"...higher education will look retro in a bad way."

Part 1 of an interview with Bryan Alexander

About the Interview series

As part of my work in this space, I seek to highlight the folks I’ve been in conversation with or learning from over the last few years as we navigate teaching and learning in the age of AI.

If you have experiences around AI and education, particularly in the higher education space, that you would like to share, consider being interviewed for this Substack. Whether it is a voice of concern, curiosity, or confusion, we need you all to be part of the conversation!

---

Dr. Bryan Alexander is a futurist, author, speaker, and educator whose work focuses on how technology, social change, and long-range trends are reshaping higher education. He is widely known for helping colleges and universities think strategically about the future through research, public writing, and his Future Trends Forum, as well as through books such as Academia Next and Universities on Fire. His most recent book, Peak Higher Ed: How to Survive the Looming Academic Crisis, published by Johns Hopkins University Press in early 2026, examines the mounting structural pressures facing colleges and universities and what institutional leaders may need to do in response.

---

Lance Eaton: Before we get into the book and your work, what does your AI usage and exploration look like? What have you been playing around with?

Bryan Alexander: I use some image generators. My normal habit, when I use images in anything—Substacks, PowerPoint presentations, video—is that I go to Flickr first and look for Creative Commons–licensed images, and that’s where I do the bulk of my image sourcing.

But I find that Nano Banana and Midjourney are great when I can’t find an image, and especially when I want to be creative, and it’s beyond my capacity to create images. I’m an amateur photographer; I can capture things, but often I can’t produce the results I want. I’m not very confident in my Photoshop or Canva abilities, and I also lack the time to really invest in developing those.

Furthermore, I’m fascinated by the results that generative AI produces for images. I find them bewildering, alien, and fascinating. I mean, I’m a lifelong fan of surrealism and experimental arts; for me, that’s one use that I really enjoy. I’m cautious about usage because I know some people are politically opposed to generative AI images, so I try not to put those out on Bluesky.

A second use is for simulation. For me, this is one of the great killer apps of generative AI on the text-bot side: being able to simulate and role-play just about any situation you like. I’ve done that for my gaming class. I’ve done that for simulations of things that I’m about to experience. If I’m going to be giving a talk to a crowd I don’t know, I’ll describe that and have it simulate the situation.

A third use is when I’m working in an area, or for a client, and I want to get the official narrative. Usually, a Google search will give me some sense of that, but I find that tools like Gemini or ChatGPT do a terrific job of giving me an official narrative quickly, so that I can work with it. That doesn’t mean I agree with it—usually the opposite—but it means I can grasp it pretty fast.

Another use is in cooking. I’m a vegan, and I’m experimenting with recipes every week. I also make food for my family, who aren’t vegan. I find that Google searches usually lead me to interesting but unbelievably bulky and hard-to-work-with web pages, especially on a phone, whereas a quick hit on ChatGPT yields a recipe. It’s also plain text that I can grab, paste into a Word document, and then add to, subtract from, and break apart for my own purposes.

I guess the other thing I do is have conversations. In everything I’ve said, I know there are political objections to saying “AI” or “conversations”. I’m saying it this way because everyone knows what I’m talking about, and it’s smoother than saying “I engaged in a simulated dialogue with an autocomplete-on-steroids machine.”

I’ll have these conversations about various issues. If I’m driving, I’ll just put on speech mode and have a conversation with ChatGPT along the way. I know this is offensive to you at a primal level because it means those are minutes I’m not listening to an audiobook.

Eaton: What? This interview is over!

Alexander: What I usually do is podcasts and audiobooks, but sometimes I’m thinking about something, and I want to talk it out with somebody. I’m driving, I can’t really write it down, so I’ll bash an idea back and forth as we’re going over the Alleghenies or whatever.

Eaton: That’s one of the ways I use it, and I really appreciate it. If I’m trying to figure something out or think through an idea, and it can keep asking me questions and pushing back on my points; it’s fantastic. And it is great on drives. I’ve used it while doing chores or going out for a walk. We’re totally aligned on that. I still love my audiobooks, and this is useful alongside them.

Alexander: I’m glad to see that. I was really worried about confessing that.

I’m also really fascinated by NotebookLM. I’ve been using that as a kind of document-repository conversation space. It’s interesting to me that there’s no real competitor to NotebookLM. Nobody else has really offered that functionality.

I find it striking, for example, the book we’re talking about, Peak Higher Ed, last April or so, when I was preparing the final draft, I just uploaded the entire book to NotebookLM. It gave me some great responses. It gave me a 15-minute podcast, a concept map, and two different lesson plans and quizzes. All of this was useful for me because it was being read, and I had a sense of what a reader would find. That was really useful in revising the book and seeing that it got the points I wanted to get across. It also didn’t mention some points that I did want readers to notice.

I thought, all right, I need to emphasize those. That was really, really useful. I mean, you could say I could just ask a person to read it, and I did. That’s true. I had a bunch of readers. But this did it in 90 seconds and gave me yet another view, another pair of eyes, if you will. I found that very, very useful. Absolutely.

Eaton: What level of comfort or acceptance did you have in basically giving the AI model your work to train on?

Alexander: It’s a really good question. Since this was for NotebookLM, I was really focused on the local use, and I was pretty comfortable with it. I’d already worked with my publisher about this. My reasons were, in part, that I think what I’m doing is valuable and I want to get it out there into the training sets, and to do that through some kind of formal process rather than having it pirated. That worked for me.

Eaton: Anywhere that you’re not using AI in your work?

Alexander: I mean, there are two things I do not use it for. I do not use generative AI to produce my writing. I’ll bounce ideas back and forth, but that’s like sitting with you and asking you what ideas we have. And I haven’t used it to produce slides. I will use it to produce some images for a slideshow, but not the core content.

One thing I hope to work on more, but haven’t over the past couple of months, just because of scheduling, is producing music or video. I’m just behind on that. And that’s more for my research purposes.

Eaton: And it’s hard to keep up. That’s something I run into regularly: how to stay competent across different modalities of AI and feel like you’ve played around with them enough to say something meaningful.

That helps lead into this next question. You hinted at it already, but I’d love to hear more about how you see the use of AI tools in future-casting, and how amateurs or the future-curious might use AI toward that end?

Alexander: There are a bunch of ways, and in the futures community, we’ve had a lot of discussions about this.

One way for people who are curious is to use AI tools to do horizon scanning. You might use something like Elicit, or a more classic generative AI like Gemini. But it really comes down to prompting, as always: saying what you want to look for. For example: “Please scan for developments in library science,” and then adding constraints: “Be sure you use diverse sources. Here’s the time window I want you to look at. Please identify trends that emerge.” That’s one approach.

A second is using chatbots to create scenarios. Again, it comes down to the prompt. You might ask, “What happens when Lance becomes the blood-soaked tyrant of Massachusetts?” How does that play out? Then you push it further: “Please tell me the impact on the aviation industry. Please flesh out the impact on microeconomic trends,” and so on.

The thing about scenarios—their glory and their problem—is that they can be very rich stories, and the problem is that humans don’t always have time to make them. With generative AI, you can use these tools to help flesh out scenarios, again including images. For example, in climate change and higher education work, there’s a big focus on solarpunk options. Asking it to imagine a solarpunk campus can yield some fascinating results. If I’m presenting a scenario as a slideshow or video, I can add those images.

A third way is something more interactive: imagining yourself inside a future scenario through role-playing. That’s the simulation feature I mentioned earlier. For example, what is it like if the United States has a post-Trump Truth and Reconciliation Commission, and I’m a citizen of New York City? What kind of world is that? You can flesh that out and walk through it.

Those are a few ways. One challenge, of course, is that generative AI is famously not literally creative and has a hard time coming up with genuinely new ideas. But I think you can prompt it toward that, in part by drawing on science fiction, which is a great source for futures work. Those are some of the approaches.

Eaton: It’s funny you say that. A previous guest, Corrie Bergeron and two other colleagues are publishing a book this year that basically looks at what science fiction says about education, to help think about what it means for AI.

Alexander: That’s a great idea. I’d love to hear what they come up with. You can tell them this if you want: I’ve been, for a few years now, working on a proposal for an anthology on science fiction and education—short stories, with an introduction that also talks about novels. That’s something I’ve been talking with a few publishers about.

You can go all the way back to Mary Shelley’s Frankenstein as one of the first really interesting examples. There’s an interesting lineage of postsecondary education in science fiction. And of course, science fiction has been thinking about education and about “crazy futures” for a long time.

Eaton: I hadn’t thought about Frankenstein as a postsecondary education space, but I see it! Ok, let’s transition into the book. Obviously, I’m dismayed it’s not in audio yet (but is forthcoming!). But I wanted to discuss the chapter on AI. Do you think you could give a very brief synopsis of that chapter and how it ties into the book as a whole?

Alexander: To set the book up: up until that chapter, it’s focused on a scenario that I came up with thirteen years ago called “Peak Higher Education,” which is the title of the book. The hypothesis I came up with back in 2013 was that higher education had gone through a generation-plus of growth dating back to the early 1980s. Growth by every metric you could measure.

That includes the number of campuses, the number of students, the number of employees (faculty and staff), the physical footprint, and the financial footprint. It had just grown and grown. There’s no real historical term for this period. Clay Shirky recommends calling it the “golden age,” and that’s not a bad term. But this was as big a development as the growth of land-grant institutions in the late nineteenth century.

Then we hit a peak—depending on the data—somewhere between 2011 and 2013. After that, again depending on the dataset, for the next seven to nine years total enrollment declined year by year. It started off in very small amounts, like two percent a year, but this was a major reversal, and not many people caught on to it at the time.

What people really noticed was the fall of 2020, with the big drop as a result of COVID, where we saw declines—again depending on the data—of five to six percent in a year. That continued until about, again depending on the data, roughly 2024.

The first third of the book frames all of that out, discusses it, and comes up with reasons why it happened. After this chapter, it tries to imagine how things might play out if we keep sliding down from that peak.

Then I introduce a couple of major forces that intersect with this. One is climate change, which was the subject of my previous book. Another is what I call diverging attitudes about the future and about progress.

In another chapter, I try to imagine: first, can AI help prevent that downhill slide? Is there a way that we can use AI to reinvigorate academia so that we grow out of this? Or, of course, the opposite: would AI worsen this? Would it push us further down the slope?

I come up with four different scenarios for how this could play out. I’ll just mention three of the big ones right now. One is the possibility. This is what people in academia really dread right now: that public attitudes come to see AI as superior to higher education. That is, more and more people accept AI, warts and all. They come to terms with deepfakes, they come to terms with uncertainty, and hallucinations or what Nate Angell and Anna Mills call “mirages.” They’ll come to terms with the politics of it and all of this stuff, and they’ll just use it; not out of enthusiasm, but because it works for them. It helps them with writing, it helps them with learning, and when they’re curious about, say, an early American Quaker who was an abolitionist such as Benjamin Lay, they’ll turn to that.

In contrast, higher education will look way more expensive. It’s important to remember for the reading audience that overwhelmingly, people outside the academy and a lot of people inside it, the conversation is focused on the sticker price of the wealthiest and most expensive schools. Several institutions cracked the $100,000-per-year barrier this year for the first time. When people see that, they freak out and panic.

They don’t know that relatively few people actually pay that amount; that we have what’s called the high-discount model. People don’t know about the discounts. Higher education is lousy at telling that story, and it’s also complicated. I had an epiphany a few years ago where I realized that higher education financing is about as user-friendly and transparent as medical pricing.

People see that and think, “ChatGPT is free,” or “six bucks a month,” or whatever it ends up being. Or “Gemini is part of everything I do.” And higher education will look retro in a bad way. It will look obsolete and irrelevant. In this scenario, AI pushes higher education further down.

You see fewer high school students want to go to college. You see fewer adults want to attend college or university. You see funders who grow impatient with colleges and universities for not advancing AI more quickly. Governments lean on us to do more with AI, including, depending on their politics, calling for us to replace staff with AI, and so on. That’s one scenario.

A second scenario is the reverse. And again, this is a question of public perception and culture, not objective technical reality. The perception flips, and people come to see academia as more reliable and more trustworthy than AI. Hallucinations remain unsolved, regulation aside, and people just get more frustrated with it. They find that asking your local community college to teach you Spanish is more friendly, more humane, and more reliable.

One thing that might drive this is if we have some spectacular disasters that people generally attribute to generative AI: a plane crash, a war, or, worst of all, a celebrity death or public suffering. That could really trigger people to turn on it and sour on it. They’d see it as dangerous or flawed, and if they want knowledge for serious purposes, they’d come back to academia. Academia would look better.

One hint of this—I don’t know if it’s taken off yet—but I saw a few Instagram clips where elementary school kids were using the phrase “AI” instead of “BS.” Like, “Don’t trust Johnny, everything he says is AI.” If that attitude goes wide, things look better for higher education.

A third possibility is that AI keeps plugging along. It doesn’t become trusted, it doesn’t become scary, but it becomes weird. People are curious about it and don’t quite know what to make of it. We see more and more AI in things like self-driving car routes. We see more AI-generated art in the world. We see more AI-generated physical objects—from clothing to home décor to monuments. More and more people are closer to AI. We have more AI friendships, more companionship, more romances. People get married to AIs. The world gets stranger. Again, this is a question of perception, and people feel that this should be strange, and they turn to academics for help. They’ll turn to you for help as an expert and say, “Help us understand this.”

And this is where academics, obviously in computer science, can help, but also academics in media studies, trying to figure out what Hollywood is doing with AI, or whether it’s a threat to media studies. If people are worried about what it’s doing to their minds, psychologists come forward. If they’re thinking about what happens to the labor market, we rely on economists, and so on.

In that world, academics become guides to the world. Now, this is based on a few things. It’s based on academics actually doing that kind of work, especially in public. And this is also based on perception. Those are three of the scenarios that I offer in this chapter.

Eaton: You also have 2 scenarios on AI collapse and therefore not being a threat to higher ed. In one of them, I was struck by this because I keep thinking about points of comparison. You mention that it could go the way of the Metaverse or Second Life. And I wonder whether those are examples of failures to launch versus outright rejection of a technology as your scenario suggests. I wonder about rejection happening to something that hit 100 million users in less than two months, and compare it to the fact that probably neither the Metaverse or Second Life gained such a following so quickly. Does AI feel more like the internet than those other examples? Even if there’s a dot-com bubble burst, the internet (or AI) doesn’t go away.

Alexander: That’s a great question. There are a lot of ways this could go down. As I’ve been saying for about three years now, we still don’t really have a viable business model.

Arguably, one might be emerging in China, because the Chinese digital ecosystem is so different from ours. Among other things, mobile apps for purchasing and mobile apps for social interaction are much more intertwined there. There are already interesting experiments using chatbots to order food or clothing, that kind of thing. That might prove to be a business model.

Here in the U.S., Microsoft, Google, Amazon, and Apple are pouring billions of dollars into this right now. They’re either assuming this is just the cost of doing business or that things will turn around.

OpenAI is just pouring out money like nobody’s business, and they’re betting on an Amazon-style model; that eventually this will turn around. Most recently, they announced that they would allow ads. That’s been tested; it’s not 100 percent effective, but it is one model. Still, they owe so much money right now. OpenAI could just go bust. People have been predicting that for a few years. It hasn’t happened yet; in fact, quite the opposite. But that alone is a real possibility. In education, we don’t like to talk about business models, but this is a key problem.

A second possibility is that the culture just turns against it. People talk about AI hype, although that’s massively overblown, especially over the past couple of years, when the tech press has been so anti-AI. But we’ve also seen anti-AI commentary appearing in all directions, definitely throughout higher education, but also in the wider world. Hollywood is terrified. We have politicians who are against it.

It’s possible that we just sour on this entirely, and there’s plenty of precedent for that. In our lifetimes, you could think about eight-track tapes, partly because they were superseded by audio cassettes. You could go further back, to the nineteenth century, and think about pneumatic tubes, which were an alternative to the telegraph. They were very exciting and very interesting, and then they went bust. They survive now mainly in bank deposits, and they’re fascinating to look at, but they’re not a dominant technology.

We could also see a political turn against AI. We saw a version of that in the first six months of ChatGPT 3.0, when the Italian government just straight-up banned it from the country. Right now, we have a lot of state-level actions trying to corral generative AI in different ways. The European Union has its famous AI Act, which they entertainingly passed without having finished writing it; something I’m always amused by.

What’s interesting in the U.S., and I think only in the U.S., is that this has become really partisan. In the Republican Party, you see a lot of support for generative AI. Project 2025 reflects that. In the Democratic Party, the criticism of AI is largely progressive: criticism of labor practices, criticism of bias, and so on.

And it’s possible that this divide will deepen and play out more and more. That’s something I’m watching closely in academia, since academic trends tend to be very Democratic-leaning. It hasn’t gone all the way yet. You still have a lot of Democrats who are pro-technology, and who see this as a business they don’t want to kill. You also see some high-profile Republicans who’ve come out against AI. I’m fascinated by Josh Hawley, who, “wrote” a book opposing AI from a religious perspective and also from a jobs perspective. I think Marjorie Taylor Greene has also come out against AI, although it’s hard to tell.

It may become a red-state versus blue-state issue. Or it may take a smaller, more benign form. Think about the polite arguments people have over print books versus e-books. E-book users say, “This is easier to carry.” Print-book people say, “I miss the smell and feel of books.” That argument goes back and forth endlessly, and we might see that kind of divide play out with AI.

One thing I’m really looking for, and not seeing yet, is a successor. If we get an AI paradigm that goes beyond generative AI, that’s one thing that could take generative AI down.

But to go back to your question, one thing that really worries me and doesn’t get much attention is the idea that generative AI may finally cramp the web. The web has never been universally popular. It has always faced opposition, sometimes implicit. Look at how companies like Facebook and Instagram try to make sure you never actually use a hyperlink. Look at how many academics still don’t like the web, for various reasons. Our biggest academic technology, the learning management system, is about as anti-web as you can get.

And yet the web keeps growing. Now, with AI-driven search—AI Mode in Google, for example—we may see incoming hits to web pages and web domains go down, which makes them harder to sustain. We could actually see the web start to shrink. We may be passing “peak web” even as we record this, which I think is a terrible and very dangerous thing.

I haven’t seen a really good response to this yet. A couple of big web media properties, like Gawker, have talked about shifting toward face-to-face events instead of web content, essentially giving up on the web.

Eaton: Gosh. Could that be your next book? A Sequel: Peak Web?

Alexander: Yeah. Peak web. Others have talked about trying to create more tailored web content, but I don’t think there’s a solution yet. This is one that really worries me.

The web is built on top of the internet. Take Apple, for example. They’re very much an internet company, but they don’t do a lot with the web. They don’t have a big web presence, they don’t support much web content, and their web browser isn’t great. Or think about tools like iTunes, which are based on the internet but not on the web. Think about podcasting—lots of podcasts have zero or minimal web presence, but people consume them through mobile apps, which completely bypass the web.

Join us for the second part of the interview in a few days!

---

The Update Space

Upcoming Sightings & Shenanigans

• Writing with AI: Making Informed, Defensible Choices Before You Submit. Free Webinar with Paperpal. Wednesday, April 22 at 3pm (ET).

• EDUCAUSE Online Program: Teaching with AI. Virtual. Facilitating sessions: ongoing

Recent Recordings, Resources, & Writings:

• Damm, C., & Eaton, L. (2026, March). From prompt to practice: A framework for transparent GenAI use in higher education. EDUCAUSE Review.

• Eaton, L., Nemeroff, A., & Sun, X. (2026). AI-assisted course design and development. In K. S. Ives, M. Cini, & R. Schroeder (Eds.), AI applications in online higher education administration: Strategies for maximizing returns and improving outcomes. Routledge.

• Margin of Thought with Priten: Season 1, Episode 5: How Can We Center Pedagogy During the AI Tech Wave? (February 2026)

• Online Learning in the Second Half with John Nash and Jason Johnston: EP 39 - The Higher Ed AI Solution: Good Pedagogy (January 2026)

• The Peer Review Podcast with Sarah Bunin Benor and Mira Sucharov: Authentic Assessment: Co-Creating AI Policies with Students (December 2025)

• David Bachman interviewed me on his Substack, Entropy Bonus (November 2025)

• The AI Diatribe Podcast with Jason Low (November): Episode 17: Can Universities Keep Pace With AI?

• The Opposite of Cheating Podcast with Dr. Tricia Bertram Gallant (October 2025): Season 2, Episode 31.

• The Learning Stack Podcast with Thomas Thompson (August 2025). “(i)nnovations, AI, Pirates, and Access”.

• Intentional Teaching Podcast with Derek Bruff (August 2025). Episode 73: Study Hall with Lance Eaton, Michelle D. Miller, and David Nelson.

• Dissertation: Elbow Patches To Eye Patches: A Phenomenographic Study Of Scholarly Practices, Research Literature Access, And Academic Piracy

• AI Syllabi Policy Repository: 200+ policies (always looking for more- submit your AI syllabus policy here)

Finally, if you are doing interesting things with AI in the teaching and learning space, particularly for higher education, consider being interviewed for this Substack or even contributing. Complete this form, and I’ll get back to you soon!

We periodically host small-group workshops and leadership sessions for higher ed teams. You can learn more about our current offerings here.

---

AI+Edu=Simplified by Lance Eaton is licensed under Attribution-ShareAlike 4.0 International