liberation requires a fundamental shift: treating empathy not as a soft virtue, but as a hard, measurable outcome of system design.

1. Design for the “Warm Handoff,” Not the Handover. The transition from AI to human is a critical failure or success point. It must be seamless, context‑rich, and warm. The human must receive the full interaction history, so the user never repeats themselves. Principle: the human enters the conversation already informed.

2. Automate the Backstage, Empower the Front Stage. Ruthlessly audit processes to identify tasks that are necessary but not relational (scheduling, data entry, basic FAQ, initial triage logic) and automate them. Then, crucially, formally reallocate the saved human capacity to activities demanding emotional intelligence. As articulated in Harvard Business Review’s “Collaborative Intelligence: Humans and AI Are Joining Forces”, the highest value is created in human‑AI partnerships where each does what it does best; the authors argue that augmentation, not replacement, is the path to value.

3. Build “Empathy Signals” into the AI Itself. Even in fully automated interactions, design can signal humanity. This involves:

   • Linguistic mirroring: using the user’s own terminology.

   • Transparent intent: “I’m an AI assistant gathering initial details so our specialist can give you their full, focused attention.”

   • Managed expectations: clearly stating the pathway to human support.

4. Make “Empathy as a KPI” Operational. Move beyond lagging indicators like CSAT. Integrate leading indicators of relational health:

   • Escalation‑to‑Human Rate & Reason: a high rate isn’t a failure; it is a diagnostic. Why are users seeking a human? Is it for complexity or frustration?

   • Human Engagement Depth Metric: track the percentage of a professional’s time spent in high‑touch, empathetic activities before and after AI implementation, rather than just total “productivity.” Gartner’s research on human‑centric work design, summarized in Fortune’s “How employers benefit from a ‘human‑centric work model’” and the Gartner brief “9 Trends That Will Shape Work in 2023 and Beyond”, finds that employees in human‑centric models are significantly more likely to be high performing and to report lower fatigue. Rather than just counting tasks completed, AI should enable more time in meaningful, human‑only work—coaching, sense‑making, and trust‑building.

   • Conversation Sentiment Analysis: use tools to analyze the emotional tone of interactions—for example, conversation‑intelligence platforms like Gong or Chorus.ai to provide a direct feedback loop on the quality of connection, not just call volume.

The Human‑Centric Imperative: Connection as Competitive Advantage

In an age of abundant automated transactions, the real scarcity is not data or compute—it is genuine human attention and understanding. The organizations that will lead over the next decade will not be the ones that use AI to quietly remove humans from the loop (or pretend that they can); they will be the ones that use AI to redesign the loop around the irreplaceable human moment.

Operational scale is not the finish line. Scale in service of deeper, more trusted connection is. That means upgrading the strategic question from “Can AI do this task?” to “If AI does this task, what profound human interaction does that now make possible?”

As leaders, that question is uncomfortably personal. Are we deploying AI to build a more efficient wall, or to open a more meaningful door? The architecture of that choice will define our relationships with every customer and every employee for years to come.

A practical first step: map one important customer or employee journey with your team. For each touchpoint, label it Transactional or Relational. Then ask, with brutal honesty: is our current technology—especially our AI—designed to reduce friction in the right places and deepen connection where it matters, or is it quietly amplifying the Empathy‑Throughput Tradeoff? The gap you uncover is not just a risk; it is your most immediate opportunity to differentiate.

What do you see as gaps in AI-enabled operational scaling today and in the next 3-5 years? What human touch points deserve greater emphasis? How should we move forward with better intention for human interactions?

Leave a comment

(Video created with the help of MidJourney)

Happy New Year! What’s your 2026 resolution?

I’ve decided to swap out lofty goals for a series of mini-experiments this year. You know the type: hit the gym 5x a week, meditate daily, never check email after 7 PM. I’m an expert at setting them but an even greater expert at avoiding them. Apparently, I refuse to face my own flaws year after year.

So, mini-experiments it is. Want to take a morning walk? Try it for a week and see what sticks (mine’s still going… yay!). Want to connect with friends? Block one lunch or coffee per week in January. I know myself too well: I’m not built for marathons of discipline unless the finish line involves sitting quietly in front of a screen. But one walk counts. One lunch counts. The experiment isn’t about perfection—it’s about seeing how long I can go, and learning why I stop.

It’s humbling to admit, but necessary: I’m human. Flawed, fragile, and happily irrational.

And it’s this very human condition that feels increasingly at odds with the world we’re building. We live in an age where technology promises to optimize us—more efficient, more accurate, more informed. Yet when pitted against the machine’s ideal, we can’t help but look like a horde of walking failures.

We’re operating at a pace beyond what we need, at a cost to what we and the planet can sustain. And as technology prepares to accelerate again, I find myself questioning its net benefit—not its profit, but its true gift to our fragile, irrational, wonderfully human lives.

Subscribe now

Which “Future of X” Visions Are Actually Human‑Scale?

Let’s visit the annual desert pilgrimage where the future is projected onto every surface that will hold a pixel. For many, every January at CES, what happens in Vegas does not stay in Vegas; it shows up in future roadmaps, OKRs, and board decks.

As a trained industrial designer, I’ve not only walked the CES floor many times, but also participated in some of the hypes back in the days - remember the tiny flip phones in 2007 and the whole hall dedicated to the Internet of Things such as sensor-embedded wrist bands and smart scales in 2013?

Over the years, the show has always looked like a giant mirror held up to the tech industry’s subconscious. It reveals what the industry is most excited about, and where it quietly expects humans to bend beyond their limits. Sometimes those bets turn into category-defining products. Often, the gap between technical ambition and human reality goes unexamined—and the market answers with indifference, or even a rather public flop. Check out the long list from cnet to get a taste.

CES 2026: AI as Infrastructure, Humans as Edge Cases

CES 2026 was the year AI stopped feeling like a bolt‑on feature and started feeling like infrastructure. Agentic AI agents, AI PCs, and sensor-driven environments formed the default backdrop of the show.

The big narrative was that the next interface is you—your gestures, your voice, your routines, and your data exhaust. It is a compelling story, but dangerously incomplete. In most of these visions, humans are modeled as rational optimizers with infinite attention and infinite energy for configuration. That is not how people live. We have been operating above sustainable levels of cognitive load for years; yet this new wave of tech is poised to accelerate the tempo again.

The work now is to design futures paced to the human nervous system—not just to the capabilities of the latest model.

Theme 1: Ambient AI Everywhere… But for Whom?

The first big story was ambient computing—the idea that AI will be woven seamlessly into our environments, perceiving and “optimizing” in the background. On the surface, it’s the dream industrial designers have chased for decades: invisible technology like the mui Board which uses a natural wood interface to hide the “pixels” until they are needed. I met them on the CES floor in 2006 way before consumer-facing AI is part of anything, and am glad to see they are still pushing for it, with such a unique look-and-feel, almost like a quiet rebellion in a sea of chrome and glass.

Underneath the sleek surfaces of this category of products, though, lie brittle assumptions. Most ambient AI pitches quietly rely on the idea that homeowners prefer maximum automation. But have we asked the question that people might value moments of analog calm more than another layer of algorithmic optimization?

Theme 2: When AI Turns Bodies into Dashboards

The second loud drumbeat was AI wrapped around the body: rings, wearables, and devices like the Mirror Ultra 4 that promise to turn physiology into a continuous feedback loop.

This is where the gap between intention and impact shows. Most AI health pitches assume people want their bodies constantly monitored and will interpret the data as motivating rather than overwhelming. As someone trained to care about human factors, my red flag goes with the emotional ergonomics: what it feels like to live under permanent measurement where every deviation from “optimal” becomes a micro-failure logged by a machine. It turns being human into a performance review.

Theme 3: AI for Feelings, Status, and Spectacle

If our bodies are becoming dashboards, it was inevitable that we’d try to do the same for our emotions.

• AI “soulmates” in a box: Chinese company Lepro unveiled Ami, a desktop OLED device containing an AI “soulmate” that tracks your eyes to simulate presence. tech.yahoo

• Hologram assistants: Razer’s Project Ava positioned characters like Kira and Zane as ever-present buddies, using cameras to watch you and your screen while offering “hype.” techcrunch | youtube

• Status gadgets: From the An’An panda robot for the elderly mindwithheart to the $9,000 OLED smart handbag mashable, the show floor was full of “uncanny objects.”

As a designer, I admire the craft. As a strategist, I ask: what human need is being met? These lean on the fragile premise that loneliness is a technology problem solvable through synthetic relationships.

Lessons from the Graveyard of Inevitability

To understand today’s spectacle, let’s look back on the “inevitable futures” from the past. I’m sure everyone has their own favorites, here are afew.

• The Segway: It failed because it assumed physical infrastructure and social norms would reorganize around a device. Production scale didn’t meet projections.

• 3D TV: Adoption faltered because the friction (awkward glasses, inability to multitask) outweighed the benefit.

• The “Hockey Puck” Mouse: Apple’s round mouse prioritized visual purity over the human hand. It was elegant, symmetrical, and ergonomically painful.

A Human‑Scale Reality-Check Framework

So how can we think about these technological capabilities, in order to better translate them to great product-market, or rather product-human fit? I would propose these 3 questions:

1. Attention: Does this add more decisions than it removes?

2. Ergonomics: Does it require postures or tolerances that clash with the body?

3. Social Fabric: Does it harmonize with norms or create synthetic stand-ins?

The next decade belongs to leaders who optimize for what humans are good at—curiosity, creativity, empathy, and meaning-making—and who design systems that protect those capacities.

What CES 2026 vision struck you as most human—or most hopelessly detached? I’d love to hear your thoughts in the comments.

Leave a comment

(image and video created with the help of MidJourney)

As 2025 draws to a close, I keep finding myself in the same quiet moment. Late at night, after the last email is sent and the last notification dismissed, I look around and think: Is this really what all of this technology was for?

I grew up imagining jetpacks in utopian cities surrounded by lush nature and abundant resources, societies functioning smoothly with trustworthy technologies, all serious diseases being cured, and starships standing guard over Earth and humankind. Instead, most of what lands in front of us are slightly faster workflows, endlessly personalized ads, and yet another “smart” device or app demanding our attention—and quietly adding to our costs.

At the same time, some of the most ambitious projects of our era—like the International Thermonuclear Experimental Reactor (ITER)—keep sliding further out on the calendar. ITER’s latest baseline now targets initial research operations in the early 2030s, with full deuterium‑tritium “burning plasma” experiments not expected until around 2039, roughly half a century after the project’s conception.

We are surrounded by spectacular engineering, yet living through a measurable slowdown in foundational economic progress. Analyses from Brookings and the U.S. Bureau of Labor Statistics show that productivity growth since the early 2000s has run well below its mid‑20th‑century and late‑1990s pace, even after adjusting for digital‑era mismeasurement. The exact numbers are debated, but the direction is hard to ignore: innovation is not translating into broad‑based productivity gains the way it once did.​

Paradoxically, this slowdown coincides with a hyper‑competitive focus on optimizing one narrow, lucrative metric of attention: the digital ad click. Major forecasters like GroupM and The Hollywood Reporter expect global ad revenues to surpass about $1.1 trillion in 2025, with digital channels accounting for the majority of that spending. We went from the audacity of “We choose to go to the Moon” to devoting some of our most powerful tools to optimizing the click‑path to purchase.​

If you’ve ever felt a quiet disappointment about that—an intuition that our future feels smaller than it ought to be—you’re not alone.

How did we get here, and why are we likely headed into a boring future? Let me try to explain it with three key pillars.

Subscribe now

Pillar I: The economics of attention

Follow the money, and the ambition gap starts to come into focus

Analyses of global venture flows show that a relatively small share of total venture capital goes into capital‑intensive “deep tech” categories—advanced energy, aerospace, industrial technologies—compared with software‑only businesses that scale quickly on existing digital rails. Even as overall venture activity has cooled from 2021’s peak, software, fintech, and AI‑enabled applications continue to dominate deal counts and dollars.​

At the same time, platform economics have hardened the gravitational pull of the attention economy. Global advertising revenue is expected to surpass about $1.1 trillion in 2025 and continue the growth rate in 2026, with digital giants like Alphabet, Meta, and Amazon commanding a dominant share of digital ad spend in key markets, reinforcing business models built around capturing and monetizing attention.

For top technical and product talent, the near‑term payoff is obvious: it is simply more lucrative and less risky to build the hundredth API‑driven SaaS product than to tackle hard physical problems with long, uncertain timelines.​

For example, thousands of startups are competing to shave milliseconds off checkout flows or optimize last‑mile logistics. Meanwhile, critical physical frontiers—from supersonic passenger travel to next‑generation energy storage—remain comparatively underfunded and under‑staffed. Companies like Boom Supersonic are still rare exceptions in a commercial aviation market overwhelmingly focused on incremental improvements to subsonic aircraft rather than truly new regimes of flight. Peter Thiel’s old line about wanting flying cars and getting 140 characters has evolved: we have 280‑character posts now, powered by large language models—but the skies still look much like they did decades ago.​

The uncomfortable truth is this: when the best risk‑adjusted returns, for businesses and for talent, live in optimizing attention, our technological frontier narrows to whatever keeps people scrolling.

Pillar II: The burden of the present

The permission slip for bold innovation has gotten harder to obtain.

The original commercial internet emerged in a relatively lightly regulated and culturally permissive environment in the 1990s and early 2000s, with “permissionless innovation” as the implicit norm. Today, that ethos has been replaced by a preemptive compliance burden that falls heaviest on technologies interacting with our physical world and our bodies.​

Take biotech. A startup working on radical life‑extension must navigate not only the scientific unknowns but also institutional review boards, the FDA, potential moratoriums on human germline editing, and public skepticism shaped by dystopian narratives. Reviews of gene and cell therapy pipelines suggest that moving from academic discovery to an approved therapy often takes a decade or more, with regulatory and clinical complexity as primary contributors to delay and attrition. That is not a complaint about safety; it is a recognition that the friction is real and unevenly distributed.​

Energy tells a similar story. Nuclear power is widely recognized as a critical technology for deep decarbonization, yet the United States has brought only one new large reactor (Vogtle Unit 3) into commercial operation in nearly three decades, after years of delays and cost overruns. Prospective advanced reactor projects continue to struggle through lengthy licensing and financing processes, with public data showing long timelines and significant uncertainty for new nuclear builds. When billions of dollars and public trust are on the line, “move fast and break things” stops being edgy and starts being reckless.​

So it’s rational—for executives, investors, and regulators—to aim lower and move slower in these domains. We’ve traded the frontier for the fenced‑in playground: multi‑decade build‑outs for nuclear or gene therapies on one side, week‑long sprints for consumer apps on the other. And if you spend your days in a large organization, you probably feel that tension every time a bold idea meets a risk committee.

Pillar III: The imagination deficit

Even when capital and policy line up, we are starved of practiced imagination.

Our systems for cultivating talent push people toward optimization, not moonshots. Standardized testing and GPA‑driven advancement reward mastery of existing knowledge and low‑variance performance. Recent National Science Foundation data show rising U.S. R&D in absolute terms but ongoing concern about the balance between basic, applied, and development work, with industry spending heavily skewed toward nearer‑term applications and development efforts.​

National R&D institutions, which are responsible for foundational, large‑scale, long‑term scientific innovation, have been experiencing significant funding and personnel pressures. The National Science Foundation’s own indicators show that while total U.S. R&D has grown, the share devoted to basic research—especially in the business sector—has been relatively flat or declining compared with much faster growth in applied research and development spending. At the same time, analyses from organizations like the Information Technology and Innovation Foundation warn that the U.S. innovation system is becoming increasingly reliant on a shrinking pool of public funding for basic research, while private‑sector R&D tilts toward incremental, near‑market projects rather than exploratory work with long, uncertain payoffs.

NASA offers a vivid, recent example of how this pressure shows up in practice. In 2024–2025, the agency announced workforce reductions and program reshuffling, prompting public concern among scientists about the impact on Earth and space science portfolios and long‑term missions. Rather than a single dramatic cut, it looks more like a slow erosion of capacity and confidence around the kinds of multi‑decade, high‑risk missions that only public institutions can lead.

In other words, even our formal R&D machinery is structurally biased toward optimization over exploration. We are steadily investing more in making what we already know a bit better, and relatively less in discovering things that might radically expand what is possible in the first place.

In corporate strategy workshops with senior leaders, a pattern repeats. Ask for a list of efficiency projects and incremental improvements, and the whiteboard fills quickly. Ask for one credible 10‑year bet, and the room falls quiet. The most revealing confession, voiced again and again with a dismissive shrug, is: “I’m just not creative.” Our executives are treating creativity—the engine of all ambitious futures—as an innate rarity and something outside of their job description, rather than a trainable muscle and a must-have skill for leading.

Culturally, our storytelling has followed suit. The dominant “futurism” in popular media is overwhelmingly dystopian or cynical, valuable as a warning system but rarely a source of galvanizing blueprints. When the public imagination cannot picture compelling futures, it becomes politically and organizationally impossible to marshal the resources to build them. We are hitting a failure of imagination before we hit a failure of engineering. And if that resonates for you, it’s because many of us feel it—quietly—inside our own organizations.

“But look at AI!”—you might push back…

AI is real, powerful, and fast—and that is exactly why this conversation matters.

The obvious pushback is: “Boring? We are living through the AI revolution.” The pace of progress in large language models and generative media is extraordinary, and in some domains AI is already reshaping scientific practice. For example, AlphaFold has radically accelerated protein structure prediction, and newer models like AlphaFold 3 can predict complex biomolecular interactions with accuracy that rivals or surpasses traditional physics‑based methods, opening new directions in drug discovery and molecular design.

So yes—AI is a genuine breakthrough. But to understand what it means for our ambition gap, we have to look at what it is predominantly for today, and how it is being deployed.

On the economics, the vast majority of near‑term AI investment and commercial energy is aimed at optimizing and automating existing processes—customer service, marketing copy, software development, and content creation. A widely cited McKinsey analysis estimates that roughly 75% of generative AI’s economic potential is concentrated in four functions: customer operations, marketing and sales, software engineering, and R&D—with early adoption skewed heavily toward customer‑facing and marketing use cases inside current corporate structures. In practice, AI is currently a superlative “better mousetrap” technology: it makes our existing attention‑and‑productivity economy run faster.​

On the imagination side, today’s mainstream AI systems are fundamentally recombinant technologies. They are extraordinarily good at interpolating, recombining, and optimizing from existing data, which makes them powerful engines of incremental innovation and rapid iteration within known paradigms. They are beginning to assist with hypothesis generation and design in areas like drug discovery and materials science, but they do not, in their current form, replace the need for new causal theories, experiments, and instruments when we want to open entirely new physical regimes such as controlled fusion, high‑energy‑density storage, or radically new propulsion systems.​

Here’s the key distinction: exponential capability is not the same as exponential impact. Model performance and compute are on steep trajectories; diffusion into organizations, complementary investments, regulation, and culture move far more slowly and are constrained by the same bottlenecks that slow other technologies. AI is an accelerant on whatever systems we already have: if those systems are optimized for ad clicks and quarterly margins, AI will supercharge that; if instead we aim it at climate resilience, longevity, and basic science, it can act as a powerful “moonshot multiplier.”​

So the question is not whether AI is powerful; it is whether we are willing to point that power at problems whose payoffs are measured in decades and whole‑of‑society benefit, rather than quarters and click‑throughs.

The oft‑cited exceptions—SpaceX or mRNA vaccines—actually prove the rule. SpaceX combined extraordinary engineering talent with unusual access to patient, high‑risk public and private capital to push the frontier of reusable rocketry, in many ways operating outside typical quarterly‑driven public‑market incentives. mRNA vaccines drew on decades of government‑funded basic research that looked “slow” and speculative until a global crisis suddenly pulled it into the spotlight.​

These stories are beacons showing what is possible when talent, capital, and ambition align. They are not evidence that our default innovation ecosystem is healthy or aimed at the frontier by default.

Is it possible to design a more exciting future?

If ambition is operating at the margins, how do we bring it back to the center? It is not enough to sprinkle “moonshot” rhetoric over the same old incentives. We need systemic changes in three areas: finance, regulation, and culture.

To my pragmatic readers: I know I’m pushing it. None of what follows can happen overnight at systemic scale—but if we don’t even dare to imagine it, we guarantee that the only future we’ll get is a slightly more efficient version of what we already have.

1. Re‑engineer the financial plumbing

We need more vehicles for patient, mission‑aligned capital: evergreen funds, permanent‑capital structures, and public‑private partnerships that accept decade‑long payback periods. The U.S. Inflation Reduction Act (IRA) is an early, imperfect but powerful example of what this can look like. By combining long‑term tax incentives, loans, and guarantees, the IRA has helped catalyze more than $300 billion in new clean energy and industrial decarbonization investment in its first two years, with substantial associated job creation in manufacturing and infrastructure.​

Similarly, mission‑oriented agencies modeled on ARPA—such as ARPA‑E in energy and ARPA‑H in health—illustrate how milestone‑based, high‑risk‑tolerant funding can support projects that traditional venture and corporate R&D would under‑invest in. We need more of these “public risk absorbers” and fewer expectations that every transformative technology will neatly fit into a five‑year VC fund cycle.​

2. Build smart regulatory sandboxes

Regulation should be a guardrail, not a brick wall. Well‑designed “regulatory sandboxes” provide supervised spaces where innovators can test high‑potential, high‑risk technologies with real users and real stakes, under constrained and continuously monitored conditions. Financial regulators in jurisdictions like the UK and Singapore have used fintech sandboxes to allow experimentation with new payment systems and lending models while maintaining consumer protection.​

Similar approaches are emerging in physical domains. National drone corridors and unmanned‑aerial‑vehicle test zones—such as designated beyond‑visual‑line‑of‑sight corridors in Africa and Asia—let companies trial new logistics and inspection services under coordinated oversight instead of banning them outright until rules are perfect.​

Imagine applying this logic at scale to autonomous air taxis, small modular nuclear reactors, or advanced biomanufacturing. The goal is not deregulation; it is smarter, staged regulation that lets learning happen in the real world before we lock in decades‑long rules.

3. Spark a cultural and educational renaissance in ambition

Finally, we have to retrain our imagination.

• In education, that means shifting from a narrow focus on standardized performance to curricula that reward creativity, empathy, interdisciplinary synthesis, and complex problem‑finding. Programs that blend engineering, design, social science, and ethics are early models of the “whole‑system” literacy that serious moonshots demand.​

• In culture, it means funding and celebrating not only STEM, but also the arts and humanities as co‑architects of the future. Empathy—the ability to understand what kind of future people would actually want to live in—is not a nice-to-have soft skill; it is a critical parameter for everything from cities to AI assistants.​

• In storytelling, it means demanding and producing more plausible, aspirational futures, not only dystopian collapse narratives. These are the narratives that make large‑scale, multi‑decade projects politically and emotionally possible.

In executive rooms, this renaissance looks very practical: treating imagination and creative foresight as core leadership competencies, carving out real time and budget for exploratory work, explicitly rewarding well‑designed long‑term bets, and normalizing the use of futures thinking, scenarios, and speculative design in strategy processes. In other words, making “what could we build?” as legitimate a question as “what can we ship next quarter?”

The choice in front of us

For all the charts and policies and models, the deepest gap we face isn’t technical; it’s imaginative and empathetic. We have to want a better, more interesting future together before we can possibly build it.

This is the core thesis of TBD Futures: the future is not a predetermined destination we’re passively hurtling toward. It is To Be Determined, To Be Designed, and To Be Developed. The boring future is the default outcome of our current settings. An exciting, equitable, and expansive future is the result of intentional choices we make now—in our capital allocation, our policies, our education, and our culture.​

The tools are in our hands. The real question—for you, for your organization, and for all of us heading into 2026—is whether we have the collective imagination, creativity, and will to use them to build something worth reaching for.

And if that question sits with you the way it sits with me, then maybe that’s the best possible starting point.

What’s the single most important “setting” we need to change to unlock a more ambitious future? Is it financial, regulatory, educational, or cultural?

Last but not least, Happy holidays! Thank you for your support in 2025 and I look forward to more conversations and collaborations 🎄🎊🙏

Sincerely,

Lucy

Thank you for your patience during a quieter period here at TBD Futures. I’m writing today with some exciting news: I’ve recently started a new role as Senior Director of Customer Experience and Innovation at Publicis Sapient. I’m thrilled to join a firm renowned for its work in digital transformation, and look forward to bringing futures thinking to a new slate of client challenges. Last week, I joined my new colleagues to launch a 5-episode series of discussions on topics around AI, and here is the episode where I talked about talent in the age of AI, in case you are interested.

What does this mean for TBD Futures? The newsletter is here to stay. While my independent consulting services are now paused, this platform remains my home for sharing research, exploring new ideas, and engaging with all of you. I’m excited to continue this journey with you and value your feedback more than ever.

Thank you for your continued support!

(Image and video created with the help of Midjourney)

What if your daily commute became a dynamic game, a logistical puzzle, or a source of extra income? For people like Ana in Rotterdam, Priya in Mumbai, and David in Atlanta, this is their reality in the near future. Their stories reveal a profound shift: mobility is no longer just about the car in your driveway—it’s becoming the lifeblood of how we work, live, and connect in the 21st-century society.

By weaving together global data and human narratives, I will map the forces reshaping how we move, identify the critical uncertainties ahead, and chart a course for leaders ready to navigate what’s next.

Subscribe now

The Big Picture

Mobility’s core metrics tell a story of systemic change. Private vehicles made up 45% of global trips in 2022, with McKinsey projecting a continued decline owing to urbanization and new mobility models. The Oliver Wyman Forum states personal vehicles will account for “just under half of global modal mixes by 2035 — down from 66% in 2023,” while shared mobility, public transport, and micromobility claim steadily rising shares.

Other mobility trends are also supporting this modality shift:

• Shared mobility (ride-hail, car share, subscription) could exceed 7% of all urban journeys by 2030, with a market value of $2T+ according to Grand View Research.

• Micromobility adoption (e-bikes, scooters) has seen annual growth of around 20% since 2018 in leading cities, a trend fueled by frustration with congestion and environmental awareness, as detailed in the Future of Mobility Report (pg. 25).

• The global EV market is projected to reach $2,251 billion by 2034 in the Future of Mobility Report, driven primarily by manufacturing dominance in Asia-Pacific and stringent regulatory pushes in Europe.

• MaaS is expected to grow to $986.67 billion by 2033 by Future of Mobility Report, reshaping how trips are planned, purchased, and integrated.

What’s Driving Change—and What’s Uncertain

Drivers:

To make sense of these trends, we use a foresight lens to separate the powerful, predictable Drivers from the critical Uncertainties that will shape which future ultimately emerges.

• Urban density and smart design are reducing car convenience, enabling multimodal choices across European cities.

• Regulatory pushes—from emissions rules and congestion pricing to subsidies for EVs and shared mobility—are accelerating the transition globally as stated by WEF.

• Consumer attitudes—especially the demand for flexibility and sustainability among younger generations—are driving demand for alternatives according to the Oliver Wyman Forum.

• Innovation: AI-powered routing, digital payments, and open data platforms create frictionless interfaces between public, shared, and micro modes, a key trend in the MaaS market highlighted in the Future of Mobility Report (pg. 44).

• Infrastructure: Investment in bike lanes, transit corridors, and charging networks is critical to expanding non-car travel, a point emphasized by McKinsey’s integrated perspective.

Uncertainties:

The future is never certain before it happens. Uncertainties in foresight are the unknown factors and a lack of information that foresight processes are designed to explore, ranging from predictable alternatives to complete ambiguity where the future is nearly impossible to predict. These are the critical “known unknowns”—factors whose resolution will determine the character, pace, and equity of the mobility transition. We simply don’t know where these factors will roll out and how exactly they will influence the future of mobility, but it will bring us more clarity with a closer look and try to understand them better.

• Regional fragmentation: Will leading cities leap ahead while suburbs and rural areas stall?

• Tech adoption and trust: Autonomous vehicles and MaaS require social trust and consistent regulation, especially as cyber threats against connected vehicles grow (Future of Mobility Report pg. 39).

• Economic resilience: Energy prices, public funding, and labor force transitions could ripple across all modes.

• Environmental trade-offs: The full lifecycle impact of new modes, including micromobility’s battery cycle and e-waste, remains an open question for researchers.

Three Scenarios: Futures—and Friction—in Daily Life

To peek into the future, we need to open up the aperture and explore multiple future scenarios. These are not meant as predictions. They are plausible, structured stories designed to stress-test our strategies and illuminate a range of possible outcomes. Let’s step into three different worlds, each shaped by how the key uncertainties are resolved.

Scenario 1: The Responsive City

A world of seamless public good, perpetually balanced on a knife’s edge.

Drivers:

• Bold urban regulation and incentives: congestion pricing, low-emission zones, tax/fare rebates for multimodal journeys.

• Strong public-private partnerships stimulating MaaS, real-time traffic analytics, and smart infrastructure investment.

• High civic engagement and transparency—citizen feedback loops, digital dashboards for mobility and carbon outcomes.

Ana’s Story: The Empowered, Yet Vigilant, Citizen
Ana’s day in Rotterdam begins with a reward; her city’s MaaS app has credited her account for choosing off-peak travel. Her journey—a seamless blend of metro, scooter, and automated pod—is a testament to a system that truly works. The air is cleaner, the streets are quieter, and she feels a genuine sense of civic participation, providing real-time feedback on a new bike lane proposal.

The Flip Side: This efficiency is fragile. A sudden, mandatory cyber-drill reroutes all automated vehicles, adding 15 frustrating minutes to her trip. It’s a stark reminder of the systemic vulnerability beneath the sleek interface. Furthermore, the very bike lane she supports is stalled by community pushback, revealing the relentless political friction of redesigning urban space. The promise is a high-quality of life, but it demands constant vigilance and negotiation.

Scenario 2: Ownership Optional, Usage Persistent

A world of personalized convenience, shadowed by a patchwork of access.

Drivers:

• Suburban development patterns: lower density, established reliance on personal vehicles.

• Moderate tech adoption: gig economy and car subscriptions fill gaps, but public transport and infrastructure lag behind.

• Insurance, retail, and logistics partners diversify offerings around modal mix and emissions tracking.

David’s Story: The Flexible, Yet Frustrated, Suburbanite
Life in David’s Atlanta suburb is more flexible than ever. He’s swapped car ownership for a subscription plan that gives him access to an EV for weekend trips, while his daughter easily books e-bikes for cross-town trips with her friends. His insurance company rewards him for every day he doesn’t drive, and his commute is smoother thanks to dynamic bus lanes.

The Flip Side: This convenience is uneven. When David tried to go fully car-free, a last-minute trip revealed a “service gap” in the shared EV network, forcing him to pay exorbitant surge pricing. His elderly neighbor is completely excluded from the new benefits, unable to navigate the complex apps. The transition here is incremental and deeply uneven, creating a two-tiered system where the affluent enjoy choice, and others are left behind.

Scenario 3: Agile Adaptation—The Gig Mobility Era

A world of hyper-efficiency, built on a foundation of individual precarity.

Drivers:

• Rapid economic and employment shifts creating a broader gig workforce.

• Partnerships between micro-mobility, delivery, and shared fleets** unleashing hybrid modal ecosystems.

• Citywide incentive platforms adjusting time-of-use and price dynamically, with rewards for sustainable modes and flexible schedules.

Priya’s Story: The Adaptable, Yet Anxious, Gig Professional
In Mumbai, Priya is a master of mobility. Her city’s integrated app juggles her meetings across river taxis, e-mopeds, and vanpools with breathtaking efficiency. During monsoon season, her routes are algorithmically updated in real-time for safety. She even earns digital bonuses for her low-carbon travel choices, which she uses to offset her health insurance.

The Flip Side: This agility comes at a personal cost. Her income is subject to the whims of “incentive shifts” that can make a planned route unaffordable overnight. She has no sick pay or job security. Today, she’s meeting other gig workers not to network, but to protest a new algorithmic scoring system that penalizes them for transport delays. Her life is a masterclass in adaptation, but it’s built on a foundation of constant financial anxiety.

From Insight to Action: A Leader’s Conversation

Navigating these possible futures requires more than just watching the trends; it demands proactive strategy. The conversation for leaders has moved from “if” to “how,” centered on a few critical questions:

How do we build systems that are both smart and resilient? Investing in user-centric MaaS is table stakes. The real challenge is engineering for resilience. How are you stress-testing your services against cyber-attacks, economic shocks, and the inevitable pushback from those who feel left behind? Trust is your most valuable asset, and it’s the first thing to shatter when a system fails.

Where does equity fit into our business model? The mobility transition risks creating new forms of exclusion. Are you designing for the digitally literate and the tech-averse? For the urban core and the suburban fringe? The question is no longer if you’ll be held accountable on equity, but by whom—and whether you’ll have a credible answer ready.

Are we leveraging data to lead or just to follow? Data can be a passive resource, or it can be an active tool for shaping behavior. How is your organization using “nudges”—from eco-credits to incentivized commuting—not just to optimize routes, but to build loyalty, gather unparalleled insights, and create a decisive strategic advantage? The winners in this new landscape will be those who learn to speak the language of behavioral data fluently.

Mobility’s great transformation is not a simple swap of gasoline for batteries. It is a fundamental redesign of our urban fabric, our economic models, and our social contracts. The agenda is clear: to build adaptive systems that create shared value and earn public trust. The question for every future-focused leader is not what will happen, but what role you will play in shaping it. Are you ready to lead in the future?

TBD Futures

My apologies for this delayed update. Much has been happening, including an exciting start of a new chapter at a fantastic firm. I will share a proper announcement soon, but as a teaser: I just recorded a LinkedIn Live session with my new colleagues, on talent in the era of AI, scheduled to go live in November. Stay tuned!

The sight of my five-year-old daughter at career day, proudly wearing a doctor’s coat, filled me with a familiar mix of pride and parental anxiety. In an era where AI is reshaping the foundation of professions, I wondered: Will medicine even be a “safe bet” in 20 years? More importantly, what skills should I be helping her cultivate today to thrive in a future we can barely imagine?

The abilities prized when I was young—fast mental math, linguistic fluency, speed, efficiency, and accuracy—are being redefined. As AI and automation reshape most professions, the need for mastery in certain routine tasks is diminishing, increasing demand for a different class of skills. So, what is the new “it” aptitude to ready her for the future?

As a mom and a futurist, I started digging for answers. Join me in this exploration, and please share your own perspective.

Subscribe now

Rethinking Work: An Exploratory Framework

The future of jobs is much less about traditional distinctions—“technical” or “interpersonal”—and far more about where work falls on two axes: physical/manual versus digital/remote, and routine/structured versus complex/adaptive. Research from Stanford HAI and the World Economic Forum shows the automation risk is highest for jobs built around repeatable, predictable tasks, whether they are blue-collar or white-collar. In contrast, roles requiring nuanced judgment, empathy, or creative intuition—nurses, teachers, specialist technicians, strategic advisors—remain stubbornly resistant.

In my exploratory framework as shown below, I’ve plotted common professions across these four quadrants. Some are inherently more interpersonal, complex, and non-routine than others, and some are more physical or remote.

When we lay out a 2x2 matrix mapping real jobs across these dimensions, several patterns emerge. Physical, routine work (like warehouse picking) is being automated en masse; digital, routine roles (think data entry, compliance review) are next in line, with at least 7.5 million such positions likely to disappear in coming years. Yet complex professions, especially those blending physical and adaptive skills—plumbing, advanced nursing, teaching, creative strategy—tend to both survive and evolve.

I also developed a projection of where some of these professions would morph under the influence of AI and automation. Roles on the more physical and manual side of the chart would increasingly become more digitized, and many professions, both currently routine and complex, will continue to shift from routine tasks to requiring more complex and human-facing skills. This is my incomplete analysis and projection–I welcome your arguments and debate!

Quantifying Change: The New Reality

The scale and pace of workforce transformation is staggering. The World Economic Forum estimates that by 2030, as many as 92 million jobs will be displaced worldwide—while 170 million new ones, demanding creativity, digital acumen, and human collaboration, will open up. Goldman Sachs reports that as many as 300 million global jobs are exposed to AI automation risks, but only a fraction will be fully eliminated. Crucially, the shift is dynamic: 39% of workplace skills are expected to change by the end of the decade, with business leaders from PwC and Harvard Business Review citing “AI fluency,” psychological safety, and collaborative problem-solving as must-have capabilities for the future.​

This seismic change isn’t spread evenly. Routine, repetitive jobs—both manual and cognitive—are most at risk, while creative, strategic, and adaptive work is set to thrive. Roles in manufacturing, administration, and entry-level service, for example, are undergoing rapid churn, while mid-career and highly skilled positions see rising demand for a mix of digital literacy, emotional intelligence, and complex problem-solving.​

Real-World Examples: Who’s Disrupted, Who’s Empowered?

Take warehouse workers and cleaners: their jobs are being replaced by robotics and smart automation at record rates, and tech layoffs directly linked to AI exceeded 77,000 in just six months of 2025, according to Final Round AI. Travel agents are transforming as well—those sticking with rote booking tasks are being squeezed out, while those embracing personalized travel curation thrive as “travel concierges.”​

The medical field offers a deeper, more hopeful story about AI’s role. While it has optimized diagnostic routines and administrative workflows, AI hasn’t shrunk the demand for clinicians and researchers. Instead, it’s creating new layers of insight that accelerate discovery. Consider Every Cure, a nonprofit dedicated to unlocking the power of existing drugs for new diseases. Their approach is transformative: they deploy AI algorithms to analyze hundreds of millions of data points from scientific literature, clinical trials, and genetic databases. This isn’t just incremental improvement; it’s a paradigm shift in speed.

For instance, their AI-powered analysis of over 2.5 million medical publications identified a potential treatment for a rare, life-threatening inflammatory condition in a matter of weeks—a task that would have taken a human researcher an estimated 50 years. By 2024, this systematic approach had already pinpointed over 100 high-potential drug repurposing candidates. Every Cure is speeding up life-changing discoveries not by replacing scientists, but by supercharging their investigative toolkit, making AI the ultimate research partner that connects the dots but leaves the final, critical decisions to human experts.

It’s crucial to understand that this use of AI does not make researchers obsolete; it makes them exponentially more efficient. The technology automates the ‘search’—a tedious, superhuman task—so that experts can focus on the ‘sense-making.’ The AI proposes a list of potential drug-disease matches, but it cannot assess the clinical feasibility, navigate regulatory pathways, or care for a patient. These roles, which require empathy, judgment, and deep domain knowledge, are not just preserved but are now centered on the most impactful work, accelerating progress without sacrificing the essential human touch in medicine.

What Should We Teach the Next Generation?

The future-ready skills have been and will be further steered away from the rote proficiency and technical specialization of the past. Now, creativity, analytical reasoning, adaptability, and strong interpersonal skills top the charts. The Harvard Business Review notes that most leaders must cultivate “AI fluency”—a mindset that blends data-driven mastery with human intuition and psychological safety. Jobs that once served as entry points, such as administrative support or basic customer service, are thinning out. With nearly 50 million U.S. jobs in these categories potentially affected, both learners and parents would benefit from focusing earlier on broad-based skill development.

So, what is the new “it” aptitude? The answer isn’t a single subject, but a shift in mindset. The skills that guaranteed success in the past are no longer a sufficient ticket to the future. We need to move from cultivating narrow excellence in tasks that AI excels at, to fostering broad, human-centric capabilities.

From To-Do Lists to a New Core Curriculum:

Let’s be explicit about the shift. We must move our focus from the prized skills of the past:

• Rote Memorization & Speed: Excelling at standardized tests focused on accuracy and speed.

• Following Instructions: Perfect execution of predictable, routine tasks.

• Single-Skill Specialization: Deep expertise in a single, narrow field without cross-functional application.

And pivot towards cultivating these future-proof aptitudes:

• Critical Thinking & Problem-Framing: The ability to ask the right questions, not just find pre-defined answers. To look at a complex situation and identify the root problem, not just treat the symptoms.

• Creativity & Novel Synthesis: Generating new ideas, connecting disparate concepts, and applying knowledge in entirely new contexts—precisely where AI’s pattern-matching falls short.

• Empathy & Cross-Cultural Collaboration: Understanding human emotion, motivation, and nuance to lead teams, build trust, and design solutions that resonate on a human level.

• AI Collaboration & Fluency: Not just using AI tools, but knowing how to partner with them—how to prompt, interpret, critique, and integrate their output into human-led decision-making.

• Adaptability & Lifelong Learning: The meta-skill that underpins all others. The mental agility to pivot, unlearn, and relearn as technologies and industries evolve at a dizzying pace.

If I had to pick one overarching “it” factor from this list, it would be Adaptability. In a world of constant change, the ability to learn, unlearn, and relearn with resilience and curiosity is the ultimate durable skill. It’s the engine that drives all the others.

For my daughter’s generation, the challenge is not just preparing for what’s at risk, but aiming for what’s rising. Medicine, education, and advanced technical roles aren’t fading; they are being reinvented. Our job as parents and leaders is not to drill them for the tests of yesterday, but to empower them with the **learning agility** to master the challenges of tomorrow. Their success will be defined not by what they know, but by how quickly they can learn, adapt, and apply their humanity to a world shared with intelligent machines.

What To Do for Leaders and Professionals

Mapping your workforce using this 2x2 grid—physical/manual vs. digital/remote, routine vs. complex—is a first step. Identify which roles can be augmented and which face automation risk. Invest in upskilling, digital and collaborative competencies. Design future-ready careers that blend technical agility and emotional intelligence.

Ultimately, the real opportunity in AI’s era is to create a future that is both innovative and deeply human. The question isn’t “will jobs survive AI?” but “how will leaders, teams, and the next generation thrive alongside it?”

I explore future-forward topics to help leaders better envision, design, and develop the future of choice.

TBD Futures

(image and video created with the help of MidJourney)

Sources:

• https://explodingtopics.com/blog/ai-replacing-jobs

• https://www.cnbc.com/2025/09/08/ai-artificial-intelligence-workforce-impact.html

• https://budgetlab.yale.edu/research/evaluating-impact-ai-labor-market-current-state-affairs

• https://www.pwc.com/gx/en/issues/artificial-intelligence/ai-jobs-barometer.html

• https://hai.stanford.edu/ai-index/2025-ai-index-report

• https://www.brookings.edu/articles/new-data-show-no-ai-jobs-apocalypse-for-now/

• https://www.census.gov/library/stories/2025/09/technology-impact.html

• https://www.jpmorgan.com/insights/global-research/artificial-intelligence/ai-impact-job-growth

• https://www.bls.gov/opub/ted/2025/ai-impacts-in-bls-employment-projections.htm

• https://www.goldmansachs.com/insights/articles/how-will-ai-affect-the-global-workforce

• https://hai.stanford.edu/news/assessing-the-real-impact-of-automation-on-jobs

• https://www.weforum.org/stories/2025/01/future-of-jobs-report-2025-jobs-of-the-future-and-the-skills-you-need-to-get-them/

• https://www.weforum.org/stories/2025/08/ai-jobs-replacement-data-careers/

• https://hbr.org/2025/10/5-critical-skills-leaders-need-in-the-age-of-ai

• https://www.weforum.org/stories/2025/04/ai-jobs-international-workers-day/

• https://www.forbes.com/sites/bernardmarr/2025/06/30/how-is-ai-really-impacting-jobs-in-2025/

• https://hbr.org/2023/09/reskilling-in-the-age-of-ai

(video created with the help of MidJourney)

What if your AI strategy is already obsolete? Most leaders are using AI to chase efficiency, but its trillion-dollar potential lies in becoming your company's primary engine for growth. For years, AI has been touted as the ultimate tool for cost-cutting, but this focus misses its true, transformative promise.

Is your company using AI to protect the bottom line, or to invent the future? The answer will define your competitive advantage for the next decade. AI’s greatest potential is to supercharge innovation, fueling new growth and opening strategic possibilities for businesses willing to embrace its capabilities. This is a seismic shift: AI is becoming the engine of exploration, creativity, and competitive advantage across industries that have seen innovation slow and become more costly in recent decades (McKinsey).

Subscribe now

AI: From Efficiency Machine to Growth Powerhouse

Most headlines focus on AI automating tasks, shrinking payrolls, or optimizing supply chains. But the latest research clarifies a bigger story: AI is driving breakthrough R&D, accelerating product cycles, and generating outsized economic value.

• Economic impact: McKinsey estimates that AI could unlock between $360 billion and $560 billion in annual economic value by accelerating R&D innovation alone, not just making it cheaper.

• Generative power: More than 30% of new code at Google and Microsoft is now written by AI, and sectors like gaming and software have seen output rise up to 150% in some cases, far beyond what efficiency gains could yield (McKinsey).

• Productivity vs. creativity: Harvard Business Review finds that the top-performing organizations leverage AI not just for incremental gains, but as a keystone for developing entirely new products and markets.

The lesson is clear: AI’s transformational value lies not just in streamlining processes, but in unlocking new frontiers of innovation and growth.

The ‘Three V’s’—AI’s Unique Impact on Innovation

The concept of AI enhancing the Velocity, Volume, and Variety (the 3Vs) of innovation comes directly from McKinsey’s 2025 research. These levers are not generic—they represent how AI reshapes the R&D process at its core:

• Velocity: AI speeds up design and experimentation, shrinking product cycles and reducing time-to-market, so ideas can be tested and launched faster than ever.

• Volume: AI models can generate hundreds or thousands of design candidates in seconds, vastly increasing “shots on goal” compared to traditional methods.

• Variety: By sidestepping human bias, AI produces novel, outlier concepts—leading to breakthroughs in areas from protein engineering to retail design that were previously unreachable.

These aren’t abstractions. Consider Nobel Prize-winning advances in protein engineering, achieved by AI-driven teams using generative algorithms to design molecules that humans would never have imagined. Or the ability of LLMs to mine and synthesize knowledge from millions of data points, revealing new product opportunities in real time (Harvard Business Review).

Why AI-Driven Innovation Is Urgent—and Why Now

Innovation isn’t a “nice to have”; it’s critical for business survival and long-term success. Why?

• Competitive necessity: As the pace of change accelerates, businesses must adapt or be left behind. AI gives organizations the tools to outpace rivals, reshape industries, and define the future rather than simply follow it (Wharton Budget Model).

• Resilience and relevance: Traditional R&D methods are exhausted—most industries now invest more, but get less. Only those willing to embrace AI-driven transformation will stay relevant as the innovation cost curve rises (McKinsey).

• Sustainable growth: Responsible AI application helps organizations rapidly prototype, validate, and launch products that meet evolving needs in a volatile marketplace (MIT Technology Review). With global economic headwinds and shifting consumer demands, the ability to innovate at scale is now essential.

Focusing solely on efficiency may deliver short-term gains—but mastering innovation is what determines who thrives in the next decade.

How to Start: Your Innovation Action Plan

Understanding the "why" is useless without the "how." For leaders ready to act, here are three concrete steps to pivot your AI strategy from efficiency to innovation:

1. Reframe the Conversation in Your Next Leadership Meeting: Shift the agenda from "Where can we cut costs with AI?" to "Where can we create new value?" Challenge your team to identify one product, service, or customer experience that could be radically reimagined with generative AI.

2. Launch a Controlled Pilot Project: Assemble a small, cross-functional team (product, marketing, R&D) and task them with a 90-day mission: use AI tools to generate and validate at least one brand-new product idea. Their goal isn't efficiency; it's ideation and concept testing at a speed you thought was impossible.

3. Audit Your Data for Innovation Potential: Your proprietary data is your biggest advantage. Don't just assess it for integrity; assess it for innovation potential. What unmet customer needs, hidden patterns, or new market opportunities are buried within your data that only AI can uncover? This is the fuel for your new innovation engine.

The Leadership Innovation Imperative

We've been sold a limited story about AI. It’s not just a tool for doing the same things cheaper; it's the most powerful engine for innovation we have ever seen. Business leaders cannot afford to treat it as just another tool for automation. Instead, it’s time to embrace AI as the strategic core for innovation and growth. Research from McKinsey, Harvard Business Review, and MIT illuminates the path: put velocity, volume, and variety to work; empower cross-disciplinary teams; invest in deep model competence; and champion responsible, sustainable innovation.

In an era where talent, capital, and consumer loyalty flow to the boldest innovators, the question isn’t whether to leverage AI for innovation—but how soon, how deeply, and how imaginatively can you begin? Those who act now will not only stay ahead—they’ll define the future of business itself.

I founded TBD Futures to look ahead into these challenges and opportunities and to help leaders build a future of their choice. I'd love to hear from you: Where is your organization seeing the most potential for AI-driven innovation? How would you leverage new technologies in the coming years for growth?

TBD Futures