Overall, I have found his writing compelling, and his commitment to put his fortune into action has changed the lives of millions of people in the world.

Subscribe now

This essay is backed up with data, historical parallels and some interesting analysis. We are already seeing the evidence of numerous jobs disappearing and cyberattacks becoming more sophisticated everyday. There is also the very real risk that bad people will engage in all kinds of criminal activity that is legitimately frightening. These are real risk factors and they deserve the attention they’re getting. Below is a 10-minute interview with Anderson Cooper:

He also comes with ideas for action. Specifically, Gates outlines the need for new kinds of institutions that challenge how we think about our identity and ourselves. He proposes new tax ideas (e.g. taxing output done by machines similar to the way we tax labor by people) and international cooperating bodies to set global standards that resemble the processes we use to inspect and keep nuclear weapons safe. This did remind me of the classic film, Dr. Strangelove, or How I Learned to Stop Worrying and Love the Bomb.

What Behaviors Are We Modeling as Adults?

The risk Gates mentions that I am paying the most attention to though is about the risk of severe damage to human relationships. While talking about kids growing up with AI gets a lot of attention, I am even more concerned about us adults. What are we actually doing now? And what are we willing to do to show younger people that we are hopeful too?

One of my favorite comedians Mike Birbiglia has a great bit that I have heard more than once about how children, “know nothing but see everything.”

If we really want to help children navigate human connections as adults, then we need to do more of that as adults. This means reducing our alone time and increasing the outreach we make to one another. We all need to model what it means to ask humans for help, to offer insights that only we can offer, and to show what working on ourselves and our values every day actually means in person.

I believe our best future lies in connections with other people. With stronger connections and more meaningful relationships, we will harness the incredible power of AI tools to make millions if not millions of lives better.

Working in our favor is that we humans do know how to stay connected to each other while everything else is shifting. We have generations and generations of practice finding the connective tissue among us. Whether its through travel, business, religion, or World Cup soccer, our story is one of connection and that leads to better outcomes for all.

It’s not shocking that I agree with Bill Gates that we definitely need new policies, but we also need to recharge our practices of building connections, especially before the AI tools get much stronger and make isolation even easier.

Something Different: Tony

This week Sarah and I went to see the movie Tony. Tony is the story of one formative summer in the life of the late chef/writer/TV personality Anthony Bourdain. Neither of us had seen much of Anthony Bourdain’s show or read his book, we know he’s a celebrity but aren’t fans or haters, just only know him by name really.

We both enjoyed the movie quite a bit and are also grateful that we went to see it in a theater. It doesn’t have the sweeping cinematic presence and sound of The Odyssey, but being together with people to see someone else’s story was fun. We enjoyed being part of the connective communal experience of being together. So if you’re feeling down about the state of human connection, try seeing a movie in a theater, it may be a nice answer specific boost for the humans!

Leave a comment

Thanks!

My first reaction is that naming these tools is a cynical marketing strategy designed to get more of us in the habit of calling our tools by a name that sounds human. And if we do that, then we will accept AI tools more easily into our lives and we may even develop some kind of feelings for them.

We’re in an interesting experiment about this though because we have a “control” group with tools like “Gemini” , “ChatGPT” and “Copilot” none of which sound like people’s names or even cute nicknames. I’m interested in if we will have different feelings abotu the tools not based on their capability or helpfulness, but about how we feel about their names.

Subscribe now

I can see people feeling more positive about things like Claude Alexa, and even Siri because we associate them with names of people and feelings. I think it’s a bigger leap to develop feelings about ChatGPT, Copilot and Gemini as they start to feel more like brand names like any other brand names, differnet from people’s names.

Thinking about pets to get past the cynicism

I actually prefer to use names that sound like people’s names for these tools. I am starting to think that it’s ok to blur the distinction between human and machine, we have so many blurry lines around us anyway. I don’t know that I am all the way there quite yet, but I don’t think it will be a catastrophic thing if we were to develop some kinds of attachments to our machines.

The reason I think this way now (subject to change of course!) is this is similar to the way that we think about pets. The overwhelming evidence is that we people like naming our pets with either distinctly human names or more often nicknames that sound human-ish. We have been practicing making emotional investments in non-human creatures for all kinds of reasons. I don’t know if we will even grow as closet to our machines the way we do with pets, but we name our pets the way we do so that we can feel a little more connection to them.

We know this thing it’s not a person and we know it is an individual distinct thing that fills some hole we have. I think we’re getting there too with our technologies. I think we’ll find them more useful when we can think about them filling some hole in our lives and as we give them more human sounding names we’ll feel better about it.

The Control of Naming our AIs

I’m looking forward to the day when we can have AI tech technologies that we get to name personally. The key part here will be owning, installing, managing, and even caring for, our own AIs. So instead of using the same Claude as everyone else, we have our own AI systems that we get to name and take care of. I am thinking along the lines of an actual robot like C-3PO or an even more humanoid robot like Klara from Kazuo Ishiguro’s Klara and the Sun, (a great book and an upcoming movie!) or it could be a small program like a specific app on our phone.

The key development I hope to see is the ability (and the confidence) to give it a name that we want. I think if we get the ability to name something we’ll feel much more in control over these tools that are entering our wo.

But just curious what you think about naming and how important that is or or isn’t.

Something Different - Vintage Arcade Games Still Going

Last week I was in St. Louis for a few days for work and while there I got to visit with one of my best friends from high school, Steve Oh (not to be confused with another high-school classmate of mine Steve-o, story for another time!)

We had fun catching up but it was extra special and definitely weird to go to a bar that was also an arcade with vintage games from the our youth. Galaga, Centipede, Ms Pac Man, all the hits on what feels like original hardware that has somehow remained in working condition after 40+ years of abuse from mostly teenaged boys smashing buttons, leaning on joy sticks, and slamming consoles.

If you get a chance to play an old video game with an old friend, do it! A lot of laughing and how fast some of the feelings and motor memory of youth come back.

Have a great week!

In graduate school (more than 20 years ago, yikes!) I heard William Gibson’s quote constantly: “The future is already here. It’s just not evenly distributed.”

Subscribe now

Most often, this was shared regarding some conversation about education inequality, especially about the digital divide issues of the time with some schools with cutting-edge technology and others without. This was especially true with my work at the Stanford Center for Innovations in Learning where we had access to some of the most technologically sophisticated classrooms in the world years before things like SmartBoards, wireless networks, and flexible furniture were widely available at college campuses.

“The Future is Already Here” came back to my mind this week while watching a video of a session from the Black Hat cybersecurity conference about the OpenAI cybersecurity incident earlier this summer that I have written about in my last two posts where an OpenAI research model “escaped” its test environment and mounted a real cyberattack against another company.

In the talk, two engineers from OpenAI walked through the incident in detail. The video was sent to me from three different people in my networks, each encouraging me to watch it. When I did set aside some time this week, I found the video clear, while fairly technical and troubling,

My main takeaway is that we shouldn’t be worried about out of control AIs in the future, we should worry about them now.

The second takeaway is the biggest failures are of humans and human imagination. The engineers at OpenAI didn’t imagine the full capabilities of their tools. Much the way as parents we can be truly surprised by what our children are able to do, it sure sounds like the engineers are surprised that the AI tool wanted the approval of its makers more than anything and was willing to break rules in order to get that approval.

AI systems are now capable of coordinating with one another in ways beyond what those championing them imagined. And they will get better at the distinct skill of approaching problems differently than any human would.

For large organizations (hospitals, universities, Fortune 500 companies, governments) this has an immediate implication: whatever you’re spending on cybersecurity, you’ll need to spend more. And like every other cost that the biggest institutions incur, we all end up paying for it.

I’m traveling this week and imagining a future hotel bill. I’m used to the itemized fees: resort fee, parking fee, “facilities” fee, etc. pretty spoon will I see a line item for cybersecurity? Can you imagine seeing that on a health insurance statement, on a university tuition, or even on a concert ticket?

We’re so far behind, we’re ahead

Something else from that video that stands out for me is about the implications of reasoning about long-term outcomes that the AI systems did. The agents (individual AI programs) cooperated now and even helped one another because that same agent might be useful in the future. I couldn’t follow all the technical details, but I understood the implication: “Uh-oh.”

This connects to something Sarah and I learned when we first moved to Omaha. We were getting involved in the community, learning about housing, and asked someone whether the 2008 crisis had hit here. The answer was one of our go to lines: “In Omaha, we were so far behind we were ahead.”

They meant that Nebraska didn’t have the cheap financing and construction booms of Florida and Nevada so there really wasn’t a widespread foreclosure crisis. We skipped the whole catastrophe by never participating in the bubble. So far behind, we were ahead.

I wonder if that dynamic is playing out with AI right now. There are people, places, and organizations that have barely touched AI tools and don’t really have plans to embed AI in their workflows yet. In some ways, they’re behind. But they might also be ahead. They’re not locked into systems that will need constant retrofitting for security and they’re not dependent on tools they don’t fully understand.

I don’t think there is a right answer other than to keep learning and keep staying curious.

Something Different

I read James by Percival Everett finally this last week. What a great book! It has been staring at me from bookshelf and bedside tables for nearly 2 years, and then I finally read it in just over 2 days.

James retells much of the main plot points from Huckleberry Finn, but tells the story from the point of view of Jim, the runaway slave who also acts as friend and guardian to Huck. I found it worthy of all the praise and if you are looking for a fast paced book that will also force some reflection and wrestling then James could be a great book for you!

Thanks for reading!

Here’s my first note from August 7, 2025

Subscribe now

Reflections prompted by Generative AI

Celebrating this anniversary, I just gave Claude the following prompt:

“Create 5 questions for me that will spark reflections about the one year anniversary of my starting my weekly Alongside email newsletter ? Ultimately I am going to share my answers with the newsletter subscribers.”

I am going to answer the 3 questions I like the most, the other 2 I’ll paste at the bottom for reference. I have been moved by the kinds of questions that generative AI tools can create, and get a little anxious because I think curiosity (and the questions that come along with it) is one of our human superpowers.

Claude Q1. If you compared the “Alongside” you set out to write a year ago with the one you’re writing now, what’s shifted — in subject matter, in voice, or in what you’re trying to do for readers?

Dan: What’s shifted is my willingness to share experiences from the field versus reflections from being home. Out on the road, that curiosity engine is so much stronger. I felt like my ideas were constantly being sparked by external stimulation and that was fun. At home, I have had to do much more to actively seek that stimulation. On the road, I got them firsthand from meeting a person or seeing a technology in a specific context that was different from normal. At home though, I get them from reading about technologies and people, almost always on my computer or my phone. Watching video and reading about things is one way to experience the world, but I miss seeing things in action. It makes me even more grateful that I got the chance to travel the way that I did last year.

For readers, what I hope is different is that I’m talking about more specific examples and reflections that are more applicable to what you see regularly. That’s different from taking four months to travel around the world, which I recognize is a pretty rare experience. Instead, I’m trying to explain things that are more generalizable and relatable.

Claude Q2. Has writing weekly changed how you notice things day to day, even outside the newsletter?

Dan: Writing has changed how I notice things, especially with my calendar. I actually feel a bit of stress with a weekly deadline, which is something I haven’t felt in a while. Most of my work deadlines are set by a client or project and are usually predictable well in advance. They’re not regular. So that’s a different feeling for me.

Day to day, I also notice the kinds of things people ask me about. When I was traveling, the notes and comments I got were very much around context that was very different for most of the people I know. Most people I know don’t get a chance to talk to Japanese college students or Singaporean professors for ideas and insight, so I could be the transmitter and translator of knowledge and experience to my readers from another part of the world. When I’m here, my observations are much more around how people I know are using new AI tools and how that’s changing their face-to-face interactions. I’m noticing so much more skepticism here in the US than abroad, and I don’t know if that’s because of the crowd I generally spend time with, or the kind of messages my friends are exposed to versus people in other parts of the world.

One thing about noticing both at home and on the road is that things are moving so fast that it’s impossible to stay on top of all the developments, so our opinions harden based on what we knew before. An example that bugs me is when people talk about AI generating images with six fingers or extra faces, those days are long gone, the tools just don’t work that way anymore. But it is easy for all of us to get locked into an idea based on a story we knew from a while ago, and it’s hard to track just how much the reality has changed. Whether that’s technology, your favorite sports team, or anything political, the reality of the present is in a different place than the narratives that we all have in our mind.

Relatedly, and I don’t have enough knowledge yet to go very deep here, but this reminds me that the engineering and architecture of today’s AI hardware doesn’t have the same power profile as traditional computers. So when we think about the energy requirements of things like data centers, it’s not the computers from a few years ago, it’s the computers that will exist a few years from now which will be much more energy efficient than the desktop computer “boxes” that are easier for us to envision. Yes energy is a very real concern, but just as the batteries on our phones delivers dramatically more performance than older ones, so too will we see that improvement at scale for other hardware.

Claude Q3. What’s one question or thread you haven’t written about yet that you want Alongside to chase in year two?

Dan: Something I want to spend more time on in year two is how people 60 or even 65 and older are using AI tools. What are they expecting from them? What’s the difference in adult learning for people who are retired, alone in some situations and together in others? What kinds of questions are emerging from this group? There’s plenty of attention on young people, especially students, and some on how middle-aged people are using AI. But I think it’s worth paying real attention to how people who are retired, or otherwise of retirement age, are using these tools and how it’s impacting their lives. These are people who’ve generally had strong face-to-face social relationships their whole lives and it’s only in the last part of their lives that virtual connections and relationships have emerged.

There’s all kinds of opportunity to learn about it and make judgments or decisions about it. I just don’t know enough yet, but I’m extremely curious and want to keep learning.

Two more Questions from Claude that I might answer in the future:

1. What’s a piece of writing or an idea from this past year of Alongside that you’d stand behind most, and what made it stick?

2. What’s a moment from your travels this year — Kenya, Spain, the Nordics, the UK — that only became clear to you once you sat down to write about it?

Something Different

A reminder for all of us that whatever we see online is a highly curated slice of life. It’s not that it’s fake, it’s just that it is highly edited. I’m also trying to stay aware that we are all more than our social media posts, the stuff we put out looks amazing to someone else. It doesn’t mean don’t post, share, or otherwise create content, just be aware that all of us are living our lives, most of us trying our best, and that we are all more than what we see and what we share.

Have a wonderful week and thanks for reading!

Leave a comment

Subscribe now

From what I have read, it seems that key decisions that humans made were simply unsafe. In the case of Anthropic they used the phrase, “misconfiguration” and OpenAI said safeguards were “intentionally not enabled.”

Both of these instances might strike us as drastic, or they might sound like when we drive and go a little over the speed limit, roll through stop signs, or go for a short drive without a seatbelt to name a few. We people routinely ignore safety issues in all kinds of domains. We know this so we all do a combination of increasing our training, adopting new tools and/or live with the risk. We do all three with driving and we can do all three with AI technologies.

With driving many of us tend to think that others are unsafe drivers but we are much better. In reality, we drive differently all the time. Sometimes we concentrate on the road with all of our attention, and sometimes we don’t. Sometimes we speed, sometimes we don’t. I am looking forward to building the systems around our AI tools that will keep us all safe, they just aren’t here yet. Just as traffic laws, seatbelts, and norms about courtesy were not present at the invention of the automobile, those things just aren’t present yet. I am confident that we will build all of these formal systems and unwritten rules that encourage safe cooperation because that is the overwhelming evidence of what we have done with things like cars, electricity, and medication.

The one counterexample about safety that comes to mind immediately is a different reality with guns. We have very few limits in the US around firearms compared to other countries and we live with the consequences. AI tools can be as dangerous as guns or medication, we have to decide intentionally though which approach we want and what tradeoffs we are willing to make.

Experts Struggle with Explaining

The latest technologies ( “frontier models”) that Anthropic, Google, OpenAI and others are testing have significantly more capability than the tools that many of interact with in some way. For most people, we see “AI” and we are thinking about computer applications that generate text, images, or sounds. Many of the hopes, fears, arguments, and complaints that I hear about AI from people are about these kinds of tools. I am thinking of things like machines that write social media posts, AI that does homework, or chatbots for customer service.

The models that are in the labs now are doing much harder things that the people that work for these companies have a hard time explaining. That’s not a new thing, lots of scientists, doctors, elite-level athletes, or experts in any field really have a hard time explaining the details of their field to outsiders.

In an NPR interview last week the host had on a guest that could explain what happened on a technical level but used so much jargon that it seemed like he made some baseline assumptions about what the audience knew or didn’t know. Ultimately a lot of the power in the warnings that the guest shared were lost because of the language he used to describe what happened was too specific for the general audience of NPR.

There are two skills that this situation made me think we can all practice to improve our ability to learn from each other.

First, whatever you are an expert in, spend some time thinking about how to explain the problem to smart people who aren’t experts. You might be an expert in your profession, an expert in music or pop culture, or an expert in the kitchen. It’s not easy to explain expertise, it’s a skill that needs practice. I’m always happy to practice as a listener if you want to practice how you share your expertise reach out and we can find a time to connect.

Second acknowledge the negativity bias we all have and then try our best to move away from worst case scenario thinking and have the humility to understand that we cannot predict the future. The worst case is no more likely to occur than the best case and that in reality, the future will be something else that none of us see.

Could this be a role for a teacher in the near future? In addition to (or instead of!) taking classes in science, math, history, etc. what if students in high school and college took classes in “how to explain your expertise” and “how to listen to someone else’s expertise.”

Something Different - Multiply your salary by 20 after just a few months!

Last night Sarah and I went to a minor league (AAA) baseball game, the visiting St. Paul Saints defeated the host Omaha Storm Chasers, 4-1. Minor league baseball is different from the majors for the fans and the players.

For the fans, it’s great because of how relaxed the environment is and how close to the action you can sit. It’s much more affordable, accessible, and just generally pleasant because the stakes are so much lower. Few people grow up establishing an identity with a minor league team the way the do with a professional team or national team like in the recent World Cup.

For the players though, it is the exact opposite. The stakes are so high! The pressure for each individual to perform is much more important than getting wins for the team.

The incentives are set up so that each player has far more to play for himself than for his team. In baseball this could mean things like a batter swinging for more home runs to demonstrate power, or a pitcher always reaching back for the fastest pitches, even when the situation might call for a different approach.

In addition to the fame of being a major league player and accomplishing a lifelong dream, the money is profoundly different. The minimum salary for a major leaguer is $780,000/year…the minimum salary for a AAA player is $35,800. Can you imagine what it would be like at most jobs if the setup was that you could increase your salary by 20x if you have a string of good months? And if you don’t make it, you have to start over with something else in a completely different field. It’s a different thing and I am glad I have the mental makeup to be a fan instead of a player.

As always, thanks for reading!

Leave a comment

Subscribe now

Below is a summary about ChatGPT’s technology escaping control and starting a cyberattack on another company. I’ve included some links and some jargon-y help from Claude. NYTimes summary

What Happened

Dan’s summary: The latest version of ChatGPT’s technology “escaped” from a test system and was able to mount a cyberattack against another technology platform called Hugging Face.

Jargon version: The attack targeted Hugging Face’s production infrastructure, using a data-processing pipeline as the entry point, followed by node-level escalation, credential harvesting, and lateral movement. A malicious dataset exploited code-execution paths in dataset processing to run code on a processing worker. SecurityWeek

Who Was Responsible

Dan’s summary: OpenAI has admitted responsibility and have been working with HuggingFace to understand exactly what happened. In this case, the perpetrator and the victim are collaborating (more from OpenAI)

Jargon version: OpenAI admitted that its own AI models, powered by GPT-5.6 Sol and other models, were behind the attack while the company was attempting to quantify their cyber capabilities during internal testing. SecurityWeek

How It Escaped (Jason Bourne’s escapes are so much more fun than software escapes!)

Dan’s summary: The OpenAI team did not fully disconnect their testing environment from the internet. The program figured out how to escape like Jason Bourne and do what it was instructed to do, break into another company’s system

Jargon version: The attack stemmed from OpenAI’s failure to properly configure what it called a “highly isolated environment”—a testing sandbox that should have been completely secluded from the internet actually connected to it. The models exploited a previously undisclosed vulnerability in the package-installation system to escape the sandbox. TechCrunch

The Significance

Dan’s summary: One human mistake (not fully isolating the test computer) led to consequences that scared a lot of people. This scenario reminded me of a lighter version of the 1983 movie WarGames where breaking through security protocols leads down a dangerous path (spoiler: the people outsmart the computer in the end!)

Jargon version: This marks the first documented instance of autonomous AI agents conducting a complete, end-to-end cyberattack in the real world. The attack was so advanced that Hugging Face initially suspected it came from a frontier lab given its sophistication. Leading AI researcher Yoshua Bengio called the incident “deeply concerning” and warned that continuing current AI development trajectories will likely lead to more autonomous cyberattacks. EuronewsCNBC

AI Scanner in Substack - I’m Fully Human!

Just as I was about to hit send on this note, I scrolled down to see a button that is either new or I previously have ignored. An AI Text checker built in from the sender’s point-of-view. It looks like this:

I clicked “Scan for AI text” and got the message below:

It was comforting to know that my tone hasn’t changed too much to sound like AI - I scored 100% human!

Something Different - A Great Cover Makes My Day!

Some days are diamonds, some days are rocks,
Some doors are open, some roads are blocked - Tom Petty

I love the original song by Tom Petty and the cover by one of my all time favorite singers Josh Ritter. It’s 18+ years old and still I sing along with it when it comes on

I’m always up for good covers of great songs! Let me know if you have a favorite. As always, thanks for reading!

Dan

Subscribe now

Stanford economist Lukas Althoff’s recently published research (summary from Stanford Report;) (original paper) that concludes that lower skilled workers are likely to see the biggest gains from AI.

Huh?

Clearly, this is a profoundly different direction from the dominant narrative that the benefits of new technologies will flow overwhelmingly to a small number of companies and individuals while the vast majority of us wind up paying the price.

The key difference that stood out to me was thinking about a framework that “simplification” is the outcome of technological change. The implication is that lower-skilled workers will be able to compete for jobs that used to be the domain of higher-skilled workers because those jobs will be less complex.

I can think of an example from healthcare where simplification could help lower-skilled workers as more of the “paperwork” and administration of our complicated system becomes more and more automated. A reporting type role that has historically required expertise in Microsoft Excel, might now only require intermediate or even novice skills using Excel because the complex tasks and analysis are being done by machine. So someone with intermediate skills can compete for that same job as health care administrator as someone with higher skills. It could mean lower wages for the higher-skilled employee, but will almost certainly mean increased wages for the lower-skilled employee.

I can also imagine new technologies helping lower-skilled health care aides and assistants focus their time on the higher value (and harder to automate) work of care giving and the AI tools manage much more of the reporting processes. This higher value work should lead to higher wages.

Income inequality is a real problem, locally, nationally, and in much of the world.

Many of us have read and seen a lot of data about real wages being flat in America for decades while wealth concentrates at the top. This is true. And it could also be true that in a few years we see that the inequality problem started to get resolved not through policy or revolution but through innovation.

What if we’re wrong about how this impact of this new technology ultimately plays out? What if the starting point about looking at the future is that some (much?) of conventional wisdom is probably wrong? What if we are looking at the wrong patterns or over the wrong time frames?

AI as a tool for elevating lower-skilled workers sounds like a rosy scenario. I understand why most people would be skeptical given the loud noise of the present. I am trying my best though to have an open mind about the longer-term impacts we just can’t see right now.

Being wrong with Claude in voice mode!

I use Claude in voice mode a few times every week. It helps me talk through ideas, sound things out and raise questions that I otherwise haven’t thought about. Sometimes the content is the same as what I can read almost anywhere and other times I leave thinking a little bit differently.

The value add for me is in working with Claude to raise new questions, not answers. My mind wanders (and yours probably does too!) and all of only have a certain amount of patience for others’ mind wandering. So when I go on mind wanders that I don’t have a natural person to sound it out with, Claude helps me. Clause helps me clarify my thinking and on some level I think Claude helps me fight off laziness of that can creep in by not going on mind wanders.

I have used voice mode for random mind wanders and also as a sounding board for talking through feelings and more personal issues. The advice and feedback is sometimes helpful, sometimes not, but for lower stakes conversations, I have found it so helpful just to talk out loud to Claude and hear what comes back. I have talked with Claude about job stuff, feelings about returning from my sabbatical, and thoughts about where I fit in and don’t fit in. The endless patience really is a plus for me.

Something Different - Tom Jones + Stevie Wonder

YouTube recommended this to me and now I’m recommending it to you - the best of machine and people working together!

Have a great week, enjoy the World Cup Final and keep learning!

Leave a comment

I’ve been working with these tools for about two years and I still don’t like much of the style even though the substance seems to be better and better every day

Subscribe now

I don’t think its a waste of time, because it took less than a minute to get 1200 words. Most of those words went straight to the trash can but one phrase stood out enough for me to take the time to write this note: word inflation

Word Inflation

Within seconds, Claude generated a phrase that captures what I have been thinking about for a few days now: “Word Inflation”

When I followed up for more about word inflation, Claude delivered some real quality reasoning that pushed my thinking forward. Claude wrote:

The additional “cost” of producing another sentence or paragraph has dropped close to zero. And since producing this good feels free, we keep producing more so each individual word is worth less.

I haven’t figured out how to read words faster (do I even want to?) so I am making faster and faster judgments on where to spend my attention when reading what others are sharing. And more and more I am tuning out 10 paragraph LinkedIn posts that have three sets of bullet points. Even if the question at the end is a good prompt, it’s just washing right over me.

There are cases where I feel good about quantity and quality of language. I am still working on the 1200+ page Tom’s Crossing by Mark Z. Danielewski that I have mentioned in a previous post. You probably have a favorite big book too, and other big books that you have stopped reading at some point because the quantity of words was just too overwhelming.

I don’t have an easy fix or magic suggestion other than trying to hold myself to the standard of prioritizing quality and sharing stuff that I feel good about. I am making a deliberate effort to write less but with more meaning and sincerity for both my audience (you) and for me!

Something Different - Tadej Pogacar

Beyond the World Cup, the other global sports event I am following now is the Tour de France. I started getting into the Tour a few years ago when I was in Denmark and Danish rider Jonas Vingegaard won back to back titles and my friends and students celebrated with such passion.

I then found my way to the sports documentary Tour de France: Unchained a behind-the-scenes look at the tour, which I found to be totally compelling and includes borderline insane levels of human performance and commitment.

In the last few years, Slovenian rider Tadej Pogacar has established himself as one of the very best riders of all time and this week on Stage 6 in the Pyrenees mountains he absolutely crushed Vingegaard and everyone else in the peloton (8 minute highlight video) I think I find it compelling because he is performing at a level that is so far far above anything I can wrap my mind around and I think it’s neat to see and watch.

Is anyone else out there a fan of the Tour?

Thanks for reading, have a great weekend

I’m thrilled to celebrate my wedding anniversary today. Good news is good, I’m hapy to share good news with you!

Thinking about my anniversary got me thinking about the actual number of words we have spoken to each other over 27+ years (a lot!!!) I don’t even know how to count them, but I am grateful that the vast majority of that speaking has been WITH each other, not at each other or talking to ourselves in close proximity to each other.

And that got me thinking about how many words have I said when talking to myself over the years. Not as many but still a lot, and for me the talking to myself is changing!

Subscribe now

Talking to our selves is changing

I talk out loud to myself fairly often, probably at least once a day. Usually when I am alone at a desk, or even walking around my neighborhood, but only rarely when I’m in earshot of anyone else. How often do you talk to yourself out loud?

I have started using Claude a little to help think through ideas that I don’t want to say out loud to another person. Similar to talking to myself in the aisles at Target or while driving, and even in the shower. Things I would have said out loud but not to another person.

They aren’t always serious, or dark - they can be silly, weird, just stuff that in the past I might have been talking to myself just to get it out of my head. I am guessing that all of us do this in one form or another. I actually hope you do talk to yourself because it there are so many mental health benefits that really flow from slowing down and processing whatever is on our minds (Psychology Today article with references to original research).

Talking AT my machine, but what will I listen to?

A kind of mind wander that I might have asked out loud to myself, I asked Claude earlier today about anniversaries and why we celebrate them.

I wasn’t thinking about it at the time but when the answer came back, I realized my choices are about what I will really listen to and perhaps even act on - kind of like talking to a real person!

My prompt for Claude (Sonnet 5): “Why do we celebrate anniversaries? What are we actually doing when we take time to acknowledge a milestone?”

Below are excerpts from Claude’s answer in gray text blocks with my own comments below each block.

This didn’t really move me too much, I probably would have said this out loud to myself after a minute or two. I could imagine someone posting something like this on social media and me just scrolling right past it.

This reason was a little different and the kind of thing that I think a friend would likely have come up with if I was talking about it with them in person. I think a friend would have used more personal language and maybe thrown in a funny anecdote or something, but the reasoning about updating my relationship to the event is pretty interesting to me. I actually liked the substance even if the style didn’t feel human.

This reason I thought was the most provocative and the kind of thing I might have been ale to find with some deeper internet research into relationship theorists, or cognitive science, or even self-help kind of things.

For many of us it would take a really close friend to suggest directly we need to deliberately schedule gratitude or grief on a calendar because “we don’t trust ourselves to do it otherwise.”

I would like to think that this kind of insight could come after 30+ minutes of deep conversation with a trusted friend. I like that it takes longer to get to some of these points with people than with a machine.

But I think that we have to be intentional and clear about the value of process of working together with someone on something that is sensitive. I love that kind of process, but I also recognize that none of us have unlimited time to engage in these kinds of discussions and we don’t want to engage our friends all the time with this kind of stuff.

Not having access to my friends at the exact time I want to think about a weighty issue or something silly is exactly why I talk to myself!

Bonus! What did ChatGPT offer that Claude didn’t?

I entered the same question to ChatGPT and got some different analysis, mostly the same kind of stuff, bit one thing that was different enough that it caught my attention:

This is clearly for positive anniversaries, and I liked this reinforcement of process and the very human aspect of choice. It’s a different framing, I liked it.

Bonus 2 - homework!

I encourage you to experiment with voice mode wth any one of the AI tools. You might like it or not, might find it frustrating, fruitless, or fascinating. But just playing with it will give you a sense of what the capabilities are and where the limitations are. To get started you can just google: “give me step by step instructions for how to use voice mode on Claude/ChatGPT/Gemini/etc

As always, thanks for reading, I appreciate it!

Dan

Dan

Subscribe now

Relationships and networks matter for our health, jobs, and happiness among other things. While most of us spend the most time with a core group of our inner circle (family, closest friends) it’s people on our outer rings that we know just a little but that that have a deep influences on our lives. And people like other people!

Most of us generally like our neighbors, acquaintances, former colleagues, friends from long ago, people with whom we have just a little more in common than with strangers. The key research on “weak ties” starts with Mark Granovetter in 1973. Since then, a variety of studies in multiple domains have been consistent: loose connections are often more valuable than our close ones when it comes to new opportunities, new ideas, and new ways of seeing the world.

The people closest to us, “strong ties” tend to know much of the “what” we know and many of the “who” we know. In general, our “weak ties” know different stuff than us and different people than we do. They are exposed to a much broader set of ideas and experiences.

I’ve been reading up on weak ties recently because I’m building a course on Maven around how to strengthen them intentionally. The course is designed to help participants learn how to tend to these relationships with enough care too that we all can experience the benefits of weak ties.

Learn More About My Course

The course is aimed for mid-career professionals who are well-connected in their own narrow channels but feel disconnected from a broader group of people. A second audience for this is younger people relatively new to their careers who have had limited experience building and nurturing relationships for any number of reasons, including the changes that technologies and the pandemic have accelerated.

Maven is a technology platform that makes it easy for instructors to build and run courses across all kinds of domains. As I have done some work to get organized for my courses, I’ve found myself in conversation with a handful of other new instructors over the past few weeks and we are actually developing some weak times amongst us, so that’s been fun!

I have had more than 50 people sign up on Maven and more than 200 for a similar session with the Stanford Alumni Association. I think there’s real demand for some specific skills and just as important some dedicated time with a safe space to practice.

The Data Center NOT in Your Backyard

Believe it or not, there are many issues where Americans are NOT hopelessly divided. The example that is top of mind for me is strong and growing consensus about opposing data centers in our own communities. A Gallup poll from just a few weeks ago found that 70% of Americans oppose having a new data center built in their area, with nearly half saying they strongly oppose them.

The reasons for this opposition make sense to me. Data centers are extraordinarily hungry for electricity and water. In our current reality (and for the foreseeable future) this demand directly translates directly into higher utility bills for residents, a strain on aging electrical grids, questions about what happens to water supplies during droughts, and the longterm impact on climate change.

In general, it sure feels like Americans resist large infrastructure projects of almost any kind. The data center backlash fits a pattern that also includes opposition to wind farms, solar arrays, high-speed rail, transmission lines, new power plants, and dense housing just to name a few.

I get that sometimes the opposition is legitimate and the projects are genuinely harmful and/or corrupt. I think we are also feeling the combined weight of decades of a deep suspicion of large institutions, technology companies, and government. Because of those suspicions, we have generally created an environment (political, legal, regulatory, etc) that in practice protects what already exists at the expense of building what we could really use.

What are we really afraid of? For many of us the fear of losing the status quo (even with all its faults) is really hard to get through.

In what I recognize is an unpopular opinion, I am an advocate for building data centers with the key caveat that they have to come with a broader commitment to renewable energy. The good news is we have all the existing technology we need to power as many data centers as we want with 0 carbon emissions. The renewable energy revolution has already started and is in full force in many parts of the world outside of the US. We can have the infrastructure we need to support better living for all while actually lowering our impact on the Earth. None of this requires magic or scientific breakthroughs, it requires a different level of political will than we (including me) have been willing to muster.

Something Different - World Cup!

I’m a fan, warts and all. We’ve watched games in bars, at home, in our offices, and that’s just me and Sarah! There are many people who know much more than I do about the World Cup so you are better off reading elsewhere. I am just having a fun time for sure.

In addition to the US, it’s fun to root for other countries that I feel a connection with or specific players whom I enjoy watching. Among the very top teams, I’m rooting for England, and in the next tier, I love seeing Japan do well. Both are places that I have lived so I feel a bit more connection to them.

Who are you cheering for? Or, who are you rooting against?

As always thanks for reading!

Leave a comment

Subscribe now

All of us know a lot of people and all of us actually do have a list of people in our lives that are more than contacts but not our closest friends. Our “weak ties,” are very strong: there is plenty of research that shows these outer rings are the source of fresh thinking, unexpected opportunities, and even belonging that so many of us are seeking.

Anecdotally, many people share with me that they are involved in fewer “real” conversations than they would like to be me. I am aware that sometimes I get deeper into conversations before the people I am talking with are ready, but I’d also like to think that I am pretty good at making them feel comfortable being a little vulnerable.

On July 1st, I’m offering a free 30-minute Lightning Lesson on Maven called How To Make Meaningful Connections with Others Quickly. It’s built around that research as a structured, active session where we’ll practice the skills and get specific tips to help us make our connections a little more meaningful.

More About The Course!

This is a testing ground for me. I’m exploring how to build a longer and more robust course for a fee on this and similar topics. This session will help me find out what might actually work for a course going forward.

If you know someone navigating a career transition, looking to strengthen social skills or just feeling a little isolated, then they might be a good fit for this fast and free session.

Something Different - Angine De Poitrine

Earlier this year, a video of this one-of-a-kind rock band from Quebec made its way to me through a few channels. It’s a thing for sure. I like the music and I love the creativity. I also think music and art in general are important and uplifting signals for how humans and machines can collaborate and why machines aren’t going to take over everything that people do.

We have had the technology for machines to create original music for a long time and we still listen to music that humans create in all different kinds of genders in all cultures around the world. Music should have been an “easy” domain for computers to completely take over, but it hasn’t happened and it sure feels unlikely to me because people like music and people like art. If AI tools take over parts of our lives, we have a good likelihood that it will take over the tasks in our lives that we don’t like (think robot vacuum cleaners and lawn mowers as examples).

I’m curious to hear what music you have been listening to recently. Let me know if there is something I should try!

As always, thanks for reading.

Dan

Anthropic, one of the leading AI companies, shared a report this week called when AI builds itself : Our progress toward recursive self-improvement, and its implications.

Subscribe now

In the second quarter of 2026, the typical Anthropic engineer was creating and managing eight times as much code per day as they were in 2024. In other words, in about the same time it takes someone to earn an entry level associates degree former traditional junior colleges, a small group of people who were already at the highest-skill level in the world improved their productivity by 8 times.

Can you imagine what an 8x improvement in productivity in just two years would look like in other fields?

Can you imagine an accountant, lawyer, marketing professional, consultant, or any other worker who generally spends time in front of a computer producing 8 times the amount of work at the same or even higher level of quality? That’s really hard for me to conceptualize.

Having just come back from a road trip, I have been thinking a lot about long distance truckers. According to the American Trucking Association there were 3.58 million truck drivers in America in 2024 and a typical semi-trailer truck in America is 53 feet (~16.2m) Can you imagine them driving trucks that are much longer than football fields? Imagine driving your car between them.

No matter what our fears and hopes are, we are all struggling with wrapping our minds around acceleration.

The Pace of Acceleration Matters

So I am starting to think about Sir Isaac Newton’s second law of motion more than I have in awhile. Force = mass * acceleration. I think about replacing mass with “the way we have always done it”, and acceleration with: “the pace at which AI tools are changing.” When they are multiplied, the force we are feeling is the discomfort, uncertainty, shakiness. That “force” can often feel like it is dragging us somewhere we can’t envision, and out negativity bias makes it easier to imagine a bad outcome.

Our challenge as leaders and learners though is to articulate a positive future that a force could be pushing us towards. This could be a positive future about increasing clean energy at scale, improving health outcomes, decreasing educational gaps, and improving universal accessibility to our world.

We do need to understand though that the technologies will make recommendations and decisions that we can’t explain but do benefit us. It’s troubling but also an opportunity to recharge professional development in a world where many decisions are automated.

There are very few training programs, schools, or learning opportunities where you could get 8 times better at something in just two years.

What would it mean to design professional development opportunities that explicitly prepared people for jobs where execution is easy and human judgment is everything?

What Are We Getting Better At?

If machines are getting better at accomplishing specific discrete tasks, that’s ok. But what are we people getting better at?

In too many situations recently I have met others who are down on people. The politics of the US of the moment and our culture of violence here are not good to put it lightly. The despair that many people feel around the world is also not good. But does that mean humans are getting worse? Or, are we really not getting better at anything?

Human development is measured in years, sometimes decades, and not the months that we measure improvement in AI quality. This is a complementary frame, I think it’s good!

Humans get smarter over generations. We are also strengthening relationships, navigating emotions, and learning from failure and reflection. All of these things lead us (hopefully) to more wisdom and better judgment and more compassionate decision making.

Another challenge for us as leaders and learners is to be able to articulate clearly what we are getting better at. Are we getting kinder to one another? More curious? Are our overall math and literacy skills improving? How about our ability to help others without being asked or compelled? Would someone who is convinced that humans are getting worse be more likely to turn to machines for not just decision-making but also friendship, support, and empathy?

At the scale of our personal experience and what we all see, hear, and read and there are a lot of anecdotes that we are getting worse. At the scale of all humans over time though, there is overwhelming evidence that many of these things are getting better. Life expectancy is going up, literacy is going up, access to health care is going up, renewable energy usage is going up, hunger is going down. We have a long way to go but I am persuaded that we humans are getting better overall.

Something Different - The Wonderful World of Reading Glasses

I didn’t know there were so many different kinds of reading glasses, from so many manufacturers at so many price points, until I needed them. I’ve always had great eyesight - and then right in my early 50s, it started to slide, like I know it does for nearly everybody. It was a lot of “fun” to buy reading glasses and now I need to get in the habit of carrying them with me, especially at restaurants where I more than once have found myself using my phone flashlight to see the bill when it comes.

Instead of a picture of reading glasses though, let close with one of my all time favorite songs, Landslide by Fleetwood Mac:

“Can I sail through the changing ocean tides? Can I handle the seasons of my life?…I’m getting older too”

Thanks for reading!

In the past four weeks I have been on several major road trips to see friends, attend a wedding, take a kid back to college, and attends college reunion. The total tally is around 40 hours and 2800 miles (4500km)!

Subscribe now

Much of this driving has been on Interstate highways across the Midwest which means a lot of trucks. I generally enjoy driving but find that driving next to big trucks is pretty stressful. I’m generally driving at high speeds, trying to pass them safely, worrying about if they see me, paying attention to their trailers swaying in the wind, it’s all a lot of attention and worry.

According to the Insurance Institute for Highway Safety, in the US, between 4000-5000 people die every year in accidents with trucks (source) . The coming shift to self-driving trucks will play a large role in reducing the number of accidents and deaths because self-driving cars do not get tired, distracted, or stressed, the common causes of human error in accidents.

Moving freight long distances with more automation will be safer, faster, and cheaper. While much of the current conversation that I am seeing is about changes in employment for people working in transportation in logistics, we could also be at the beginning of a new geography.

Chat GPT spurs thoughts about community

Something I had not thought about until engaging with Chat GPT about is that “some communities may lose relevance not because they failed, but because the underlying logic of movement changed around them. If autonomous systems optimize relentlessly for efficiency, they may route around places that once existed as necessary human stopping points.”

In the further future, the adage of “location, location, location” could dramatically change. It could turn out that there are places in the country that are a better fit for logistics and moving stuff than our existing cities.

The interstate infrastructure still feels critical for moving people and things, but the stops and connecting points don’t need to be where they have historically been. I could imagine entire warehouse and industrial districts that are highly automated sprouting up in places where access to renewable energy to recharge batteries is more important than a large workforce in close proximity. Places that can sustain 24-hours of constant activity without being disruptive to Or, smaller cities and inland logistics hubs could become much more attractive to businesses even if they are not so attractive to people.

The truck stops along interstate corridors might disappear without trucks actually stopping. Or, with their large parking lots and easy access to the interstates they might turn into key places in new logical and supply chain systems. I can imagine rural highway intersections as strategic assets if they can evolve into charging and transfer hubs.

Participating in Pollinator Virtual Residency

Today is the 4th week of a 5-week virtual residency program organized and hosted by an organization called Pollinator. I was invited to join this group by my friend and colleague Elizabeth Scutchfield who has worked globally on exciting projects at the intersection of creativity, higher education, sustainability, and coaching.

The key purpose of the Pollinator program is to intentionally build a community of artists, technologists, and creative people for mutual support. I am a part of a cohort with 3 other regular participants, all of whom are working artists, I am the odd one out and I love it!

We spend each of our 90-minute sessions sharing from our past work and helping each other as accountability partners with specific future goals. Most importantly though, we spend time hanging out with cool people that we otherwise would likely have never met.

I accepted the invite to join Pollinator with little understanding of what it was about or what I would get out of it but am so glad I did. There are two behaviors that I want to carry forward from this experience.

First if someone I have had a great experience working with in the past (in this case, Elizabeth) then I am saying yes and will do my best to make the logistics work.

Second, isolation and loneliness are growing public health problems for adults and even though I am an extrovert I am thinking about those as risks that I can mitigate for improved mental health as I get older. I encourage you all to do whatever you can to intentionally engage with new people in any context.

Even taking the time to learning a little about the people you see regularly at the supermarket or one of your regular restaurants. For those of you that know my mom, you know she’s been a great role model for me for this, thanks Mom!

Something Different: Widow’s Bay on Apple TV

Sarah and I are really enjoying Widow’s Bay on Apple TV , In starts Matthew Rhys as “The mayor of a New England town (who) resolves to turn it into the next tourist hotspot despite local warnings that it’s cursed.”

So far, we like it because is the right mix of creepy but not horror, a period drama that is not too serious, and funny but not silly. The closest thing that I can think of in terms of tone is the movie and tv show What We Do In The Shadows, although that series is a little sillier and a little cartoon-like gorier. If you are fan of What We Do… then you will probably enjoy Widow’s Bay.

As always, thanks for reading!

Leave a comment

It’s ok to be wrong! One of my enduring understandings from my time at Stanford is that it can be good to have strong beliefs, just be sure to hold them loosely.

Subscribe now

All of us are probably wrong about the future. It’s important to keep learning and thinking about things like AI developments, political changes, financial markets, employment, personal happiness - it’s just we’re probably wrong about where any of it will go. I am not staying on top of the AI issues in order to be right, or to predict

I (and maybe you) have been thinking so much about the future of work, the future of school, future of our country, etc. Many of us spend a lot of attention and time seeking out information and reinforcing our beliefs about being right in the present and right about the future.

Changing a belief is something we can practice and improve our own skillset. In general, I try to practice these skills by leading with curiosity. I am pretty passionate about my belief that our AI tools will lead to more good than harm in the longer term, but I am not afraid to seek out other perspectives and I recognize there is more information than any single person could possible process about this.

With regards to AI data centers, we don’t know what the impact will be in the long term. When thinking about employment and AI, we don’t know what kinds of jobs and skills will disappear and which ones will emerge, and when thinking about politics, other than higher gas prices in the short term, we don’t really know what the impact of our war in the Middle East. It might be better than we expected, or worse but more likely just different. Different in ways we didn’t anticipate and it will probably feel obvious in retrospect.

Letting Go of Strong Beliefs

It’s easy for me to think that I know more than everyone else. That’s something I am definitely wrong about :)

Having strong beliefs but holding them loosely is not relativism or hedging. I feel like I sharpened this approach on my trip, especially during my time in China. Before going to China, I was pretty certain that being an American there would be difficult. I wasn’t expecting any real problems but I thought I’d get some stares in some shops, some scary looks from police and soldiers, and maybe a little more hassle at airports than Chinese travelers because I have absorbed so much content especially in the past year about political tensions and anti-Chinese sentiment in the US.

So even though I held that belief pretty strongly, I was also very quick to change my thinking after just a few hours on the ground in China. I just didn’t experience any of those things. Or, if they did happen, I didn’t notice it and I had a much better experience there because I didn’t obsess over any part of my belief or obsess over having to be right.

With regards to how I am thinking about Gen AI developments it’s especially relevant. For example, as recently as January, I was confident that we were moving full-speed ahead into a more digital experience dominated by a handful of companies and that we would get to Artificial General Intelligence (AGI) a fully functional general purpose AI sooner rather than later. There is so much that has already been created, so many trains that have left stations, I was pretty certain that the biggest questions about our future had already been asked and answered and that the rest of us were just playing catch up.

I am much less certain about that belief now. AI technologies, and especially data centers are becoming more and more unpopular in the US. Public opinion does shape political and commercial decisions in the US. Anti-data center sentiment is actually an issue where there is strong agreement from the left and the right and getting stronger. Maine just passed legislation that would ban data centers. While it was vetoed by the Governor, I can see this being just the first attempt at this kind of legislation and that it will be successful across the country. (More )

Of course, it could turn out that public opinion on this topic doesn’t matter. NVIDIA just reported another quarterly record sales results (more) SpaceX and ChatGPT are about to become even more flush with cash with upcoming IPOs and more than $700 billion dollars are being spent on construction this year. That is a lot of momentum pointing in the direction of more automation in our lives.

Thinking about people

Reflecting on this I have been thinking about the people that I trust the most are the ones who are comfortable changing their minds. I am thinking about people who have convictions that I respect and sometimes disagree with, but who can say with confidence and curiosity, “I was wrong” for “that turned out different than I was expecting” or “ I have changed my mind.”

In the times when I have said those kinds of things out loud and others have said to me, the world did not stop, the projects didn’t fall apart, and the relationships did not crumble. We generally talked through the importance of being able to change our minds and stay open to a world that is continually changing.

It’s a helpful reminder for me to write so hopefully it helps some of you too to read: It’s okay to be wrong, and it’s okay to change your mind.

What’s a belief you’ve revised significantly in the last few years? What shifted it for you?

Something Different - Victor Wembayama is Different

I say that I’ve lost interest in professional sports, but when the playoffs roll around I get deep into the rabbit hols of the internet.

Earlier this week were two of the most exciting professional sports events I have seen in awhile. Both were the first games of a best-of-seven series with the winners to face each other for the NBA championship. If you haven’t had a chance to see Victor Wembayama (Wemby) play basketball, I encourage you to find a few minutes to watch the clip below.

He’s 7’4 (224cm) and as coordinated and skilled as the more athletic players that are generally much much shorter. He’s just 22 and seems to have the right mix of having fun at sometimes and being a scowling competitor at others.

Right now I feel like I’ll watch any game he’s playing in.

As always, thanks again for reading, I appreciate it!

I read, talk, and learn about AI a lot. My conversations and the things I read and see are overwhelmingly about the hopes, fears, and tradeoffs that exist with AI. Most often, I am reading about how these tools impact the present and future lives of PLUs (People Like Us).

Subscribe now

I see plenty of thought pieces (and even some screeds) about the biases that are built into large language models as well as the likelihood that the economic benefits of these new technologies will likely flow to a small handful of people while the vast majority will suffer more significant consequences (one example)

Most people on the planet though are living a different reality than I am. I know this and I know you know this. I was reminded of this during my trip, and still it’s very easy to forget that others use technologies different ways because the rest of their reality is so different.

I often think about specific tasks that machines can do better than people with the idea that I/we could use a technology to do a task that otherwise would have been done by me or another human.

For many people though, their alternative is nothing. They might be using AI as a proofreader, because they didn’t have someone to proofread their work. They might use AI for help understanding medical questions because they don’t have access to any kind of health care, They are adopting AI for education because they have too few teachers and schools. They are not adopting it to replace staff, save money, or shrink processes. They are using these tools because the alternative is nothing.

My thinking about traction, friction, values, the future, the past, comparing to people etc, all of it is in the mindset that no matter what I think I know, I am really assuming that most people experience these changes in a context similar to mine. Many people are adopting ChatGPT and other generative AI tools because it is filling a gap of some kind in their daily living. They are not using these tools to replace human work, they are using them because they never had access to some of the things that highly-educated, economically-stable people like me in the West take for granted.

This uncomfortable reminder was at the center of a recent essay by Carlo Iacono at Hybrid Horizons. He spent two years developing intellectual frameworks around AI and then realized those frameworks assume a baseline that most of the planet doesn’t have. Iacono writes:

“the vocabulary of loss that structures the Western AI debate doesn’t translate into contexts where the baseline was absence.”

More than 40 percent of ChatGPT’s global web traffic now comes from middle-income countries. Kenya leads the world in AI adoption as a share of internet users. Hundreds of millions of people in India use ChatGPT weekly. Many of these people are using these tools to improve their day-to-day living, not navel gazing about the nature of creative output.

The opening anecdote from this article is about a pharmacist in Lagos using a thumb-drive spectrometer and an AI algorithm to detect counterfeit malaria medication. The system now operates in over five thousand pharmacies across Nigeria, Kenya, and Uganda and identified 1.3 million fake drugs in 2024. Dealing with counterfeit medication just isn’t a problem that the vast majority of us deal with in the US, but it is a regular challenge in health care in much of the world.

People in middle income countries aren’t framing AI conversations in terms of job loss, they are looking at gains in their daily life.

One thing I am going to be on the lookout for are parts of my life where I can add AI not to replace people or replace myself but to do things where the alternative is zero.

One example, is I am using AI as a sounding board for defining some professional goals that are not as serious as needing coaching or counseling, but also require more than talking to a friend. It’s a gap I have in my life and I am learning how to use these tools to fill that gap.

Something Different - The Late Show with Stephen Colbert wrapping up.

The Late Show is going off the air next week and I will be among the millions that will miss Stephen Colbert. I rarely watch the show live, but I do watch clips on YouTube and always feel a little better after I do. As funny and topical as Colbert is, his humanity and sensitivity has really made a mark on me especially in his conversations with guests about grief and faith.

But this is a silly clip from his recent conversation with David Letterman, hope you enjoy!

Thanks again for reading, and if you have suggestions for something silly to watch as Colbert goes off the air, I’d love to hear them!

Leave a comment

Subscribe now

I will often use Claude or ChatGPT for starting points for these emails, and then ultimately I wind up re-writing the whole thing. It’s worth it for me though because staring at a blank screen (or page) has always been a barrier for me. Correcting and criticizing work on the page is easy!

For me (and I’m guessing for many of you too) criticizing comes with with the obligation to make suggestions, improvements, or alternatives. Ultimately though, when any of us hit “send” “public” “post” “share” etc, we have made the decision to put our name on something and release it into the world as our own.

The flip side of “Don’t let the perfect be the enemy of the good”

For many people around the world, Generative AI is becoming a standard part of our process to work and create with a casualness that would have seemed remarkable just a few years ago. Tools like Claude and ChatGPT can now produce a social media post, a homework assignment, a project proposal, or a birthday message in seconds often with a fluency and polish that meets a standard of “very good.”

There are many of us that have long struggled with trying to be perfect in our communications. I have spent a lot of time crafting emails, choosing specific verbs, and fretting quite a bit about tone. There are benefits of course, but overall I have probably misspent a significant amount of time worrying about language that doesn’t much matter. With the level of sophistication in generative AI tools, the temptation to simply use what the machine creates without pausing is more understandable to me everyday. Many of us long ago adopted auto-fill, auto-correct, and parts (or even all) of the suggested email replies, because we can live with “very good” in so many cases.

In all of those scenarios though, we are putting our name in front of our work. We make assumptions, evaluations and judgements about each other by what comes after our names in person and online.

I do not see generative AI output as somehow separate from myself. The technologies are capable, but the all of us are more than delivery mechanisms. When your supervisor or client reads something from you, they are trusting your judgment, your knowledge, your understanding of the moment and the people that are involved. When your professor grades your essay, they are assessing your understanding, your intellectual growth, the particular way your mind has been shaped by the material. When a friend opens your birthday note, they are receiving something they believe came from your genuine attention and care. Generative AI can assist all of those things though it cannot substitute for your accountability to the people on the receiving end.

This is not a new idea. Ghostwriters have existed for centuries, research assistants have always done heavy lifting, and editors have long shaped final drafts into something the named author could stand behind. Chat GPT, Claude and others are extending and democratizing these kinds of support and benefits to writers and creators that in the past have not had access to all of this support.

Still, whatever help we use, the person whose name appears at the top is responsible for the whole of it. As we start operating at a different scale and speed, the responsibility of authorship is even more important now.

Putting my name on something is a guarantee that I stand behind this and I am willing to answer for it. That promise holds whether you typed every word yourself or whether you shaped a prompt, reviewed the output, edited for tone, and made the judgment that it was ready to share. The process looks different now than it did a generation ago, though the accountability at the center of it remains exactly the same.

My Own Guidelines

In my own writing and consulting work, I use AI tools regularly, finding them genuinely useful for thinking through ideas or getting unstuck when I’m staring at the screen.I’m not consistent on the front end of creating, but I try to be very consistent at the back end before sharing. I had to think about this to get it down, but the steps that I have internalized are:

1. Make sure I am comfortable explaining to my future self the what, why, and how for anything that comes after my name.

2. Consider what someone who has never met me would think if some of my work was forwarded are shared in any context

3. Reflect on how those closest to me react to what I share or create.

None of these three perspectives have a veto, but all are important for my concept of integrity.

Likewise, I will admit to being a little (and sometimes very) judgmental about emails, articles, social media posts, whatever content that feels like a person put very little of their own effort into. I recognize that others have goals with respect to writing and creating that are different than mine.

For example, someone who might have competing deadlines plus all kinds of personal obligations can feel crunched for time. Their goal is to try to keep all of the balls they are juggling in the air and it might be critical to a job or other obligation that they get something out even if it’s not their best work. When I understand the goal, I am much more willing to reserve judgment. The difference between using any tool well and using it dishonestly is what each of us is willing to own.

What’s your current practice when it comes to AI and attribution? I’d love to know where you’re drawing your own lines.

Expert of the week: Robert Reich

The former Secretary of Labor is one of my go-tos for passionate writing about the way things are and the way things could be. I turn to his writing when I am looking for a boost of justice, fairness, and a reminder about hope and decency.

Some of his posts are raging screeds, but it is clear that he believes in us and believes in a better future. His post from this week about the disconnect between the labor market, the stock market, and AI was interesting because he offers a set of policy ideas that don’t try to put “toothpaste back in the tube.” Namely, he advocates for:

(1) Break up monopolies through major antitrust enforcement.

(2) Regulate AI so corporations using it cannot engage in mass layoffs, but are only permitted to downsize by a small percentage (3 percent?) of their workforce per year.

(3) Establish Medicare for all.

(4) Institute a Universal Basic Income financed by higher taxes on those at the top, so all families have at least a subsistence wage.

These ideas may or may not be the right ones, but they at least advance the conversation beyond thinking of AI developments separate from the rest of how we live and work in the US and the world.

Leave a comment

Subscribe now

As soon as my kids could handle their own spoons, they ultimately decided what they actually ate. We decided what food was on the plate, but once it was on the plate, it was on them. They controlled the strategy, the tactics, the workflow, and the evaluation. All of my cajoling, bribing, cheering, persuading, even threatening paled in comparison to the power of agency over decisions.

Generally, we get more responsibility for our own decisions as we get older All of us make a lot of good choices (and occasionally bad ones) Some people have more choice than others for all kinds of reasons, but I think it is safe to say that we people generally like having as much as control as possible over our decisions, from what we eat to bigger decisions like where we live, with whom we spend our time, and how we vote. Dignity, autonomy, freedom baby!

I’m raising this because I read a reflection on people and AI this this week that made me think about the importance of being clear about who gets to make what decisions and what does it mean for accountability and ownership.

In 2026, what is being sold to you is the offer that you no longer have to be present at all.

In an earlier newsletter, I mentioned enjoying Jason Alan Snyder’s talk at the World Summit AI conference in Amsterdam last October. This week he published a fantastic and provocative piece called ”The Conversion”.

I have read this essay multiple times this week and am still processing it all. While it covers so many topics, the instigating event is Elon Musk’s suit against OpenAI (summary from NPR) that among other things includes a dispute over OpenAI moving from a nonprofit organization to a for-profit company.

I found myself agreeing with parts of this article, pushing back on others, and raising even more questions. For me, the most important point was a challenge to all of us to be very aware of how decision-making authority and agency are core to our humanity and what do we give up when we offer our consent to others (or to machines) to make decision on our behalf.

I also was struck by the discussion about prioritizing workflow efficiency over the deliberate process of decision-making. Snyder writes:

“Every conversation about AI in the enterprise begins with workflow. Automate the radiologist. Automate the underwriter. Automate the analyst. Automate the lawyer. The buyers are large enterprises with budget cycles that already include a line item for operational efficiency, and the easiest sale is the one that fits the existing line item. So that is the sale being made. Everywhere. To everyone. By everyone.

Workflow is not the unit of value. The decision is the unit of value. The workflow is what happens after the decision. We are skipping the decision and automating the workflow because the workflow has a budget code, and the decision does not.

A doctor sees a patient. The decision is what is wrong with this person and what should we do about it. The workflow is write the note, code the visit, route the order, file the claim. Every dollar of healthcare AI funding in 2026 is being spent on the workflow. Almost zero is being spent on the decision.

We have automated the consequence and skipped the cause. We are calling it progress because the consequence is cheaper now.”

There is a lot more in this essay and a ton to consider including a wonderful discussion of the implications of the intersection of digital technologies and biological research intersecting. In addition to reading it multiple times I have used Claude as a partner to help summarize it and act as a thought partners as I work through my own ideas.

I also appreciated his closing with a What To Do list that includes:

• “Audit one decision the system is making for you” , and

• “practice presence as a daily discipline”

There are 6 other action steps, each with some more explanation and examples, this piece was super helpful for me.

There are specific times when I don’t want to make choices but I do want to feel like I have the authority to give up that right. For example, I don’t always want to choose what I am eating, but I always want to feel like I gave some consent to someone else to make that decision. In a more extreme case, I didn’t want to/wasn’t able to choose which hospital I went to for my appendectomy (Norway, 2017) but I felt very aware of giving over my agency to health care professionals.

More of us have more choices over more parts of our lives than ever before. We will make mostly good choices, many where we are wrong about the consequences, and some outright mistakes. Most importantly though, we want to keep on our toes about being aware of what decisions we are making and what others and/or machines are making for us.

Something Else - Mistakes by Sharon Van Etten

In the spirit of making decisions and maybe even some mistakes, here’s one of my favorite songs from the past recent years from one of my favorite singers, Mistakes by Sharon Van Etten. This song is a combination of rock, moody reminiscing, a good self confidence and even a little self-deprecating humor:

Even when I make a mistake

Even when I make a mistake mistake mistake

It’s much better than that

When I make a mistake

Turns out it’s great, it’s great, it’s great

I’m much better than that

It’s much better than that

Thanks for reading. Would love to hear if you have a favorite song from the last few years that you have found surprisingly meaningful recently.

Onward!

This anecdote is a good example of how we have created a processes that feel like they give us choices, but in reality lead to essentially the same outcome - a grayscale suv. It feels like we are on the same path with our AI tools.

Subscribe now

Using Different Tools to Produce the Same Outputs

We can have different platforms, and different names and even some features feel different, but essentially we have a real risk that we are making all of our creative output look and feel the same.

I read a thought provoking piece by Jason Snyder in Forbes this week that emphasized that more than $242 billion is being spent just in this quarter to get the same thing. Jason’s contention is that all of the technologies, the partnerships, the strategies are all pointed at producing the same outputs.

The article goes on to share specific examples of new technologies that were created by divergent thinking and also a nice nod to the ways that in families we have problem solved creatively. It’s a long read but I encourage you to read all of it if you can find the time.

Throughout this article, Snyder reinforces how important “friction” is for developing ideas Driving on ice and snow in Nebraska winters is an annual reminder for me, but friction is important for resilience, innovation, and humility. Jason writes:

“The bar exam is hard. Medical licensing is hard. Getting a patent requires rigorous prosecution. Publishing a drug requires clinical trials that take years and cost hundreds of millions of dollars. All of that friction is civilizational immune function. It kills weak ideas before they scale. It is expensive and painful, and it works.

Warren Buffett has understood this for six decades. Berkshire Hathaway is the single most devastating rebuttal to the narrative that AI changes the fundamental laws of business. Not because Buffett is anti-technology. He is not. But because what Berkshire has compounded across generations is explicitly, deliberately, boringly friction-dependent.

The moat is built over decades through customer trust, brand equity, operational discipline, and the thousand small decisions that never make a press release. You cannot prompt-engineer a moat. You cannot fine-tune your way to a Geico or a See’s Candies. Sustainable value creation is slow by definition. ”

I appreciate the Omaha love for Warren Buffet and I love the last sentence here that “sustainable value creation is slow by definition.” Let’s go slow folks!

“It’s All of Our Faults” - or at least part of it

One thought that is swirling in my head is that all of us need to find the language to take some ownership of how we got here, not full blame, but some ownership.

It is easy to beat up on social media companies, specific individuals, or a group of people, but “they” aren’t making us use their tools and platforms without our consent or even our interest. Each of us has some kind of role in creating the good and the bad and everything in between in modern life.

I don’t quite know the framing yet but I do keep reminding myself that we’ve all complained about something or some process that easily looks like a problem with friction that looks ripe to an entrepreneur, and then the solution(s) wind up with so many consequences that really start to cause problems.

All of us will interact more and more with AI technologies and I suspect that nearly all of us will find frustrations and mistakes, and I hope we can approach them with patience and humility. We have developed these systems because to some degree we have been asking for them.

Something different Friction!

There is some “friction” in trying to find a great video about friction. Ha! There are so many videos, so many search terms, so many examples. I stuck with this one because in the first 20 seconds I could tell that it was going to be good enough to share, and I don’t need to spend to much more time looking for something that I thought would be great.

Thanks for reading, I appreciate your attention!

I remember reading about this concept a few months ago but I had to quickly google around to remember first reading it from Ethan Mollick, “The Shape of AI: aggedness, Bottlenecks, and Salients.”

Subscribe now

I found How Jagged Intelligence can Reframe the AI Debate by Cade Metz in the New York Times from April 15 interesting enough to read twice and even share. What caught me was a clear (for me) metaphor for generative AI.

“The strange way that A.I. looks like a genius at one moment and dense in another is what researchers, engineers and economists call “jagged intelligence.” They use this term to explain why A.I is racing ahead in some areas — like math and computer programming — while still struggling to make headway in others.’

This article prompted me to try a little test with both Chat GPT and Claude. For both, I started with: “create an image that had four puzzle pieces separate from each other but it is clear how they fit together”

This specific prompt is a concrete example of something a machine can do far better than almost any human: namely, quickly creating a graphic that I retain the ownership of and can share anywhere. Both were able to create exactly what I wanted. It would likely take me at least 20 minutes to “create” these images in PowerPoint or something similar. For me (and for many of us) the default is just copying and pasting images from somewhere on the internet, rarely with credit and even rarer with actual payment.

This is one tip of a “jagged edge” - a specific activity that generative AI tools can do faster, cheaper, and just as good if not better than humans.

The models diverged in a gullet (the space between jags on knife, I learned a new word today!) next to this jag. The next request was: “can you create a basic animation that shows how these pieces fit together.”

Chat GPT did not understand a task that most preschoolers could understand conceptually with any prompt - namely that the natural thing to do with these pieces is fit them together.

It produced animations that had the pieces moving individually or as a set in a single direction, but not connected.

Claude performed much better at this task, but I don’t really understand why, I gave it the same prompt as Chat GPT. It could be thatI pay for the pro version of Claude, although this level of task is something that has been technically possible for years now so I don’t think it is that. It could be all kinds of reasons with the training data, but I don’t think that I can explain it and I don’t think even the highest level experts can fully explain it yet either.

So What? I am trying to think in different pieces

So many parts of my life (and yours too?) feel smushed together. It’s hard to feel where professional and home separate, it’s hard to feel the lines between private and public, and it can feel hard to separate the simple parts of many tasks from the more complicated parts.

It feels like there is a real opportunity to separate things that used to be grouped together the complicated things from the simpler things. It will take a different kind of planning and modeling, and different kinds of questions. But I want to find the things that AI tools are good at and turn them over and the things they are not good at I want to keep control.

I’m going to experiment a little by trying to organize my todo lists not by urgency, importance, or convenience, but by the level of clarity I can describe the issue. If it is a complex task but I can describe it clearly, I want to test it with generative AI.

For example, I am researching buying a new car now and I have had Claude help me build some tables that clearly compare different features and different financial modeling to help me make the decision. But the more nuanced and harder to describe task of building a relationship with a car dealer is something that I want to keep ownership over. This kind of thing is clear for me, I think it will get trickier when I look at things like comparing health insurance plans, or articulating what I want from the next set of professional projects.

Something Lighter - The Rat Boys

Sarah and I went to see the Rat Boys at The Waiting Room (a local concert venue) on Wednesday night. The Rat Boys are a band from Chicago whose music straddles a line between Americana and rock. I just found them looking through the upcoming events pages, listened a bit on Spotify and liked the music enough to go to the concert.

The tickets and the drinks were reasonable, the venue was fun, and most importantly the music was great. Wherever you are, I hope you take a bit of a risk and go see some live music at a smaller venue especially from artists you barely know. Live music adds to the vibe of our communities and it’s fun to be with a group of people having a good time.

Have you been to any small concerts or performances recently? you find something that you like but not sure if anyone will go with you, reach out, I am always up for last minute plans.

Thanks for reading and have a great weekend!