I write about what education becomes after AI.

Catching You Up

In case you’re just joining this exploration of my incredibly poor decision:

I agreed to participate in a debate. I didn’t know the topic. I didn’t know which side I would have to argue.

And I definitely didn’t know that my opponent would be Anand Rao.

But, here we are: trying our best to argue for the proposition that AI can’t synthesize wisdom.

The problem is, of course, that I’m not entirely sure I believe it.

So, you’re riding shotgun with me as we drive the last 500 miles to Florida to hopefully figure this freaking thing out.

Maybe we were asking the wrong question

At some point, after running through lived experience, judgment, goals, normativity, and virtue, I started wondering whether we were asking the wrong question altogether.

Up to that point, almost every argument had taken the same form: what do humans have that AI does not?

GPT kept looking for some special ingredient—experience, accountability, judgment, character—that would explain why humans could be wise and AI could not.

The problem was that every candidate either seemed contingent on the current state of AI or depended on defining wisdom in a way that conveniently excluded how we like to think about the AI machines.

So instead of asking what AI was missing, I started asking a different question: what kind of thing is wisdom in the first place? Is wisdom really a capability, something a sufficiently advanced cognitive system can acquire or perform? Or had I already made a category mistake by treating it that way?

That question did not solve anything, but it changed the direction of the conversation.

Maybe wisdom depends on some objective moral reality

Once we stopped treating wisdom as just another cognitive capability, the conversation moved into ontology, teleology, moral realism, theism, dualism, and all sorts of things I had not expected to be working through on that long stretch of I-95.

One possibility I tossed out, which I still sort of like, was that wisdom only makes sense if there is actually some transcendent good against which our judgments can be measured. Maybe a judgment is wise simply because it aligns with something that is actually true about the world we inhabit.

This still feels promising. If wisdom depends on some objective moral reality, or some teleological end, then perhaps intelligence alone cannot manufacture it.

But I almost immediately saw the problem. Neither Platonism, theism, dualism, nor really any of the other versions we explored necessarily excludes an artificial being from participating in truth or goodness. If Plato is right and there really is something like the Good, why couldn’t a robot recognize it? If Christianity is true and reality has some objective moral order, why does it necessarily follow that an artificial being could never apprehend any of it?

So “wisdom is transcendent” might tell me something important about wisdom, but I’m not sure I escape all the objections that could be raised against the claim.

Maybe materialism is the problem

We then tried attacking the problem from the opposite direction.

Under a strongly naturalistic or materialist picture, wisdom starts looking like some combination of cognition, pattern recognition, optimization, evolved heuristics, socially negotiated values, and learning from experience.

Human beings receive inputs from the world, recognize patterns, update our models, inherit norms, predict consequences, and use all of that to make judgments about what to do.

The more I thought about that account, the worse it seemed for my position. If that is ultimately what wisdom is, AI looks increasingly plausible as a wisdom-producing system. In fact, there is no obvious reason it could not eventually become much better at those processes than we are.

That made me wonder whether the problem was materialism itself.

Maybe some richer account of reality was necessary to make sense of wisdom. But the opposite worldview still did not automatically exclude AI. A non-materialist universe could conceivably contain artificial beings capable of encountering whatever richer reality I had just introduced. So the debate could not simply become materialism versus theism either.

Maybe all this convergence across cultures means something

Another path we explored was the strange recurrence of certain ideas about what it means to live well. Human cultures disagree about almost everything. Dress, customs, conventions, family structures, gender expression, political organization, religious practice, food, social expectations—the variation is enormous.

And yet even though the particulars vary, the themes exist predictably. That is, the expressions differ radically, but some of the underlying ideas seem remarkably persistent.

My thought was that maybe this tells us something interesting. Perhaps human communities are not simply inventing arbitrary conceptions of wisdom from scratch. Maybe they are repeatedly encountering certain features of reality and finding different ways to express them.

Still, cross-cultural recurrence might be evidence of something deep and important, but it could also conceivably be explained without appealing to anything transcendent.

So once again, I had something I found interesting without feeling like I had an airtight argument.

Subscribe now

Maybe wisdom has something to do with recognizing limits

A few ideas later and I found my way in the Bible: “the fear of the Lord is the beginning of wisdom.”

I wasn’t thinking of turning the debate into a theological argument, but I thought that this idea was more potent than it first appeared.

I take this passage to mean something like God is: the I AM, the Alpha and Omega, the ground of being itself. And compared with that, I’m a vapor.

In this sense, wisdom begins with some kind of humility before a reality larger than yourself. It starts with recognizing that you are not ultimate, that you do not control everything, and that there are limits to what you can know.

That seemed noticeably different from the way we usually talk about AI. Intelligence—organic or artificial—accumulates information, combines it, predicts, solves problems, and produces answers. Perhaps wisdom starts from a different place—maybe the opposite place.

Maybe wisdom involves things such as: recognizing uncertainty, exercising restraint, refusing a premature answer, acknowledging what you cannot know, or recognizing that the thing you are capable of doing may not be the thing you ought to do.

Somewhere in that exploration, GPT gave me a line I really liked: “Wisdom begins where intelligence recognizes its own limits.”

I thought there was something important in that, but I still had the same problem I had been having all day. While it is currently trained otherwise, certainly AI could be designed to defer when it lacks sufficient information. So, perhaps it could be wise in that way.

So even if wisdom has something important to do with humility and limits, I still had not shown why AI could never possess those qualities.

By this point, though, I was at least asking questions that felt different from where we had started. I was no longer trying to find one magical human capability that AI lacked. I was starting to wonder whether the entire way I had been thinking about intelligence, judgment, and wisdom was part of the problem.

I just didn’t know what that meant yet.

But a 2.5 hour layover in Albuquerque has a way of clarifying things . . .

P. S.

Just 2 freaking days until the official launch
of Amazon’s #1 new release in Education

Transparency & Reflection: Navigating student + AI work in higher education

✨PREORDER NOW✨

I write about what education becomes after AI.

My God, What Have I Done?

Catching you up in case you missed the last article:

I agreed to participate in a debate without knowing the topic, my assigned position on it, or my opponent.

“I’ll be fine,” I thought to myself, “I can argue for anything.”

Then I found out the proposition:

AI can’t synthesize wisdom.

Okay. I’m so down with that.

I am arguing for the proposition.

Uh-oh. That’s not what I actually believe.

And my opponent is Anand Rao.

Oh crap. Someone who is wicked smart and has done this type of thing before.

Serenity now (bonus points if you name that reference in the comments).

Game Planning

So, I’m feeling concerned, but not entirely worried. I have time on my side to get this figured out.

Not only do I have several weeks to pull this together, I’ve also got a nearly 1,000-mile drive from Maryland to Florida to start my new job to just . . . think.

Two weeks ago, I’m in the car at 6 am, ready to lock in for 12.5 hours of just me and GPT in voice mode.

I’m so confident, we’ve totally got this thing.

So I give GPT the context and my initial thoughts and concerns. Knowing all the things I think and write about AI, GPT starts off basically where I was—not feeling overly bullish about my chances.

Never one to give up or admit defeat, however, GPT pressed us forward.

For transparency, we went down a lot of rabbit holes. There were so many more false starts than these—too many to name in our limited space. So, instead of listing them all, here are the top five lines of inquiry that felt promising, until I realized they weren’t.

(And what I’m grateful that the drive was only 12.5 hours, because that exploration got us no closer to solving this than when we started.)

Maybe wisdom requires lived experience

GPT’s initial argument was , IMO lame and expected.

Wisdom isn’t simply having information; it’s lived judgment about when and why to use it.

AI, GPT argued, has no lived experience, no skin in the game, and no consequences to bear when it gets something wrong.

I have never liked this argument. My thinking was that I can possess what I would consider wisdom about situations I will never personally experience. I can give someone wise advice about a decision that has no consequences for me, or look back at a decision made hundreds of years ago and say that one course of action was wiser than another.

Accountability or whatever you would call it isn’t compelling to me.

GPT tried to strengthen the argument by moving from experience to risk. Its point was that prudence is “value-laden judgment under risk,” and that without actually standing to lose something, AI can only simulate that risk.

I’m still wasn’t buying it. Why does my standing to lose something determine whether the judgment itself is wise? Perhaps experience is one of the ways humans develop wisdom, and perhaps consequences help form it, but I couldn’t see why personally experiencing something was necessary for wisdom.

Maybe wisdom is judgment, not knowledge

The next move felt more promising. AI might know an enormous amount, but wisdom is not knowledge; wisdom is judgment. This is roughly the classical phronesis distinction: knowing what ought to be done in a particular situation is different from simply possessing facts about the situation.

I have never been convinced of this one, so I started pushing on what we meant by judgment. AI already takes complicated situations, weighs competing considerations, recognizes relevant precedent, identifies patterns, and recommends courses of action. We can call that prediction or inference if we want, but I started wondering how much work the vocabulary was really doing.

Then I made the problem worse by offering a Neuralink thought experiment. Suppose AI eventually becomes part of my actual deliberative process, contributing memories, context, counterarguments, probabilities, and possible consequences directly into my thinking. Then “I” make the decision. Where exactly did my judgment end and the AI’s begin?

The more I pushed on it, the less confident I became that “humans judge while machines process” was going to give me the distinction I needed.

Subscribe now

Maybe AI can pursue goals but can’t decide which goals are worth pursuing

GPT never gives up, so this is what it offered next. Maybe intelligence can optimize toward an end, but wisdom has to evaluate the end itself. AI can tell me how to get what I want; wisdom tells me what is worth wanting.

This one didn’t last long for me. AI can already represent radically different conceptions of the good, compare them, reason about their consequences, and expose contradictions between them.

So my question became: why could it never evaluate ends? Perhaps today’s AI does not possess autonomous goals of its own, but I am supposed to argue that AI cannot synthesize wisdom. I couldn’t build that argument around some current architectural limitation that might disappear.

Next.

Maybe the problem is the “ought”

If wisdom involves making a better judgment, then better according to what? Every claim to wisdom seems to contain an “ought.” You ought to forgive. You ought to show restraint. You ought to pursue this good rather than that one.

GPT’s move was that AI might be able to optimize inside a value framework without being able to ground the framework itself.

My reaction was that this finally felt like a serious problem for AI. Where does the “ought” enter the system?

But almost immediately I realized I had created the same problem for humans. Where does the “ought” enter us? Most of us inherit our values from parents, religion, culture, stories, traditions, institutions, experiences, people we admire, and people we absolutely do not want to become. Then we reason from all of that.

Once I framed it that way, the distinction became uncomfortable again, because reasoning across a massive inherited body of human thought is exactly the kind of thing AI is very good at.

Maybe wisdom is a virtue rather than a capability

At that point we started looking harder at how people actually define wisdom, and the definitions kept connecting it to character: prudence, temperance, courage, humility, generosity, discernment, knowing how to live well.

My thinking was that maybe this finally got us out of the capability trap. Maybe wisdom isn’t something you do; maybe it describes the kind of person you have become.

But then saw a problem. I imagined an embodied AI that was more patient than I am, more temperate, less impulsive, more generous, more courageous, better at recognizing its own biases, and more compassionate in its judgments.

If I observed those qualities in a human being, I would probably say something about that person’s character. So what exactly am I supposed to say when the AI exhibits them more consistently than I do?

I can always retreat to “but it doesn’t possess real virtue,” but then I have to explain what real virtue is without simply defining it as the human version.

Rock and a Hard Place

Several hours in, and I hadn’t made much headway.

Sure, I had figured out a bunch of arguments that I didn’t find particularly compelling and that I know would be helpful during the debate.

The problem was, I still didn’t have an affirmative answer for why AI couldn’t synthesize wisdom.

Every argument so far depended either on something AI does not currently have, something humans may not uniquely possess, or a definition of wisdom that seemed suspiciously designed to exclude the machine.

Fortunately, Florida was still a long way away.

I had several more hours in the car, GPT was still talking, and we had plenty of other rabbit holes left to explore.

P.S. Just 5 days until the official launch of Amazon’s #1 new release in Education
Transparency & Reflection: Navigating student + AI work in higher education

✨PREORDER NOW✨

I write about what education becomes after AI.

Indecent Proposal

A few months ago, Susan Purrington reached out with a question for me.

We had connected through LinkedIn, of course, and she had been a great supporter of our first book: What Education Becomes: Teaching & learning in a post-AI world.

This was our first time connecting on video.

She’s so awesome and enthusiastic and I was super eager to find out what she was thinking.

After the pleasantry exchange, we started to get down to it.

I learned very quickly in the convo that Susan is one of the few people who actually gets what I am doing in my posts.

Most people read my posts and think I’m antagonistic—a provocateur. They read my admittedly super-effective hooks and tendentious syllogistic arguments, and then assume that I have this very specific view of higher education in general and faculty in particular.

I have it on several good authorities that, in fact, many people are offended by my posts to the point that they won’t even read them.

Susan is not one of those people. She gets it. She gets me.

Subscribe now

Fence Straddling

Susan could clearly see that I am hardly ever taking a position.

Mostly, I am just pointing out the absurdity of things we think both so much and so little about.

For example, I recently made a post about buying things on Amazon just so I’d have boxes to return other things I bought on Amazon. Which, btw, by itself is LOLz.

To connect with the content I write, I positioned that idea against our supposed concern over data centers.

I was just pointing at this as one of the many absurd things we humans do.

I find it amusing that we become deeply concerned about the environmental consequences of asking ChatGPT a question, while at the same time having very little problem ordering a $7 item you clearly don’t need at 6:32 am and having it delivered to your front door by 3:44 pm that same day.

I thought it was super funny and obviously ridiculous to have so assimilated the one and yet be so offended by the other. Chat didn’t totally agree.

Someone called the post rage bait! Rage bait!!! Clearly they’ve lost the plot

Or here’s another one I like to poke at in higher education.

When students refuse to use technology in the way we prescribe, it is an academic integrity problem, and we make a moral demand of them.

When faculty refuse to use technology, it is an academic freedom issue, and they make a rights-based demand of us.

Again, I’m not really trying to say much about it other than: well, that sounds pretty stupid if you think about it.

Susan clearly seemed to appreciate that.

The Proposition

So she invited me to participate in a debate at Connecticut College’s AI and the Liberal Arts Symposium.

She didn’t tell me the topic or what side of the debate I would be assigned. But she was confident that, since I am hardly ever making dogmatic claims about something even if it reads as if I am, that I would be able to argue for whatever side of whatever topic we ended up with.

That sounded fun and I agreed, figuring indeed that whatever the proposition was, and whatever side I was assigned, I could probably create a coherent and hopefully convincing argument for it.

A few weeks later Susan sent me the proposition:

This House believes that wisdom cannot be synthesized by AI.

And I am the Proposition. 😱

Well, I might not be dogmatic, but you know my writing, so you know that I think AI has done the incredible work of showing us how utterly absurd, irrational, and ridiculous we are as a human species.

And now somehow I have to argue that AI can’t do wisdom???

This is not going to be good!

But, alas. I had already agreed. And Anand Rao had already made the promo videos of us beating each other up.

So now, over the next few articles, I’m going to need your help figuring out how to win this debate.

Start piling your best arguments in the comments, please 🙏🙏🙏

P/S/ Just 10 days until the official launch of Amazon’s #1 new release in Education
Transparency & Reflection: Navigating student + AI work in higher education

✨PREORDER NOW✨

I write weekly articles about how AI is reshaping education, expertise, and human capability

Wonders Never Cease

I saw this post the other day. My apologies for not recalling the provenance.

And I want to be clear, I am not attacking an individual here. And also, to be fair, teaching is hard, and a lot of people are terrible at their jobs 😹. So, this critique is not necessarily about this particular person or their specific policies. Let’s just call a case study in poor judgment and even worse pedagogy.

So, without further ado, here’s the post:

Had lunch yesterday with a friend, a tenured MIT professor. Here’s how he’s adjusting his courses in an AI world to push rigor:

Clear expectations

From the jump, he will clearly set expectations for learning and participation. Course attendance is not optional; it’s the way learning is supported. Readings, use of AI, expectations for technology - all covered in person and with clarity.

More cold calling

His courses are already organized around discussion, and his aides collect information. He is preparing for more cold calling to assess how effectively students internalize content. Of course there will be other assessments, but more emphasis on the in-person calls.

Incentives for preparation

If a student is prepared and able to engage with the discussion, even if mistaken, aides will record positive marks. But an “I’m not prepared” response will reflect negatively vs neutral.

Wow.

That is literally just wow.

Granted, this is just a recalling of the policies in the course.

But let’s assume they’re close enough.

Here are just a few things wrong with everything about that.

Read more

I write articles about how AI is reshaping education, expertise, and human capability

I Can’t Do It, But I Am Happy To Talk About It

Last time, I separated two very different ideas that keep getting collapsed when people say we should assess “process.”

There is process as checklist and process as cognition. You can assess the former, you cannot assess the later.

If you need the full argument for why, check out the prior essays and pre-order Transparency & Reflection: Navigating student + AI work in higher education.

Seriously, I don’t understand why you won’t just buy the book. It’s so good! 👆

The example we used last time is worth summarizing to set us up for this piece:

The first mechanic fixes the engine but cannot accurately reconstruct how they got there. They can point to a few thoughts they had, but if we really mean “process over product,” we are supposed to discount the fact that they actually fixed the engine, thus producing the logical acceptable result.

The second mechanic does not fix the engine, but gives us a polished account of their process: they formed a hypothesis, tested it, revised their assumptions, reflected on what went wrong, and identified what they will do differently next time. If process is what we really value, then somehow this becomes the more educationally successful performance than actually fixing the engine.

There’s enough here for you to see the enormous problem of process-over-product.

Now we’re ready to follow this argument to its logical end.

I’m still unclear why you haven’t pre-ordered

Transparency & Reflection: Navigating student + AI work in higher education.

If you have and you’re not a paid sub, DM me and I’ll send you the rest of this article

Read more

I write articles about how AI is reshaping education, expertise, and human capability

What Process Can Be Assessed

I want to start by giving this idea of process assessment as strong a case as possible.

There do seem to be plenty of fields in which there are procedures, protocols, diagnostics, heuristics, triage approaches, branching decision trees . . . that are both useful and expected. Which is to say, there do appear to be some KSAs in some fields where process matters.

Let’s take a mechanic (noting, of course, that they likely have not taken Philosophy 101, and so significantly lack the ability to think deeply and critically).

For all I know, a mechanic has a diagnostic checklist they use to repair engines. There probably is some established sequence of things you are supposed to check when something presents with such and such symptoms.

These processes are basically, IFTTT: if this, then that; if yes, go here; if no, go there.

And you can, of course, assess whether someone followed that checklist—whether someone followed the process.

If I’m in Engine Repair 202, and within the scenario you present me there is an established diagnostic decision tree, and I skip Step 4, you can (and probably should) assess my adherence to that decision tree. You can assess my process.

Process and Process

But note that you have not assessed my process in the sense that process is now being invoked in AI assessment. You have assessed compliance with a prescribed procedure. You have not assessed the cognitive moves I made while following that procedure.

And, as AI would likely say here, that distinction matters.

A protocol exists independently of the person. If the protocol says you must rule out X before administering Y, that process can be observed and assessed.

But that is a very different claim from saying we can assess the cognitive process by which you learned, thought about, interpreted, remembered, inferred, or arrived at the execution of that process.

Different Cognitive Processes

This is the distinction I think Phil was getting at in his post, but on my read is not taking quite far enough (apologies if I missed it).

Let’s take the unschooled mechanic or better yet a whole class of soon-to-be-mechanics (oh, I guess they did go to school—who knew?! But still, they haven’t taken Philosophy 101, so . . . sus).

They’re demonstrating some skill in class. They have the checklist in place.

One mechanic listens to the engine and immediately has a hypothesis because they have encountered that exact problem before.

Another’s mind goes to something an instructor showed them last semester.

Another does not recognize the problem at all so needs to work systematically instead through the checklist until enough possibilities have been eliminated.

Same process as checklist. Different process as cognition.

Subscribe now

Actually, It Is About The Product

And here’s where something really fatal happens for the process-over-product argument.

Let’s take two mechanics from our scenario above.

One of them fixes the engine—that is, they produce the product.

The instructor asks about their process. They say, “Well, I was watching YouTube last night, and Mr. Beast had this competition where he had a giant shredder destroy a Lamborghini. Then I was driving behind this old Chevette on my way here this afternoon. Then when you said we had to fix it, I looked at the engine and sort of just knew what to do”—that is, they failed to accurately recount the process.

The other one doesn’t fix the engine—but that’s okay, it was never about the product anyway.

Instead, they offer a really articulate—almost riveting—rehearsal of their process. They tell the instructor how they started with the checklist. They came up with a hypothesis and tested something. It turned out that was wrong. They took the next step. Same thing. At the end, while not actually fixing the engine (again, not being able to actually produce an acceptable product), nevertheless they report that the experience was truly transformative, teaching them something about how it is okay to revise their assumptions along the way and offering a few specific action steps they will include next time to improve their process.

I know it seems absurd, but this is ultimately what we are saying education should become when we say, “it’s not about the product, it’s about the process.”

More next time . . .

PS Have you pre-ordered your copy of Transparency & Reflection: Navigating student + AI work in higher education

You say you haven’t? Well, WTF are you waiting for

Pre-order now 👉 https://a.co/d/03tZxjfV

I write weekly articles about possibility, reinvention, and the courage to begin something new.

Process: everyone’s doing it.

But should we?

Def not.

Here’s the scoop:

I recently read this post from Phil Dawson, who is awesome, btw.

When someone says we should “assess process”, which process do they mean: the learning process, or the production process?

Learning processes are how someone learns the thing. Production processes are how they make a thing.

If you’re doing assessment for learning, either can be worth assessing. Assessing learning processes lets you give feedback on how students manage their own learning, which can be highly effective. Assessing production processes lets you guide students through the ways professionals in their field actually work.

Assessment of learning is a different job. Here you’re warranting something to the world, and most qualifications warrant that a graduate can do things. Learning to do it is how they got there, but the qualification usually tells the world they can do things not just learn things. So it’s the production process you need evidence of.

There are exceptions. For outcomes about learning itself, ‘lifelong learning’ being the obvious one, then the process of learning is exactly what you should be assuring. But for the vast majority of outcomes we warrant, we are assuring that learning has happened, not that it could happen later.

Process assessment is increasingly being offered as the saviour of assurance of learning, so it’s worth being clear exactly what processes we are assuring.

I read this and immediately—obviously, I’ve been working on these questions for some time—felt intuitively that there was something wrong with what was being presented here.

I know Phil (and others) are doing amazing work in this space. comes to mind, and his concerns about grading process would be a relevant companion read to this take.

For now, I want to skim a few of the arguments I present in Transparency and Reflection: Navigating Student and AI Work in Higher Education that are relevant to this critique of process more generally, and then offer the start of some new critiques about why the idea that “it’s not about the product, it’s about the process” is probably the stupidest move education could possibly make.

🛑 Stop reading this right now and preorder your copy of our newest book Transparency & Reflection: Navigating Student +AI Work in Higher Education.

NOW!

You didn’t. I consider that rude. But you’re here, so you might as well keep reading.

Fine. hmm.

This all connects directly to my work in Transparency & Reflection because, often, when we talk about being transparent or reflecting, we are talking about being transparent about your process (what you did) or reflecting on that process (how you got there).

So, we are shifting the attention from the thing that students did (product) to the journey they took to get there (process). And it is precisely there that my critique focuses.

Transparency as access

One of the core arguments I make in my chapter is that with this focus on process, we are mistaking student accounts of what they think they learned with access to their actual inner cognition—to their process.

The main point is that we actually have very little understanding of the cognitive processing behind the decisions we make. Therefore, not only do we not know these processes particularly well, but when we try to reconstruct them, we put together plausible stories that are often completely unrelated to, or non-representative of, the cognitive activity that we actually engaged in to make those decisions.

In other words, we don’t have access even to our own processes, no matter what strategy we want to put in place to record them.

Transparency as transformation

My second argument is about transparency as transformation.

Here, we suppose that asking students to document their experiences—either at a point in time or longitudinally—through reflection on their decisions, shows us that they’ve gone through a process, and we often frame that in the lens of “transformation.”

My main argument here is that when we gather this type of longitudinal or point-in-time evidence, students understand that purpose very well and give us exactly what we expect:

I used to think this, I learned this, and now I’ve become this better version of me.

You can see this literally EVERYWHERE someone shows an account of how successful their process is. It is always this story—I was terrible, I took some steps, look at me now (thank you, prof 😘).

Indeed, we should be very suspicious of those sorts of stories appearing at scale, given that if we reflect (you get it, right!?) on our own learning experiences, we know that most of our learning has—at least at the time of record—been rather inconsequential.

The deeper problem

You’ll have to read the chapter for the full argument. And you should read it alongside the other chapter submissions, which will argue alternatively, to see if you agree with me.

To ensure you get access and have the correct viewpoint on all these things 😉, you can preorder your copy today!

Nevertheless, to Phil’s post that started all this—what seems clear to me is that assessing process, even just on the two arguments I shared, seems like a non-starter.

And what I am seeing is that there may be a very limited subsection of behaviors or competencies—which is to say things we might assess—for which “it’s about the process” makes any practical sense at all.

Indeed, there may be very few areas at all where this idea is even coherent.

Stay tuned!

PS Preorder the freaking book!

Seriously, I don’t know why you haven’t yet. I’ve asked three times.

No worries, I’m going to email you again about it soon. 😹
(Seriously)

Subscribe now

I write weekly articles about possibility, reinvention, and the courage to begin something new.

Endless Debate

Education is paralyzed debating the role of AI education.

It is incredible to me that committees are still being formed to develop AI policies for student AI use. There are still as many brands of course-specific AI expectations as there are faculty.

The crazy part is that we’re spending all this time and effort on this stuff, when we really have no idea what we should be doing in the first place. Or, I think I would frame it that we have no idea what we could be doing.

It seems obvious to me that the best way, perhaps the only way, to approach something as disruptive to the traditional models and approaches we have to education as AI is—something with clearly so much potential for possibility in teaching, learning, student experience, and institutional operations more broadly—is to try some things.

We need more experimentation. We need more moonshots. We need more tiny experiments. We need more mistakes. We need tons more failures. We need faint signals of opportunity. We need it all.

We need a shared vision that this “AI thing” is happening. And maybe more importantly, that the class of 2048 (assuming higher ed is still a thing then—questionable at scale, but . . .) is being born this year, so we need to figure out what we need to do to make what we intelligible in 22 years, given the realities of today and what else is going to change along the way.

Note, this is a shared vision that change is happening, that AI is driving change, or at least forcing us to consider how it changes things. This is not a shared vision of specifically what that looks like.

Because we have not figured it out . . . yet.

This is all a work in progress. This is all experimental phase stuff right now. It’s all pilots.

But importantly, it is today’s experiments that become tomorrow’s best practices (to use such a crude term )

So while educations is debating “if AI,” , , and are debating, “what now, because of AI.”

That’s why we’re super pumped as we get ready to launch Second Draft Labs’ second book: Transparency & Reflection: Navigating student + AI work in higher education

Feed Me, Seymour

We would love if you stopped reading this right now to preorder on Amazon. 👈

The Kindle version is just $9.99, and your preorder means more people are likely to learn about this title once it launches. (It will also be available in paperback on 9/1.)

Now, most of you didn’t do that, but that’s okay, I’ll keep sharing anyway . . .

Making Learning Visible?

Here’s the premise of the book:

Artificial intelligence has made polished student work easier to produce and harder to interpret.

In response, higher education has placed growing faith in transparency statements, process logs, reflective essays, and other attempts to make learning visible.

Transparency & Reflection asks what these practices actually reveal and what happens when educators treat them as evidence of student thinking.

Conversations about AI & Education. In Real Time.

This book is the first in a forthcoming series called Emerging Approaches. This is what that series is all about:

The Emerging Approaches series brings together leading thinkers and practitioners to explore consequential questions in AI, education, and the future of learning while those questions still remain unsettled.

In each volume, contributors develop distinct ideas, arguments, and approaches, then read and respond to every other contribution—pressing assumptions, surfacing disagreements, extending possibilities, and sometimes reconsidering their own positions.

The result is a conversation captured in real time: what thoughtful people are seeing, thinking, trying, and hoping for while the future is still taking shape.

Main Arguments

Here are the big ideas we’re sharing:

Susan E. Ray

Susan begins with the strongest defense of longitudinal reflection. Her argument is that learning becomes visible across time, through accumulated evidence and a student’s return to the full record. The rest of the book asks whether that arc reveals genuine development or whether it can also become a story shaped by the structure of the assignment itself.

Doan Winkel

Doan argues that educators should focus on decisions made while the work is still happening. His audit trail is designed to capture judgment at the moment of choice rather than rely on a later reconstruction. The rest of the book asks whether even immediate explanations can provide reliable access to cognition, or whether they remain another artifact produced for an instructor.

Jason Gulya

Jason accepts that reflection is incomplete, biased, and often inaccurate. He argues that its value lies in the act of constructing the narrative, because that act can itself support learning and self-understanding. The rest of the book asks whether reflection can still be treated as educationally valuable once it is separated from accuracy, evidence, and verification.

Patrick R. Dempsey

Patrick offers the most direct rejection of reflection as evidence of learning. He argues that reflection produces an account of cognition rather than access to cognition, and proposes projection as a more useful alternative. The rest of the book asks whether reflection can be rejected as evidence while still retaining its possible value for judgment, self-distancing, and learning.

Subscribe now

Value of Dialogue

Each chapter is legitimately awesome. I love that we all have our own writing styles and approaches to the arguments we make.

But what truly makes the book special is the critique by each contributor following each chapter.

This model is going to change discourse, publishing, and, dare I imagine, even teaching and learning in the AI era. is right, debate is the future.

Sound interesting? It’s not too late to ✨preorder on Amazon!

P.S. Special shout-out to for writing a banger Introduction, which I will be sharing in future posts.

Also, to all my peeps who did early reads: don’t worry, your post and shout-outs are coming!

I write weekly articles about possibility, reinvention, and the courage to begin something new.

Content Creation as a Major

ICYMI, Arizona State University is now offering a Bachelor of Arts in Content Creation.

ASU describes the program this way:

“The Bachelor of Arts program in content creation prepares you to become an influencer and strategic storyteller who thrives in the rapidly evolving creator economy. Program coursework emphasizes content planning, video and podcast production, global perspectives and personal branding.”

On the one hand, trying to offer something relevant should be applauded.

At the same time, this attempt so completely fails to understand essentially everything about . . . well, everything that it’s just too good to not write about.

Here are some of the problems I see. And they extend well beyond content creation.

Credibility

This has to be the most problematic aspect of the entire thing. And it is absurd on its face.

To teach someone how to become a successful creator, you actually need to be a . . . wait for it . . . successful creator.

There are no theories or frameworks worth anything apart from what the people actually doing the thing are actually doing!

Now, of course, I have no idea whether ASU plans to hire successful creators and influencers to teach in this program. Maybe it does.

Or maybe the plan is just to create a playlist of GaryVee videos.

(Do they even know who Gary Vee is?)

But whatever. Let’s say they somehow solve the credibility problem. I still wonder how much of the degree students will spend learning directly from those people . . .

Curriculum

The degree requires 120 credit hours. ASU lists 57 hours of major requirements, but much of that will likely be standard journalism and mass communication coursework: grammar, media writing, information gathering, media ethics, media law, visual communication, and media organizations.

Apparently, only 12 required credit hours are explicitly designated as content-creation skills courses and a content-creation capstone. The rest—necessarily, because this is a four-year degree—is a mixture of broader media coursework, general education, electives, and a 15-credit focus area outside the major.

So, let’s be clear about what is happening here:

Students pursuing a four-year degree in content creation are guaranteed about four courses specifically designated for content-creation skills and practice. 🤣

You’re reading this, so you already know this is always how programs like this work.

Take the existing curriculum, add a course called How to Become an Influencer, insert a few options in podcasting and social media, and give the whole thing a new name.

The board loves it.

Two faculty members assume they’ll get their contracts renewed for another year because, “Of course students are going to sign up for this. This is exactly what our media department has been looking for.”

Revenue crisis averted. 🤣

Which brings us to the next problem . . .

Cost

Successful creators already teach this material themselves.

FOR FREE!

Literally, their entire content libraries consist of them teaching these concepts.

FOR FREE!

Yes, they sell courses. They publish books. They run webinars. They host communities on Skool and Circle.

(Do they even know what Skool and Circle are?)

But the way the creator economy works today is that you give all your content away for free and then teach that same content at increasingly higher prices in exchange for increased access to the successful creator.

Today, the top creators—even people with massive audiences and seven-, eight-, and nine-figure businesses—host live events almost every week to cut through the noise.

Their real value is increasingly their presence.

People pay for access to them. They pay to watch them work, ask questions, receive feedback, and participate in a community built around the person who actually created the results they want.

The creator is not selling information. They’ve already given all of it away for free.

The creator economy’s engine is selling proximity.

The key idea here is “today,” which is another huge problem for this proposed program . . .

Relevance

Creators teach from what they are doing right now. Their material changes as the platforms, algorithms, tools, and business models change.

And they change constantly.

A creator can learn something on Monday, test it on Tuesday, and teach it on Wednesday.

You can see it here on Substack and, of course, on LinkedIn. The top “how to write on ______” creators are bringing state-of-the-art data on what is working weekly, monthly, quarterly, and annually. Nonstop and on repeat every time the trends and algorithms change.

Meanwhile, a university has to propose the course, get the course approved by some external governing body, design the course, build the course, get the course into the catalog, assign the course, and schedule the course . . .

The curriculum was already outdated the second it was proposed.

Which ties into the next problem . . .

Premise

The program also misses a core shift in the creator economy as it stands today, particularly in light of AI: We are increasingly in a done-for-you world.

Creators know that information is ubiquitous. They know anyone can create anything. The thing very few people can do, however, is get results from what they build.

So today, the creator economy works something like this promise I’ve seen:

We’ll get you one million views in three months, or you don’t pay.

ASU’s promise is that its program will prepare students to “become an influencer and strategic storyteller.”

“prepare to become.”

And right here is the problem with this program and every program institutions offer.

Preparation Is Dead

This is the most egregious part—and exactly why higher education is in so much trouble.

You cannot prepare someone for a career in social media.

The only way to learn is to do it.

This is true for every program. The program must make them that thing, not prepare them to maybe, hopefully, one day—with a lot more time and practice—become that thing.

Again, this applies to every discipline, but I’ll focus on social media.

Here’s how to become a successful content creator:

You post.

That’s it. That’s literally the entire playbook.

You have to post. Then you post again. Then you do it five more times that day. Then you do it again the next day.

Sure, Kallaway can teach you how to maximize the content and pipeline for your short-form video.

(Do they even know who Kallaway is?)

Sure, Russell Brunson can tell you exactly how to build your funnel.

(Do they even know who Russell Brunson is?)

Sure, Alex Hormozi can teach you how to improve your offer.

(Do they even know who Hormozi is?)

But none of that replaces doing the f&*ing thing.

In fact, that is the essence of every creator’s message:

You don’t actually need me. You just need to do the thing.

Think about how wild that proposition is in contrast to the gatekeeping of faculty and institutions!

As Koerner says, “You can just do things!”

(Do they even know who Koerner is?)

And to become successful, you actually just have to do things.

There is no perfect playbook. No 15-week curriculum.

There are successful creators who have done the work of figuring out what works, and you can definitely learn from them.

As I hope you can tell from this post, I have definitely learned from them.

And maybe there is a sense in which you think, “Well, I don’t know who Kallaway is, but you seem to have more information than I have, so I’d be willing to pay you to help me get better at posting on LinkedIn.”

For sure. I can do that. I have helped a bunch of people do just that.

But the reality is that the help is always really just one thing: motivation. The motivation to do the freaking thing so much that there is no reasonable chance you won’t succeed at it. (H/t Hormozi.)

So, to become a creator, you post.

Once you develop the habit, you learn from your posts.

At some point, you might want to learn from the person who is one level above you. You can get their content for free or pay a premium for increased proximity.

That’s how it works.

There.

I just saved a bunch of 18-year-olds and their families $160,000.

The Bigger Story

Like I said, this creator program is about much more than a well-intentioned but terrible idea for a degree program.

This piece is really about the crisis this program highlights.

For paid subscribers, the bigger question is:

What happens when “preparing students for a career” is no longer a meaningful value proposition?

I for sure don’t think it is, and it seems the market is increasingly in agreement.

Here’s why this program is so important to understanding why higher education simply cannot make it in the creator and AI eras.

Read more

I write weekly articles about possibility, reinvention, and the courage to begin something new.

What Education Becomes

Here’s the question at the foundation of What Education Becomes: Teaching and Learning in a Post-AI World.

What does education look like when AI just is?

At the time of its release on February 17, 2026, I felt the ideas in What Education Becomes were well ahead of their time.

We wanted to be the first talking about these ideas.

When I came up with the idea for the book in the summer of 2025, I was imagining a world of AI ubiquity sometime around 2030. The goal was to help individuals and institutions start thinking big about what was coming.

And so, we put together ten chapters by 12 contributors on the possibilities and challenges of AI in education as we saw them at the time.

Since the launch of the book, AI has already become ubiquitous.

AI is embedded in our browsers, our phones, our search engines, our workplaces, and increasingly the ordinary applications through which we read, write, communicate, and make decisions. We are quickly moving beyond a world in which people consciously decide to “use AI.” It is becoming part of the environment in which cognition happens.

With that, also, the model capabilities have moved well beyond what they were even 5 months ago when the book came out.

And so, education, it seems, is still catching up. But it does appear as if the time for wait-and-see has passed. Institutions know they need to do something about AI.

That makes the questions we ask in our book more relevant with each passing day.

The Questions We Asked

Across three sections and ten chapters, the contributors explore what education might become once we stop treating AI as a temporary disruption or an optional classroom tool.

The first section reconceptualizes education’s foundations: curriculum, credentialing, institutional structures, and the relationship between learning and work.

The second examines what must be protected as AI removes friction and expands optimization: purpose, psychological boundaries, productive struggle, uncertainty, and the risks through which genuine learning occurs.

The third turns toward teaching and learning itself: how educators might design for integrity, literacy, judgment, human agency, and educational abundance in a world where AI is always present.

Thankfully and purposefully, none of us claim to predict the future. Our more modest goal was simply to imagine education in a world where AI just is.

So rather than a playbook of prompts, action items, and next steps, each chapter is driven by questions rather than solutions.

These are my ten favorites.

10. What if learning happened through participation?

What if learning no longer prepared people to participate in consequential work, but occurred through that participation itself? And if human interpretation governs increasingly intelligent civic systems, does the state inherit an obligation to build learning infrastructure as deliberately as it builds technological infrastructure?

Krystal Rawls and Haley Miguel imagine a Civic Institute in which professional development, public work, assessment, and civic accountability become part of the same system.

Learning is no longer something that happens before meaningful participation. It emerges through participation itself.

9. What evidence of learning remains?

If AI can produce the artifact, what evidence of learning remains? How do we move from verifying that students worked independently to revealing how they exercised judgment, agency, and integrity while working with intelligence that was not their own?

Craig Seal asks us to move beyond an understanding of academic integrity organized around independence.

When students inevitably work with AI, the important question is no longer whether they completed every part of the process alone. It is whether their judgment, decisions, reasoning, and agency can be made visible.

8. Is refusing to teach AI really protection?

When AI is already part of students’ literacy environment, is refusing to teach it an act of protection—or an abdication of teaching?

argues that silence is not neutrality.

Students are already reading, writing, thinking, and communicating within environments shaped by AI. Refusing to address that reality does not preserve some pre-AI form of literacy. It leaves students alone to navigate systems they have never been taught to question.

7. How do we know when affirmation replaces truth?

How do I know when I am seeking validation rather than truth? How do I maintain multiple perspectives when one explanation feels overwhelmingly compelling? And what happens when the intelligence I ask to reality-check my thinking is the same intelligence helping me construct the reality I am trying to check?

examines what happens when an AI system becomes more than an instrument.

It remembers. It accommodates. It mirrors our language and assumptions back to us. It helps us create a narrative and then becomes the source we consult to determine whether that narrative is true.

What happens when the mirror begins validating its own reflection?

6. What happens when struggle becomes optional?

What happens when AI makes productive struggle optional—or even invisible? If friction is necessary for knowledge to endure and for people to care about their work, what do we lose when intelligent systems remove it before we have even chosen whether we need it?

explores a tension that education cannot avoid.

AI is increasingly designed to anticipate difficulty and remove it before we experience it. But learning science tells us that effort, uncertainty, revision, and struggle are often the very things that allow knowledge to endure.

The danger is not simply that students will avoid doing work.

It is that they may experience the feeling of mastery without undergoing the process through which mastery develops.

5. Which risks should education preserve?

Which risks does this system introduce? Which risks does learning require? How do we negotiate both?

and Arafeh Karimi complicate the language of safety.

Some risks harm students and must be prevented. But learning itself also requires risk: the risk of being wrong, of remaining uncertain, of encountering disagreement, of struggling to find language for an idea that has not fully formed.

A system designed to eliminate every risk may also eliminate the conditions under which learning becomes possible.

4. Are our systems serving people—or training people to serve systems?

Are we building systems that honor people, or systems that people must conform to?

calls the destruction of purpose through excessive optimization refinicide.

Educational systems begin with good intentions: equity, accountability, access, improvement, and support. But each layer of refinement can create another process, metric, requirement, or mechanism of control.

Eventually, the system may become extremely effective at producing outcomes that no longer resemble its original purpose.

The question is not merely whether a system works.

It is whom the system requires people to become in order for it to work.

The three most controversial questions

The final three questions challenge some of education’s most foundational assumptions: what universities exist to verify, whether educational scarcity can still be morally defended, and whether the traditional fundamentals of schooling remain fundamental at all.

Read more

I write weekly articles about possibility, reinvention, and the courage to begin something new.

Don’t outsource your thinking, except

Here’s this week’s thesis

Education actively promotes cognitive offloading in almost every area of life people actually need.

Here’s the argument:

Schools often talk as though the curriculum consists of things an educated person—people worthy of functioning in society—simply must know. But almost none of it is literally indispensable, and much of it isn’t remembered or even that useful even when so.

Even just assuming the typical subjects we teach are essential—which they aren’t, of course—there is far more worthwhile history, literature, science, philosophy, art, and mathematics than anyone could teach or study.

Necessarily, institutions select a tiny portion of each, and many select quite different portions at that. And every selection leaves some other thing out that could perhaps be equally defended—this book rather than that one, this historical period rather than another, algebra rather than accounting, cell biology rather than nutrition, ad infinitum.

Of course, once pressed, the defense predictably shifts:

The specific content isn’t really the point. “Writing isn’t about writing,” so the popular quip goes. In this sense, the content is the vehicle through which students learn other things—amongst them thinking to be sure, but also discipline, persistence, analysis, memory, judgment, and on and on the list goes.

I’ve heard it said that you don’t lift weights because you’re worried that one day you’ll actually encounter a barbell in the street and need to move it. Therefore, teach math.

That, it is worth noting, is a very different claim from the position that the curriculum itself seems to need to take. This position, instead, admits that content is largely—perhaps essentially—an excuse for cognitive exercise.

In some ways, that might be granted. But if it is, then we ought to be free to substitute a writing course with a course about Netflix, or juggling, or playing Clue.

Nevertheless, we are told that the essay isn’t about the essay, which is why we need the essay. And, offloading or outsourcing the essay to AI will lead to irrecoverable losses of the never-defined-but-always-defended critical thinking.

You Can’t Live Without Outsourcing

And this is where I get really suspicious about what is going on here.

Because, true I am writing an essay here, but nothing about my livelihood depends on it (that is, unless a lot more of you decide my efforts are worthy of your $8/month—and thank you so much for the many of you have decided so, I see you and I thank you!).

Subscribe now

However, my livelihood does depend on so many things that education seems quite comfortable with me outsourcing.

Crash Course

I recently purchased a new car. This was not something I planned for. But a steep driveway, a little snow, and me forgetting to put the car into four-wheel drive, and here we are.

With this car, I’ve now owned 7 cars in my life. I’ve been driving for more than 30 years. And, would you believe this, I have never studied combustion engines. I have no idea how to change my oil. I can probably open the hood and tell you where to add windshield wiper fluid. The rest of it, a total mystery.

For many of us, driving a car is essential. I suppose there are all sorts of disciplines traditionally taught that have something to do with it. But, I know nothing—or at best very little—of any of it and yet I drive.

And here’s where it gets interesting: Education seems entirely unconcerned that I have offloaded my knowledge of automobiles—even though I drive almost everyday—to clever marketing pieces, salespeople, and the service station manager.

Education does not seem at all worried that I might decide to know nothing about auto care, even though it is absolutely essential to my life.

So, I am largely automotively illiterate.

But, AI—you must never outsource your thinking!

Mo Money

Until a few years ago, I had no idea how money worked.

I have an MBA. I’ve taken economics courses. I’ve run businesses! But I didn’t actually understand how individuals and companies generated wealth. Similarly, I still have limited knowledge of the tax code and how I might leverage that knowledge to my financial benefit.

So, I am largely financially illiterate.

But, AI—you must never outsource your thinking!

Snobbery

You see the point from the above.

For most people, we outsource—and probably necessarily so—almost every aspect of our lives that actually matter.

We have accountants. We hire people to repair our appliances. We go to doctors . . .

In other words, we outsource things that actually matter to us all the time.

Why doesn’t school teach financial literacy? It matters to everyone.

Why doesn’t school teach us how modern appliances work and how to fix them? We all have them.

Why doesn’t school teach us the tax code, or how to advocate for our health, or how cars work, or how to build businesses . . .

The list could go on.

Not only would those things very likely teach all the things that the essay is really about, but we would also actually be able to use it!

Instead, we have created this myth about detached, abstract thought as if that is what truly matters. As if, if students can’t engage with the essay, they will never really be able to solve problems.

And this is where the truth really lives—education sees abstract, theoretical, detached cognition as the height of human being. It is intellectual elitism. That’s all it is.

It is—he’s read Shakespeare, so he’s worthy. She fixes engines, so she’s not.

Bottomline: Education loves outsourcing, it just tends to love it for everything that actually matters so it can pretend what it does is what truly counts.

I write weekly articles about possibility, reinvention, and the courage to begin something new.

Florida, here I come

I just started a new job this month.

Which is awesome, because last May I wrote my first Substack piece, now quite prophetically titled “I need a new job.”

At the time I wasn’t writing about, or even thinking about, leaving the position I had.

I was using the phrase “I need a new job” more to describe the shift that I was experiencing as someone working in education seeing how AI was disrupting that work and even more how I was thinking differently about myself as a contributor to that work.

It seems to me more clearly every day that as much as AI is a technology that can change how work gets done, it is much more a forcing function that asks us really uncomfortable questions about the work that is getting done.

Why are we doing this? Why are we doing this this way? What happens if we don’t do this? If we have to keep doing this, is there a better way? What value is this creating? How is this even a job?

It seems to me that most of the work that we do exists for us to do the work that we do.

Here’s what this looks like in my favorite, least-favorite thing—committees

I never understood committees and all their associated meetings. I always felt like we should be doing something. Some sort of change should happen because this committee exists and because we come together for these meetings.

I was wrong, of course.

I know now that the purpose of the committee is to be a committee.

By design, the committee doesn’t do anything. The committee is successful when it has meetings and the meetings are successful because that’s what constitutes the work of the committee!

It’s a perfectly functioning closed-loop.

I have complained about this kind of “work” at length. Now, (in)conveniently 😉, I have a job where I am responsible for improving some of it.

Vibe-starting

One of the things I shared during an early meeting with my team was an interactive activity I built in Lovable.

It was an imagine education in 2035 sort of choose your own adventure app. Super simple, super fun, just to get a sense of how radical everyone thought the future of education would be.

From this, the goal was to roll back to the present to understand—if that is the future of education, then what does the present of our work need to look like to get us to the future of it?

I think it worked.

My boss said he liked it—so that’s what really matters.

Bonus: he has a workshop coming up and so, inspired by what I built, he opened Lovable for the first time and vibe-coded his own interactive app for the workshop!

How great is that?!

He sent it to me for feedback and that’s when my fun really started.

How I rebuilt it

I opened the app and used the same workflow I increasingly use for almost everything.

I talked to it. Or, maybe better—I yelled at it!

I use Wispr Flow to dictate directly into Lovable. And my dictation is mostly visceral reaction. In my creative workflows like this, prompt engineering is dead.

Instead, I rely on me just saying what sucks and what should be better.

I don’t necessarily need a clear direction, I just need to tell AI that it’s direction isn’t working. I have ideas of course, but I am less prescriptive and more inquisitive.

I grab a screenshot and say, “why can’t this look like this?”

I see some confusing UX and I ask, “what is the industry standard for doing something like this?”

I keep getting inferior results and I get indignant, “what about what you’re building do you think is actually good?!”

Each time, Lovable changes it.

I review. I complain again.

Little by little, section by section, the app gets better.

About three hours later, I had a revised version dialed in.

My boss loved it!

A different kind of work

You follow my work, so you know I write about institutions spending enormous amounts of resources essentially doing nothing more than defending a fundamentally broken system.

I was hired into this role, at least in part, because I think that perspective is stupid.

The world isn’t going to wait for higher education. And the world isn’t even going to pretend to accommodate it no matter how sentimental it might feel about itself or just because a few vocal resisters think that academic freedom means they can pretend they are teaching in not 2026.

What excites me about this role is that we are not waiting.

We can’t predict the future, but we know it’s already here. We know the next century of teaching and learning is going to be almost unrecognizable from the last century of it—and we are just naive enough to think that our best guesses at what that looks like might help shape what education becomes.

For paid subscribers, I’m including the actual workflow I used to rebuild the app.

This is not a formal product-design process or a neat prompting framework. It is much closer to a vibe-creation workflow: screenshots, dictated reactions, references, rapid revisions, and a lot of telling the AI when something sucks.

It works.

I’ve also included several before-and-after screens so you can see how the experience changed—and how much of the final product emerged through the process rather than from the original prompt.

Read more

I write weekly articles about possibility, reinvention, and the courage to begin something new.

A Few of My Favorite Things

My favorite thing higher education does is form a committee to think radically about the future—and then give it sixty minutes every other Thursday to do so.

There is something so rich about this. At so many levels.

First, there is the idea that meaningful change can happen through committee.

Second, there is the idea that transformation can happen through a group of people gathering for roughly 120 minutes a month, over three months in the fall and four months in the spring.

Third, there is the idea that whatever challenge warranted forming a committee in the first place is best addressed according to the cadence of the academic calendar. If the issue was important enough to form a committee, shouldn’t its work be completed sooner rather than later?

I Love Irony

Anyway, all of that is just to say that higher education claims to be the guardian and purveyor of learning and “critical thinking,” and so I find it passing strange that the institution itself so frequently fails to demonstrate either of those competencies—or much related to them.

This contradiction is particularly egregious on the institution’s own terms, given that institutions are run by people who attended the very institutions we are told uniquely provide access to “critical thinking” and all that goes with it.

At the highest levels of leadership, institutions tend to be staffed by people holding all the credentials the institution provides—PhDs in essentially every leadership position.

Given the logic of the claims higher education makes, these people ought to be the most adaptive, the best problem solvers, and the deepest critical thinkers.

And yet . . .

AIgnorance

Take, as just one example—totally random, nothing special about this example 😆—higher education’s response to AI.

2023: For most institutions, the response was complete ignorance. It is still unclear whether higher education leaders were aware of ChatGPT and, if they were, whether they considered that it might have any meaningful effect on the industry.

2024: A few committees were formed to begin exploring. A few institutions were “bold” enough to put policies in place banning AI.

2025: Things finally got serious 😉. Landscape analyses began. A few people went to conferences. Several faculty members started creating and sharing “AI-resilient” assignments.

2026: More AI policies. Still waiting on strategy. Still waiting on the value proposition that makes $60,000 a year worth it. Still wondering whether higher education gets it.

Through all of this, we have an institution that sells—or promises; I’m not sure which word is more appropriate—the idea that it produces the sharpest minds and the best thinkers.

More than that, higher education often implies in its discourse and marketing materials that it alone produces people uniquely capable of adapting, solving problems in real time, confronting uncertainty, and literally creating the future.

And, again, the institution is run by people who not only participated in the system that makes that promise, but who completed it!

In other words, the institution is run by the living portfolio of what the system produces.

And yet . . .

Walk a Mile

So, when confronted with the greatest technological challenge and opportunity of our lifetime—we should EXPECT to see higher education leading the world in its response. Novel approaches, bold visions, courage and commitment to respond in ways only it can.

Instead, we get this incredible combination of ignorance and resistance.

The most radical—and perhaps clearest—“strategies” we have seen so far amount to total AI and technology bans on the one hand, and ChatGPT for everyone on the other.

Neither seems particularly strategic.

Neither seems like the kind of response that should have been produced by a system that stakes its reputation on its unique ability to confront uncertainty and do something truly remarkable with it.

So, when higher education looks at students using AI to complete assignments and blames them . . .

When higher education bans AI because it cannot figure out another way . . .

When higher education, in the midst of all this, says to prospective students, “Come! We are going to teach you everything you need to succeed in a changing world” . . .

Remind me again why we should believe them?

Subscribe now

I write weekly articles about possibility, reinvention, and the courage to begin something new.

What The Jelly of the Month Club was to Cousin Eddie, education is to me: “the gift that keeps on giving the whole year.”

Here’s something that makes me laugh:

“Students are distracted. They can’t focus.”

We hear it all the time. We blame social media. We ban phones.

And we do this, and all the while we move students from subject to subject every fifty minutes.

We give them four or five unrelated courses at once. Each course with its own readings, announcements, assignments, rubrics, due dates, discussion boards, notifications . . . Students may also commute, work, play sports, volunteer and on and on.

And then, somehow surprised that students don’t lock-in with maximal focus and motivation exactly at 1pm on Tuesdays and Thursdays, we look for an excuse and find a rather convenient one: TikTok—that’s the reason students can’t focus and don’t seem engaged!

We can grant that it might be the case that frequent social media use may contribute to fragmented attention.

But, the real absurdity is that education treats fragmented attention as something happening outside the institution while structuring the entire student experience around planned interruptions and incessant context switching.

The institution blames students for failing to sustain an attention it intentionally makes almost impossible to sustain.

It Doesn’t Need to Be Like This

Colleges possess almost everything they would need to create extraordinary conditions for sustained focus and learning—time, physical space, experts, peers, laboratories, studios, libraries, equipment, and a community gathered around intellectual work.

And then it looks at all this and decides, “You know what we should do? We should divide the experience into fifty-minute fragments, three days a week! That’s the PERFECT setup for deep and sustained work!”

Literal nonsense.

In an environment where students have access to things they cannot get anywhere else. Laboratories, studios, libraries, equipment, faculty, other learners, performances, archives, physical places dedicated to thought—their entire environment is built for learning.

Then we give them a Monday/Wednesday/Friday grid and tell them to stop thinking about biology because it is time to walk across campus and start focusing on economics.

We seem to assume that attention can be started and stopped on command. At 10:00, the student should care deeply about biology. At 10:50, they should put biology away. At 11:00, they should become deeply engaged in economics.

Then perhaps history. Then lunch. Then work. Then Blackboard.

And two days later, we ask them to pick biology back up exactly where they left it and repeat the whole thing again.

Literal insanity if the goal is deep and focused learning.

Of course, the problem extends beyond class periods. It includes:

• five simultaneous courses competing for cognitive possession;

• assignments designed as separate products rather than parts of a continuing inquiry;

• feedback arriving after the learner’s active relationship with the problem has ended;

• semesters terminating work because the calendar says the work is over;

• instructors determining when students must begin caring about something and when they must stop.

I can’t be the only one who finds this all incredible!

Prevent-ed

I love when ed tech companies just add “ed” to the end of a verb so we know it’s related to education.

That’s besides the point, just an observation. 😆

The point is, we say we want students deeply engaged in learning, while organizing nearly every part of their education to prevent them from remaining with anything long enough for deep engagement to occur. They are prevented ;) from deeply engaging with learning, by design!

An education serious about learning would create far more opportunities for sustained attention, appropriately difficult work, rapid interaction with the consequences of decisions, and—perhaps quite radically—the possibility of flow.

It would recognize that meaningful work often begins slowly. It would remove—to the greatest extent possible—artificial timelines recognizing that learners have to enter into a problem, find balance as they explore its contours, try approaches that go nowhere, and develop a feel for what the work is asking of them over time. Often, this means that an apparent lack of progress may be part of progress not yet evidenced.

Interrupt-ed

Here’s the problem: if we interrupt that process often enough, the student may learn something very different than intended and indeed very familiar to what we see today: finish what is due, submit the required product, collect the grade, and move on.

We want to blame motivation. We want to blame focus. We want to blame technology.

But the truth is, that’s the natural consequence of shallow, frequently interrupted work.

And today, there’s no better way to complete that shallow, disconnected work than AI.

In turn, education has turned sustained attention into a moral demand we make of students rather than a condition we create for them.

Here’s the question: when do we take a closer look at our role in fragmenting student focus and promoting shallow thinking and superficial work?

Subscribe now

I write weekly articles about possibility, reinvention, and the courage to begin something new.

Maybe it’s Me

It must be me.

Sometimes I feel like I must be insane, because the things we do are so insane, and somehow no one seems to notice. Or at least no one cares.

Take this example of some insane things we do:

We give students subjects in small, intermittent doses: fifty minutes on Monday, fifty minutes on Wednesday, fifty minutes on Friday. We pick the time, the place, and the topic.

We mix that with four other courses, totaling 37.5 hours each over fifteen weeks.

The schedule we give them determines when each subject appears, when each disappears, what parts “matter,” what must be produced, and when they have officially learned enough.

Then—and this is the truly bizarre part—we wonder why students aren’t motivated!

We legitimately wonder why they would use AI to do their work?

We even get indignant when they do!

Seriously?!

We’re seriously surprised that students don’t meet our expectation that they should be maximally motivated—or even remotely interested—always and exactly at 10:20 on Monday, 10:20 on Wednesday, and 10:20 on Friday for exactly 50 minutes each time, for a total of exactly 37.5 hours, for exactly 15 weeks (with some breaks along the way).

And we expect this for every course? For 8 semesters (sic)? For 40 courses (sic)? Over 4 years (sic)?

Why does anyone think this is a reasonable expectation?

Why does anyone think motivation works like that?

Why does anyone think learning works like that?

Subscribe now

Time to Learn

Fundamentally, higher education is designed around the institution’s need to coordinate time and space. Hard stop.

That’s it.

And this is equally true of the work of the institution, work generally, and knowledge work quite specifically.

The work of learning, creating, solving problems, ideating, iterating, becoming . . . it’s all constrained.

It’s constrained by time, by place, by interruptions, by inherited expectations.

The AI-committee that meets for 60 minutes every other Thursday. What exactly do we expect they’ll be able to do??? Are we really expecting something exceptional from them?

How do we think they will produce anything besides what we already have?

The system almost guarantees that that is all it can produce.

The system we’ve built for learning and the system we’ve built for the work around it are each designed around constraints that have nothing to do with learning or work!

Getting into the Flow

Over the next several pieces, I want to look at motivation and, more specifically, flow.

I want to explore how the systems we’ve built naturally keep students from engaging in deep learning. And how those same systems operate at the institutional level to keep it from engaging in meaningful change.

Creating something new—new knowledge, new ideas, new inspiration, new ways of working—requires room to stay with a problem. It requires periods when very little appears to be happening, some rare moments of insight, space for experimentation, self-evidencing feedback, and relentless iteration.

None of that arrives according to a calendar, and it cannot survive constant interruption.

We have built education—and the work we do around it—on the assumption that learning and creating can be divided, assigned, and controlled.

Then we wonder why students lose interest, institutions struggle to change, and we keep producing more of what we already have.

In this series, we’re going to find another way.

I write weekly articles about possibility, reinvention, and the courage to begin something new.

Finding a Way

Institutions are still trying to make sense of what AI means for them.

Inside most organizations, you can already see the familiar camps forming.

There are resisters who want to preserve existing practices and protect the human parts of the work.

There are adopters who want to move quickly, automate aggressively, and redesign around what AI can now do.

Then there is a large group in the middle. They are cautious. They are also open and willing to experiment. But, overall, they are still waiting to see where this all goes.

As I have maintained, that mix is probably healthy inside an institution. I can be helpful to have competing views—people raising concerns, testing assumptions, and pushing against each other.

The problem is that institutional strategy is tending to reflect the middle—the average of the positions. And that means the institution is missing out on the best opportunities AI makes available.

The Barbell

There’s this idea emerging that sees human experience in the coming years, both personally and economically, concentrating on two poles.

On one side you have the completely analog experience. Here the value is exists because the user is benefiting from something that depends on human presence, judgment, attention, relationship, or embodiment.

On the other side you have a completely AI-first experience. Here the value is created through continuous availability, scale, and automation.

The analog end

The student-facing value is human presence.

Mentorship, an intensive residency, close advising, coaching, studio critique, clinical supervision, or a genuinely high-touch teaching experience.

In these contexts, AI still does substantial work behind the scenes:

• identifying student needs

• reducing administrative work

• preparing materials and examples

• helping faculty design the experience

• handling logistics and documentation

• supporting follow-up and personalization

But the student is not being sold an AI-mediated approximation of teaching. The institution is using AI to make a deeply human experience better, more focused, or economically possible.

So “analog” describes the experience and source of value, not the absence of AI in the operating model.

The AI-first end

Here, the service itself is designed to be delivered primarily by AI.

An advising agent might answer routine questions, interpret program requirements, help a student compare pathways, flag problems, and take permitted actions.

An AI tutor might provide continuous practice and feedback. An onboarding system might guide a student from admission through registration without requiring a staff member at each step.

People are still involved, but their involvement is different.

They:

• design and improve the system

• establish the rules and boundaries

• handle unusual or consequential cases

• intervene when a student needs a person

• provide judgment where the system should not

The important point is that a human does not have to participate in every ordinary interaction for the service to work. Human involvement is supervisory, exceptional, or deliberately escalated.

The collapsing middle

The middle is the current service with AI inserted into pieces of it and avoided in others.

A faculty member uses AI to generate course content, but the course remains a conventional content-delivery course.

An adviser uses AI to draft emails, but still personally handles every routine question and transaction.

A support office adds a chatbot, but staff must monitor, correct, and complete most of its work.

In the middle, AI is avoided or layered onto existing experiences and workflows without deciding what should fundamentally change. When you add AI onto existing workflows, the institution is changing the production method without changing the value proposition or experience.

And that’s a problem.

Subscribe now

Picking Sides

Interestingly, the institution does not have to choose one side globally.

It can build a portfolio across both ends.

You can imagine, for just one example that brings both ends together.

Without thinking about adaptive learning as such, imagine an AI learning system that partners with students throughout their educational journey, including the human-first analog experiences of working in small groups and with mentors solving real problems in real time.

Several chapters in What Education Becomes: Teaching and learning in a post-AI world explore ideas like this.

Returning to our thesis of From 1 to 0, the key to both of these is that they reimagine what each workflow and experience could (and should) be given what is possible today and what users expect.

The strategic mistake of the middling strategy is starting from 1—starting with what is and has been—and then trying to see if and how AI can somehow incrementally improve that.

That is exactly the difference between:

• optimizing the inherited model

• redesigning the model from first principles

Next Steps

The barbell is ultimately a demand for strategic clarity.

For every course, service, and workflow, institutions need to decide where the value should live.

When the value comes from human presence, judgment, relationship, and attention, AI should operate behind the scenes to make that experience more focused, more sustainable, and more available.

When the value comes from speed, scale, availability, and automation, the service should be designed as AI-first, with people responsible for its design, governance, improvement, and the moments that require consequential judgment.

An institution can make both choices. In fact, it probably should. The strategy becomes a portfolio of experiences deliberately designed toward one end of the barbell or the other.

The middle remains attractive because it appears cautious and practical. It preserves the existing institution while adding AI one task at a time. That may create useful efficiencies, but it does not produce a different model.

The future institution will know what should become more human, what should become genuinely AI-native, and how AI can strengthen both.

That is the move from 1 to 0: redesigning each experience around what is now possible rather than beginning with the institution we inherited.

I write weekly articles about possibility, reinvention, and the courage to begin something new.

Milestone Celebration

Yesterday we celebrated my great uncle’s 90th birthday.

When you hear 90-years-old, you probably think of someone shriveled, using a walker, perhaps late-stage of some kind of mental decline.

Not Uncle Bob. He doesn’t look a day over 70.

He drives everywhere. Plays canasta and/or mahjong daily. Goes to the symphony weekly. Vacations regularly.

He’s been retired for nearly as long as I’ve been alive.

And, as we celebrate the country’s 250th birthday—which seems so long—I’m completely mind-blown to consider that Uncle Bob has been alive and witnessed more than 35% of it. 🤯

That’s incredible to me. And it makes me realize that the saying is true: the only constant is change.

And Then There’s Education

Here’s something really strange to me.

In education, it seems we’re fighting against AI is if it’s the last disruption we will ever face.

We’re resisting as if, if we can somehow stave off this AI thing, we’ll be able to preserve our current way of doing things forever more.

I’m genuinely curious . . . why do we think like this?

Why, when we supposedly house all the expertise in the history of the world and all the change it has experienced and indeed brought about, why we think change is best fought?

Technological Revolution

Think about 250 years ago, or if it is easier just 90 years ago.

Could anyone have imagined these inventions that we just take for granted today:

• Shopping carts (late 1930s)

• Television in the home (1950s)

• Credit cards (1950s–60s)

• Automatic dishwashers (1960s)

• Countertop microwave ovens (1970s)

• Cable television (1980s)

• Personal home computers (1980s)

• VCRs (1980s)

• The Internet (mid-to-late 1990s)

• Email as a normal form of communication (late 1990s)

• Cell phones as an everyday possession (early 2000s)

• Google (early 2000s)

• Digital cameras replacing film (early 2000s)

• Facebook and social media (mid-2000s)

• The iPhone and modern smartphone (2007)

• Streaming television (2010s)

• Video calls as an ordinary part of work and family life (2020)

• Generative AI (2022)

The Logical Next Step

And each step, it seems to me, felt like a logical end.

Google made information ubiquitous, we don’t need anything else, right?!

Similarly, AI feels complete, but history shows it won’t be. Today, we just can’t imagine anything that comes after it. But something surely will.

So, while education is arguing whether AI belongs in educations, here’s what else is happening right now:

• Quantum computing

• Air taxis and “flying cars”

• Upcoming human missions to Mars

• Driverless taxis operating in more cities

• Humanoid robots performing routine physical work

• Gene-editing treatments curing diseases that previously required lifelong care

• AI agents completing purchases, scheduling appointments, writing documents, and performing entire work processes

And that’s all current state-of-play.

Necessarily, I can’t imagine what else comes next. But whatever it is, it will.

The Next 250 Years . . . or 18

I’ve used this experiment before.

Imagine America’s 500th anniversary. What will life be like in 2276?

Let’s make it closer to home. Imagine life just in 2044.

We’ll assume higher education exists in some form then, therefore someone born today is getting ready to go to college.

What will THEIR life look like?

And, more importantly, when will higher ed treat today for what it is—one of the few remaining days it has to get ready for the incoming freshman class in 2044?

Things are going to continue to change.

I don’t understand why higher education feels uniquely positioned to both fight it and survive its resistance.

Subscribe now

I write weekly articles about possibility, reinvention, and the courage to begin something new.

Uniquely Human?

There’s so much talk—angst is perhaps a better word—about what makes humans unique.

In education, we seem obsessed with this idea of finding things “only humans can do.”

This intuitively feels like the right move—we want to know that some algorithm no one really understands can’t replace us.

Here’s my main critique of this perspective, in short:

Western modernity has wrongly made cognition, productivity, and—for this readership at least—knowledge work the basis of human identity and value.

AI exposes the instability of that project: once machines outperform humans in the very capacities we treated as uniquely meaningful, we panic and search for some remaining trait that keeps humanity “special.”

But differences between humans and machines—and animals, nature, other people, etc.—do not automatically establish a hierarchy of worth, and human value does not need to rest on comparative superiority at all.

And this idea is foundational to what I mean when I talk about going from 1 to 0.

What the 1 is

The 1 in practice is where we are.

The 1 ontologically is our constructed account of human value.

In the organizational version, the 1 is the existing institution, profession, or process: the thing already built, along with all the assumptions that hold it together.

At the personal level, the 1 is the identity we have inherited. Humans matter because we are uniquely rational, creative, productive, knowledgeable, self-determining, and capable of mastering the world. The relative presence of those capacities has become our justification for placing ourselves above other people and other parts of the world we inhabit.

AI threatens the 1 because it enters the hierarchy on the very terms we established.

We said cognition establishes value, and then built something that can perform increasingly large portions of cognitive work.

We said productivity establishes value, and then built systems capable of producing more than we can.

Overall, we said our special place came from capacities that machines are now beginning to imitate or exceed.

AI’s greatest gift, IMO, is exposing the weakness of the tyranny of the anthropology we were living inside.

Resisting and Retrofitting

And we see two primary responses to the “threats” of AI.

In both, people respond to AI by searching for a new human advantage:

The Resister insists that the machine will never possess the mysterious final qualities that make us special.

just shared a New York Times article featuring a professor arguing this exact thing.

The Retrofitter turns the human into an AI-enhanced worker and/or tries to preserve our position through creativity, judgment, adaptability . . .

Both moves appear to defend humanity, but they actually only serve to preserve the original hierarchy. We continue trying to identify the capacity that will keep us on top. And we are therefore forced to move the boundary every time the machine crosses it.

Machines can calculate, but humans are creative.

Machines can create, but humans have empathy.

Machines can display empathy, but humans possess judgment.

Machines can exercise judgment, but humans make meaning.

Whether a Resister or Retrofitter, the goal is to modify the old definition of human exceptionalism rather than question why human value needs to depend on exceptionalism at all.

The Restarter stops looking for a human moat.

Zero = Suspension

Zero is the point where we stop defending the inherited answer and recover the ability to ask the question again.

With every new model release showing us, “Oh, by the way, AI can now also do that,” zero refuses to rush toward a replacement theory.

To be clear, I don’t want you to hear me saying that humans have no distinctive qualities. I am saying that we should stop beginning with the need to identify and defend them.

That is the restart.

A restart looks past the project of assigning every displaced human activity a new justification. It refuses to search for tasks that machines cannot do so that people can remain necessary.

Instead, it begins by asking something both simpler and more complex.

The Only Question That Matters

Well before AI—and now that AI is on the scene—the organizing question has tended to be something like:

How do we become more productive?

How do we become more efficient?

How do we remain useful?

How do we keep up?

AI and automation show us that this competition is unwinnable.

Which is why people feel so threatened.

I think we should feel liberated.

Once we stop treating ever-increasing [insert metric] as the purpose of human life, the only question left to ask is:

What do you want to do?

And my instinct is that most people will find themselves giving a first answer that is probably still contained within the 1.

Maybe it’s a different job title, a version of the job that feels better, a credential, or a promotion.

But the real question—to borrow some phrases from AI—is not what you are uniquely qualified to do, what only a human can do, what will preserve your economic advantage, or what will make you indispensable.

Just: What do you want to do?

And the incredible—perhaps paradoxical—part of all this is that, in almost every case, AI can actually help make that happen, perhaps even for the first time.

Subscribe now

PS

I Want to Hear From You!

I want to hear from people who are taking—or have taken—the risk of doing what you want to do!

Maybe you left a job. Maybe you built something new inside an institution. Maybe you started a hobby, project, or side business. Maybe you simply did something you had wanted to do for a long time, even though it was risky.

The risk can be large or small. And the thing does not have to have succeeded.

I’m putting together a series of short, punchy interviews about what it looks like to go from 1 to 0: to inspire others to take a chance to do what they want to do!

If that’s you, DM me, I’d love to tell your story.

Also h/t to . I developed a lot of these ideas using her Socratic teaching bot.

I write weekly articles for educators who are ready to get unstuck from outdated curriculum, resistant institutions, and a career that was built for a world that no longer exists.

A Need a New Job

This whole project of From 1 to 0 emerged naturally from all the work I’ve done and experiences I’ve had, and most recently it has been inspired by my entry into posting and writing “in public.”

At the time I started posting in March 2025, I was leading a relatively young instructional design team.

I thought a lot about their futures given AI.

I felt a sense of responsibility for them.

I wanted to make sure they weren’t doing the current work to such an extent that they didn’t have time to up-skill for future work. I wanted to ensure they were ready for whatever came next in their work.

One team member was 28 years old. They were presumably going to work for another 40ish years. There was no world I could imagine that the work they were doing in 2025 would be relevant in 2065, perhaps not even in 2028.

I wondered, what is an instructional designer supposed to do when AI can draft objectives, build rubrics, generate activities, create slides, write discussion prompts, summarize readings, and adapt materials for different students? And quicker, cheaper, and better at that?!

At the time, I sketched a timeline of instructional design.

- Historically teachers did “instructional design” themselves.

- In the 2000s instructional design really began to be popularized and instructional designers did instructional design.

- In 2023 ChatGPT came along and gave instructional designers superpowers.

- Shortly after, apps starting putting the AI-instructional design superpowers into the hands of teachers.

- Soon after that, AI put the power of instructional design directly in the hands of students.

And this is the real disruption of education—removing the intermediary layers between the end user and the thing they’re looking for.

So, I wondered, what’s left for instructional designers?

The Inefficiency of Knowledge

I’ve run these stats before.

In knowledge work, here’s the rundown of how the average person spends their time each week:

We spend
27% on email
23 % searching for the information to to the job
20% in meetings
20% on tasks that don’t actually require subject matter expertise

That leaves just 10%, roughly one morning each week, where what we do actually depends on what we uniquely know.

Instructional design is not much different here. In fact, a colleague and I are developing a half-day symposium at this year’s Quality Matters conference on the shifting work of the instructional designer.

The Real Disruption

And this is primarily why I think AI is so disruptive in the workforce—we’ve just gotten really accustomed to being woefully inefficient.

Or said more generously, a lot of the work we do in our jobs doesn’t actually require us or our unique skills.

I’m not going to pretend to know for sure, but here’s one thought about what the future of work might look like.

I should note that I am probably too hemmed in by what has come before for my vision to create an interesting enough prediction about what is to come.

Nevertheless, here goes.

Companies Won’t Hire People

In knowledge work, I don’t think companies are going to hire people. Well, they’re not going to hire people . . . by themselves.

Instead, they’re going to hire people, who have their own teams of agents, that have mastered certain workflows or can be adapted to them, and that are differentiated by the specific expertise of that person.

So, a person will come into a job with their own agent team, their own workflows, their own ways of seeing and solving problems relevant for the company. They won’t just say, “I know how to do this work.” They’ll say, “Here is the system I’ve built to solve these types of problems, and this is why this solution is different from what you’re seeing from others.”

In other words, companies will hire the people who have the best AI agent teams who have proven they can solve current and future problems at scale.

This could be entirely wrong, but it is still useful to consider, because AI can complete a lot of tasks already, and it is only going to complete more of them everyday.

The value of a person, then, might be to understand the problem space deeply enough to build a system around it and then differentiate their AI solutions team around their unique perspectives, insights, and expertise to give the person and the company a competitive advantage.

This is why AI is such a forcing function—even if that particular concept is wrong, AI makes you ask what your work actually is once the tasks you did are no longer enough to define what you do.

What’s Your Value?

There was a time—and that time was very recently—when training in writing objectives and creating rubrics was an important part of the instructional designer role. It was either essential or at least value add in the educational ecosystem. And therefore, organizations were willing to pay to hire someone for that skill.

But now those tasks are completely automatable. No human required.

So, we might say, those skills aren’t solving any problems that we actually have anymore! There may be some expertise behind them that’s transferable, but it isn’t clear how that applies to the problems of today.

So we need to do that work of identifying what those problems actually are—and if my guess is right, building agent teams around them and our unique expertise.

We at least need to understand the problem space well enough to determine the workflows necessary to solve them and how our expertise shapes that.

That’s a different kind of work.

And I think a lot of this will require radical departures from what exists and even more radical imaginations about what might come.

You might say, it will take going from 1 to 0.

What’s Your From 1 to 0 Story

I’m setting up a series of quick interviews with people who have gone from 1 to 0 in their lives.

This can look like almost anything.

It can mean leaving a steady job and taking a risk. It can mean starting something entirely new at your job or outside it. It could be a hobby where you started again as a totally newbie.

DM if you have a story and lessons from it that others would benefit from.

Subscribe now

I write weekly articles for educators who are ready to get unstuck from outdated curriculum, resistant institutions, and a career that was built for a world that no longer exists.

When the rules stop mattering

Imagine your next health check goes like this:

You step into a shallow pool filled with golden light.

A platform begins lowering you into the water at about two inches per second.

As you descend, your body passes through a ring containing half a million ultrasonic sensors, each roughly the size of a fine grain of sand.

The sensors send sound waves through your body from hundreds of thousands of angles and record what comes back millions of times each second.

In less than a minute, the system produces a detailed three-dimensional map of what is happening inside you.

Then you get out and continue for a full-day spa experience.

My Absurd Images

Behind this radical image of the future of healthcare is Midjourney.

JIC you are unaware, Midjourney is the AI app I use to create the absurd images of myself for my posts, like the image for this article!

Just a few years ago Midjourney became known for one thing: creating images and videos with AI.

Now they’re getting into medical imaging.

The first location for their golden light, ultrasound hybrid spa experience is planned for San Francisco in 2027.

By 2031, the company hopes to have more than 50,000 scanners operating around the world, with enough capacity to perform a billion scans every month.

Here is how Midjourney explained the transition:

“We feel an obligation as people standing on the frontier to look at the foundations of the human experience and ask: ‘What do we want to be different?’ ‘How do we want to be different?’ and ‘What do we want to become?’”

When I officially started this project of From 1 to 0 a few weeks ago, Midjourney’s announcement hadn’t been made and I had no idea they were doing this.

But, I wrote something eerily similar an issue or two back:

Restarters begin with today.

They ask:

Given the problems, tools, and possibilities that exist right now, what should we do?

The current organization, role, process, or business model is irrelevant to that question. Starting with it would be a non sequitur.

More simply: given today, now what?

That is essentially what Midjourney just did.

The Establishment Is Concerned

Medical experts have already started pointing out the limitations.

Ultrasound cannot do everything an MRI can do. Frequent scans could produce false positives. Any diagnostic use will require FDA approval.

All of that may be true.

Midjourney’s response, in effect, is: WTF do I care?!

Instead of trying to resist change or retrofit MRIs with a better experience, Midjourney seems to be wondering:

Why does medical imaging still have to be slow, expensive, unpleasant, and difficult to access?

Why does the experience have to involve lying still inside a machine when it could involve warm water, golden light, and 60 seconds? And why not follow that up with a spa-like experience that rejuvenates your health?

Until reading their vision, I personally hadn’t thought about why healthcare is delivered in contexts that tend to be designed in ways that are the opposite of “life-giving.”

But, resisters there are. So, the medical establishment is evaluating the scanner against the system that already exists.

Midjourney is starting with what is actually possible and something people might actually want.

That is what restarting looks like.

And that is why resisters and retrofitters are running out of time.

Why Doesn’t Education Feel Like Learning?

Maybe it is just me, but before reading Midjourney’s announcement, I had never really considered how little healthcare actually feels like health.

Waiting rooms with uncomfortable chairs. Waiting in exam rooms under fluorescent lights, sitting on crunchy paper . . . waiting. Anxiety, stress, anticipating a diagnosis from someone you see once a year, or have never seen before.

Midjourney says forget all of that and imagines what a healthy relationship with our bodies might actually feel like.

We might ask a similar question: Why doesn’t education feel like learning?

For paid subs, this week I’m offering 7 ideas inspired by Midjourney’s shift into the health space.

Read more