HigherEd Series: What Should Students Learn When AI Can Do the Work? with Nils Gilman of Berggruen Institute

Nils Gilman is Senior Advisor to the Berggruen Institute, an intellectual historian by training and a widely published essayist on the future of higher education and AI's impact on work and learning. His writing has appeared in Noema, The New York Times, Foreign Policy, and many other venues. He holds a B.A., M.A. and Ph.D. from UC Berkeley.

Hosted by Matthew Rascoff
This episode is part of the Edtech Insiders HigherEd Podcast Series, featuring conversations with leaders shaping the future of higher education.

💡 5 Things You’ll Learn in This Episode

1. How AI is disrupting traditional assessment.
2. What students should learn when AI can do the work.
3. Why trust, taste, and goal-setting matter.
4. How the “multiversity” could be reimagined.
5. What AI-enabled, social learning could look like.

✨ Episode Highlights
[00:14:20] AI disrupts the term paper and problem set.
[00:17:41] Trust, taste, and goals as uniquely human capabilities.
[00:21:59] Rethinking the classroom as a social experience.
[00:25:37] The future of the “multiversity.”
[00:47:09] Can personalized education scale?

😎 Stay updated with Edtech Insiders! 

Follow us on our podcast, newsletter & LinkedIn here.

🎉 Presenting Sponsor/s:

Tuck Advisors was founded by entrepreneurs who built and sold their own companies, frustrated by other m and a firms, they created the one they wished they could have hired but couldn't find. One who understands what matters to founders and whose North Star KPI is the percentage of deals closed. If you're thinking of selling your ed tech company or buying one contact Tuck advisors now.

Every year, K-12 districts and higher ed institutions spend over half a trillion dollars—but most sales teams miss the signals. Starbridge tracks early signs like board minutes, budget drafts, and strategic plans, then helps you turn them into personalized outreach—fast. Win the deal before it hits the RFP stage. That’s how top edtech teams stay ahead.

This season of Edtech Insiders is brought to you by Cooley LLP. Cooley is the go-to law firm for education and edtech innovators, offering industry-informed counsel across the 'pre-K to gray' spectrum. With a multidisciplinary approach and a powerful edtech ecosystem, Cooley helps shape the future of education.

Innovation in preK to gray learning is powered by exceptional people. For over 15 years, EdTech companies of all sizes and stages have trusted HireEducation to find the talent that drives impact. When specific skills and experiences are mission-critical, HireEducation is a partner that delivers. Offering permanent, fractional, and executive recruitment, HireEducation knows the go-to-market talent you need. Learn more at HireEdu.com.

Transcript

Copy Transcript Copied

[00:00:00] Alex Sarlin: Innovation in pre-K to grade learning is powered by exceptional people. For over fifteen years, Edtech companies of all sizes and stages have trusted Higher Education to find the talent that drives impact. When specific skills and experiences are mission-critical, Higher Education is a partner that delivers.

Offering permanent, fractional, and executive recruitment, Higher Education knows the go-to-market talent you need. Learn more at hireedu.com. That's H-I-R-E-E-D-U.com. 

[00:00:33] Nils Gilman: My assumption in the piece is that AIs are already really good at doing many things that we expected students to do, like writing papers, like solving problem sets.

And they're also really good at the kinds of things people are going to do in the workforce, whether it's writing marketing materials or coding things for software companies. So the value add people in that equation isn't going to be the ability to write code. It isn't going to be the ability to write marketing copy.

It's going to be something else. And what I tried to do is kind of a first principles assessment of where I see the AI tools currently going based not on their current capabilities, but on their fundamental architectures. There's certain things that they actually just can't do because they're not really meant to do it

[00:01:20] Alex Sarlin: Welcome to Edtech Insiders, the top podcast covering the education technology industry. From funding rounds to impact to AI developments across early childhood, K-12, higher ed, and work, you'll find it all here at Edtech Insiders. 

[00:01:36] Ben Kornell: Remember to subscribe to the pod, check out our newsletter, and also our event calendar.

And to go deeper, check out Edtech Insiders Plus, where you can get premium content, access to our WhatsApp channel, early access to events, and back-channel insights from Alex and Ben. Hope you enjoy today's pod.

[00:02:00] Alex Sarlin: Welcome to the Edtech Insiders Higher Education Series hosted by Matthew Rascoff. Matthew is an education innovation leader, a Stanford lecturer, and a regular guest on our podcast, whose work focuses on expanding access to knowledge and opportunity through digital learning. In this series, he speaks with leaders shaping the future of higher education.

We hope you enjoy the conversations. 

[00:02:25] Matthew Rascoff: Welcome to another episode of Edtech Insiders and our special miniseries focused on higher education innovation. We have a fantastic guest today, Nils Gilman, who is senior advisor at the Berggruen Institute, a former administrator at the University of California at Berkeley, and author of a very new and interesting provocative essay that we're gonna have a chance to talk about today about higher education and AI, originally published in Persuasion, then republished in The Chronicle of Higher Education, that we'll include in the show notes.

Nils, thanks so much for being with us. It's really a pleasure to have you on the podcast. 

[00:03:03] Nils Gilman: Matthew, thank you for having me on. 

[00:03:05] Matthew Rascoff: Maybe we could just get started with a little bit of your own biography, your own formation. I know you're an intellectual historian. Maybe you can give us a little bit of that lens on yourself.

What's your own intellectual history, and how did you come to where you are today, professionally and intellectually? 

[00:03:20] Nils Gilman: It's been a bit of a, a meander. I don't wanna spend too much time talking about myself, but I, I'll, I'll talk about the parts that are relevant. I went to Berkeley for college and for graduate school.

As you said, I, I studied intellectual history, and specifically I studied the history of the social sciences in the United States, above all in the post-war period. So I, I've looked a lot at the way the growth of universities, the differentiation of different disciplines, interdisciplinary work over time.

So originally I was studying that as a historian. I did not go into academia, however. I, I went in, into industry initially, doing stuff in tech and management consulting. Eventually found my way back to Berkeley, where I ended up as the chief of staff to the chancellor there for several years. So I got a sort of inside front seat view of the way in which a big, complicated university like Berkeley operates.

And since I left there nine years ago, I've been working at a research-oriented nonprofit in Los Angeles called the Berggruen Institute, where I've actually been building up a series of research programs on areas like the future of capitalism, transformations of the human, the planetary as a successor concept to globalization, and so on.

So the bottom line on all this is I've done a bunch of different things in my career, but the through line has been that I've been really interested in the way in which technology and social thought interact with each other and how they get institutionalized in various forms, whether that's universities or think tanks or what have you.

[00:04:43] Matthew Rascoff: The piece we're gonna talk a little bit about today, I would say is unafraid of technology. I mean, it's aware of it, but, you know, I think for a historian, it's certainly a more sophisticated view of technology. Like, how did you come to that? Was that, like, the Bay Area ecosystem? Is that going to college at Berkeley?

Like, what was your technological formation, as it were? 

[00:05:04] Nils Gilman: Honestly, this was one of the many pieces of fortuitous luck I've had in my career. I, I finished my doctorate in early 2000, and that was the very height of the dot-com boom, the first one back in the late '90s. And it, uh, you know, eventually the bubble burst eventually a year later.

But, you know, in, in early 2000, any fool could get a job in Silicon Valley, and I just happened to be in the right s- Mm-hmm, mm-hmm ... place at the right time, and I got a job at a enterprise software. It was not a dot-com firm, it was a, it was a, a real technology firm. And I spent half a dozen years in the early aughts, first half of the aughts in tech firms.

Ultimately ended up my last tech, formal tech job was working at what was then a pretty new company called salesforce.com. And through that process I got to know- Wow ... I worked closely with engineers. I was in engineering organizations. I'm not an engineer myself, obviously, but I worked really closely with engineers and learned a lot about technology.

And ever since then, you know, a lot of my practice since I left tech has been about thinking about how technologies emerge, how they get adopted, the kind of social and political impacts that they have, and doing that both in a register of h- you know, sort of historical perspective, but then also obviously with an attentiveness to what's happening now and what's likely to happen in the future with these technologies.

So I've stayed pretty closely attached to these questions. And specifically with respect to AI, it's something I've been thinking about really since I was at Berkeley working in the chancellor's office. We were starting a program, what eventually now has become a college of data science at Berkeley. And so I was working really closely with a lot of the people in the computer science faculty then.

This was back in 2014 and '15, and that's really where I started to get, I guess, indoctrinated you might say, in, into these kind of cutting edge technologies, and I've kept up with the field as much as I can over the years since then. 

[00:06:46] Matthew Rascoff: I think part of what was so refreshing about your essay, and maybe we can pivot to that, is that it preserves your historical framework, obviously, and it starts with this story from Clark Kerr, and maybe you can tell us that story.

But it's a futurist piece. It's sort of about where we're headed. And it does seem like that is part of the discourse divide, that the technologists are just so historically uninformed about anything before, like their own formation, their own education 18 months ago, whatever, however they do their periodization.

And historians tend to be so anti-futuristic. Like they think that's, you know, presentism is the worst accusation you could make as a historian. Futurism is probably even worse, is it not? And like, and I think part of what you're doing here is bringing together these two discourses that are really unable to communicate, at least in the Bay Area circles that I travel in.

[00:07:35] Nils Gilman: Yeah, I think that that's right. I definitely think that there are some people, my former colleagues, intellectual historians, who I'm sure look askance at what I've done with my education, which is I sometimes describe as turning the telescope around and instead of looking backwards, trying to use the same tools- 

[00:07:50] Matthew Rascoff: Exactly

[00:07:51] Nils Gilman: to look forward, right? 

[00:07:52] Matthew Rascoff: Right. That's what I see you doing in this piece. Yeah. 

[00:07:55] Nils Gilman: A lot of the work that I do when I try to think about the future and prognosticate or diagnose where we might be headed is in fact using kind of methodologies that I originally learned as a historian. And let me be specific about what I mean by that.

So, you know, when you're a historian, you're looking at some episode, right? First World War. Why did the First World War break out? What was the course of it, and why did it end up the way it did? Now, as a historian, you know how things turned out, and it's impossible really to, when you're looking backwards in time, to not be aware that, you know, eventually France and Britain, the United States would win, the continental powers would lose, and you can't write a history of First World War without that awareness.

However, one of the dogmas you have as a historian is that the people who were involved at the time didn't know how it was gonna turn out at the time. And obviously, when we look at our own present day, we don't know how things are gonna turn out. However, when you look backward in time, you can see that there were all these structural reasons why in the end it probably was fairly, if not inevitable, very likely that a combination, certainly once the United States got into World War I, that the Entente Powers were likely to win.

And we know that because of industrial power, because of manpower, et cetera, et cetera. So I sometimes think of when I try to think about the future in the same way, like what are the structural elements that are present that we can diagnose right now about our present world? And then what are the ways in which they might interact in complicated ways to produce a variety of possible futures?

It's not that the future is deterministic, but that we do have some structural patterns. There are some things that are likelier and some things that are less likely. There's always contingency. There's always agency. Those things are important to acknowledge. But I think if you do the structural analysis of a situation, then you can begin to get some ideas about where things might be going.

So when we think about that with respect to something like AI, you know, you gotta talk to people who are in the industries that are building these foundational models, that are deploying these technologies to build new companies or deploying them inside of existing organizations to transform them, and you start to try to look for patterns about the way that's going.

What things are successful? You know, what's the rate of change? You know, what are the reasons why, you know, this rate of change might change in the future, either slow down or speed up? And you always gotta, of course, take it with a grain of salt how the technologists at these companies, they have every incentive to make everybody believe that these things are gonna transform the world in radical ways, that the technology's gonna improve recursively forever.

Eh, that may be true. Might not be true, and you should always be skeptical about what people are saying when they're talking their own book. 

[00:10:20] Matthew Rascoff: Well, let's talk about the higher education book and kind of point that telescope a little bit at universities, which is, you know, the theme of this podcast. You start your piece, maybe you can just tell us about the piece and like how you came to write it, because it started in Persuasion, then it was republished in The Chronicle.

Maybe you can just give us the three-minute summary of the piece. I have many questions about it. But this opening with Clark Kerr, obviously, who described only the apparent incoherence of higher education, but in fact, you know, he viewed the multiversity, his model, as the institutional master key of mid-century American civilization.

So, like, give us this framing of the piece and its origins, and maybe you can just summarize it briefly for us. 

[00:11:02] Nils Gilman: Yeah. I mean, so Clark Kerr, first of all, let's just level set who Clark Kerr was. Clark Kerr was, I think probably, it's not very debatable, I think he was the most important education administrator of the 20th century.

He was the president of the University of California and then pioneered what became known as the Master Plan for Higher Education in California in the mid 1950s, which massively expanded the University of California system. You know, many, many more people were going to college as a percentage.

Population of California was growing dramatically. There had been two universities of California before, the Berkeley and UCLA, and under his guidance, the system would expand to eventually the 10 university system that it is now. And he had a very specific vision for what the university ought to be. He used, he coined this term in some lectures he gave in 1963 of what you just said, the multiversity, the idea that a university should be many, many different things.

And he, the reason how he had this vision was that he saw the university as the key to helping societies modernize and develop and economically develop. And he had this vision for a stratified education system, so you would have the elite UC systems, you would have mass education provided by the California state systems, and you would have open access through the California community college system.

And this would create, you know, highly trained workers, it would create original research that would power new industries, and also create workers who were trained who could go work in all those industries, right? So he had kind of a master idea that this complex conglomerate of different functions would actually be the thing that would produce a modernizing, economically growing California and United States.

So that was kind of the model he had. And it's really important to note that like when he was coming up with this in the 1950s, the general model for complex organizations was conglomerates, right? You had businesses that had many, many different lines of business, many, many different functions. That was the, the kind of dominant way of thinking about the way in which big businesses should be organized.

So it was natural enough to think that a big business like the University of California should also be organized as a conglomerate with lots of different functions. Now, conglomerates, of course, went out of fashion in the corporate world back in the 1970s and '80s, but it's still the form in which the University of California and most universities, most big research universities are organized.

And that model has gotten more and more, I guess you could say threadbare over the last 50, 60 years, and people have been cr- criticizing it for, for decades, but it's held together for now. So that's sort of the conjuncture we were in, and now AI has arrived, right? And, you know, with the arrival of ChatGPT 3.5 in particular, now almost four years ago, I believe that the model is gonna be completely blown up, and universities are gonna have to reinvent everything that they do as a result.

And so what I tried to do in this piece is think about, you know, what does that reinvention have to entail, and how do or- universities potentially need to be reorganized or new institutions of higher learning to emerge that can both take advantage of the new technologies and also won't be exposed and vulnerable to these new technologies?

[00:14:04] Matthew Rascoff: And just to put a very specific point on it, what is it about AI that you think is existential in ways that, let's say, MOOCs, the internet, and previous technological revolutions were not because institutions were able to absorb those and were not threatened by them the same way? 

[00:14:20] Nils Gilman: They're not threatened by them the same way.

I mean, there's been resistance to those things also at many universities, it's worth noting. But it's not nearly as an existential thing. I mean, look, the fundamental, the core of pedagogy in higher education has been the term paper and the problem set. That's basically the way in which students were expected to produce content and to be evaluated in their performance.

And the result of your ability to write papers or do problem sets, take exams also, which replicated the idea of writing, you know, writing and/or solving problems, that was meant to test and enhance people's cognitive abilities in whatever the field that they were studying. So you know, if you're wanting to be studying history, you become a better historian by reading a lot of sources and then synthesizing them and writing a paper.

Or if you're studying physics, the way you get better at physics is by working through these problem sets until you become familiar enough with them so that you can recognize the same problems and solve them more and more rapidly and more and more efficiently and more and more correctly over time, right?

The issue with AI is it completely blows that apart. Now, if you get a problem set or you get a paper assignment, all you gotta do is feed the problem set into the AI or feed the paper topic into the AI, and it'll spit out an answer which is at least as good as what most undergraduates are able to do.

And so there's no actual putting the students through their paces and training them. There's also no way to evaluate whether they're actually learning anything with these mechanisms. So it's a much more direct assault on the entire pedagogic process that's been used for, for decades inside of universities.

[00:15:52] Matthew Rascoff: It sounds like your view of the pedagogic process is that assessment is at the center of it. That's where the trust in these institutions comes from. That's the basis of the value of the credentials, I think you called it the currency, you know, that they trade in. 

[00:16:06] Nils Gilman: Yeah. 

[00:16:06] Matthew Rascoff: And if we can't trust assessment, then we can't trust all these other things that are downstream of assessment.

Is that right? 

[00:16:11] Nils Gilman: Yeah, that's right. I mean, look, let's, let's be very clear. Credentials have always been a loose proxy for capability. You know, if you graduate with a degree in, you know, physics and you have a three point eight GPA, the assumption is that you were better at physics than a person who had a three point three GPA, assuming they went to the same university.

Now, that's always been a bit of a fudge, right? 'Cause it depends on exactly what classes you take. Some disciplines have more grade inflation than others. Different universities will have different quality teachers. But you know, there is a hierarchy of universities, and then there are the grades were also supposed to be a hierarchy.

And as a result of those things, they were a proxy that employers used, among others, among others, to evaluate whether these people were hireable and what kind of skills that they had. Now all of that has kind of been blown up, and we're gonna need to come up with really different ways both to actually put people through their cognitive paces, put students through their cognitive paces, and also for evaluating whether they're actually learning from that process, right?

And that can then become a different way of, you know, the way they would present themselves to potential employers and present their skill sets to potential employers. 

[00:17:15] Matthew Rascoff: My read of your essay, though, is that it's not just about new assessments on the cognitive goals. We also just need a set of new goals that are organized around the things that AI cannot do.

And, you know, I read them as trust, taste, and the constitution of goals. So is it the former? Like, is it just like we need a better assessment system that, that can't be hacked the same way? Or do we really need to rethink the outcomes of learning and the purpose of learning? 

[00:17:41] Nils Gilman: I think it's both. So look, one of the things I say in the essay, which is not uncontroversial, I, I should say, and, and to be totally candid, I'm not even 100% certain that I'm right about this, but my assumption in the piece is that AIs are already really good at doing many things that we expected students to do, like writing papers, like solving problem sets, and they're also really good at the kinds of things people are gonna do in the workforce, whether it's writing marketing materials or coding things for software companies, right?

So the value add people in that equation isn't going to be the ability to write code, right? It isn't gonna be the ability to write marketing copy. It's gonna be something else. And what I tried to do is kind of a first principles assessment of where I see the AI tools currently going based not on their current capabilities, but on their fundamental architectures, right?

There's certain things that they actually just can't do because they're not really meant to do it. They're not being designed to do it. It's kinda like there's cer- some records in sports that are never gonna be broken because people aren't even trying to break them anymore. Nobody's trying to throw 40 complete games in a, in a baseball season anymore, so that record's never gonna be broken, right?

So in the same way, there's certain things that at least the current architectures of AIs aren't even trying to do, and some of those things, I think, and you just named three of them. One is build trust between people, right? The ability to evaluate another person and establish a basis for trusting people.

The AIs are not trying to do that, right? That's not their, what their architectures are designed to do. They're also... You know, that's an essentially social and intra-human thing, trust, right? Likewise, taste. You know, what is taste, right? I mean, this is something that a lot of people are saying that the AIs aren't really able to do.

I'm not sure that they're never gonna be able to do this one, but I doubt it because what is taste actually? Taste refers to the ability to have a distinct and high quality or the perception that somebody... If you say somebody has good taste in, say, home furnishings, right? It's that they have similar taste to you and that they have elaborated it in some kind of original way.

If you, you know, are a natty dresser but you just get all of your clothes from one designer, that doesn't mean you have good taste. It just means you have money and you can buy Loro Piano or whatever, right? That's not good taste. Good taste is the ability to put things together in a kind of original way which other members of your epistemic community understand to be original to you, and that's what makes somebody have a good taste in the eyes of others, is that they are within a, a certain kind of aesthetic community and they've elaborated something that's specific to them.

Now, I think an AI could help you with that potentially, right? You could brainstorm. You could say, "Hey, look, here's some pictures of outfits I've got. Here's a piece I've already got. What are some things that I might buy to complement those things?" So I think that there's ways in which it can help you with that, but it's not actually giving you the taste, right?

Because at the end of the day, you're gonna have to make a decision. This is the particular piece. It can help potentially find things online to show you that you maybe want this, maybe you want that. That's brainstorming with you about things, right? AIs are really good for that kind of thing, but they're not actually gonna give you the taste.

Ultimately, that's gotta be something that you do for yourself. And third, goal constitution I think is another one, right? At the end of the day, what you want to do is not something you can outsource to somebody else, right? Like, why do you even wanna do X as opposed to Y? Well, at, at the end of the day, you've gotta make an internal assessment that X is more valuable to you than Y, and you've gotta figure that out on your own.

Like, you know, your boss can tell you what to do, but then your boss, you know, that's a, that's a regression problem, right? At the end of the day, somebody's gotta figure out what this organization is about, what you wanna do, and that's also something that at the end of the day you have to assess for yourself.

Is this goal that I constituted for myself still the same one that I want as I get further and further into the process of trying to achieve that goal? These are things that are essentially things that are about you internally, and it's not something that you can outsource to a machine, and the machines aren't even trying to do those things, right?

So the architects aren't trying to build machines that will give you taste or that will constitute your goals or that will build a trust relationship between you and your spouse or you and your business partners. 

[00:21:39] Matthew Rascoff: One thing I appreciated about your essay is that you're willing to take on the organizational and structural implications of these ideas.

So walk us through, like, what is the new design of a learning environment or a learning institution that is built around those goals and is a model of assessment that you think is defensible and trustable for the long term? 

[00:21:59] Nils Gilman: I'm gonna get back at the end to whether this scales or not because I'm not sure it does.

But when I, I, I tried to reason from first principles and I- when I was writing this essay and I said, "Okay, so like if what we really wanted to do was train people in those things like trust building and taste and goal constitution, really what you need to do is have an inverted classroom in ver- in a, in a lot of ways.

What you need to do is have the classroom be a social experience. It's not just a professor standing in front of the room and lecturing, right? You can get a lecture on any top- I mean, already n- before now you could go to YouTube and get human beings giving you lectures. But now you can get a bespoke lecture from Claude on almost any topic you want, and it's gonna be about as good as what you might get from a lot of professors, right?

So that's not an efficient use of a professor's time, and it's not a good way to, you know, have the students, you know, get information. They should be, you know, there should be as- assessment of your personal learning goals and that you should get specific custom information fed to you and th- this is something that the AIs absolutely can do.

What they can't do is have you sit next to your classmates and have a social experience where you're debating about is this interpretation better? Is that interpretation better? Does this particular image look good? Does this particular image look, look less good? You know, I'll just give an example of one that I heard a design professor tell me about She gives students an assignment and has them use Midjourney to create images.

You know, for example, create an image of a futuristic city, and then the students will all sit there and play with Midjourney for five or 10 minutes, create images. She then will project all the images in front of the room, and then she facilitates a conversation among the students about which of these images they like best, right?

Why do they like image A versus image B? What are the relative merits of it? So now you're having an actual conversation among the students where they're assessing the relative, you know, aesthetic or technical or whatever merits of these different images. That's something, again, you're now having a back and forth.

The students are beginning to learn their own taste. They're not necessarily gonna agree which is the best image, right? But you'll be able to, like, have to defend your position, listen to others carefully. So you'll learn all of those kinds of, if you wanna call them soft skills, that are really important.

And when you think about, like, what goes on in most workplaces, a lot of the work actually is that. It's about persuading your colleagues that you should do X or Y, right? So I think there are all sorts of modalities like that, but that are extremely labor-intensive. And this is the point that I think is, is really crucial.

One of the things I say in the essay is, you know, the university system that I think comes closest to this model today, there's so many universities, maybe there's better ones out there, but, like, the one that comes closest to it today that I, that I know about is the kind of Oxbridge tutorial model that Oxford and Cambridge have, where you work really-- If you're a student there, you work very closely, directly one-on-one with a tutor.

Maybe there's two people, and you are working through content with them continuously, and they're giving you ongoing feedback, you know, a couple times a week. That's super labor-intensive, right? But it's actually, these are extremely high-quality learning outcomes because you're getting intensively worked.

It's like having a personal trainer as opposed to just being in a spin class with 25 other people. You're gonna get much more bespoke custom fit training that's gonna be designed and fit for your particular needs and your particular interests. Obviously, that's super labor-intensive. And one of the kind of paradoxical things that I conclude in the essay is, heck, this might actually be the solution to the academic job market crisis, right?

Like, we're gonna need to hire way more teachers. If we go with this model, we're gonna need way more people providing that kind of educational training to the students who are going to the university. And it may not even be a university, right? It may be some other kind of learning institution. I know that some of the people in the audience here are working at various kinds of Edtech companies or other kinds of learning platforms.

And it may be that these kinds of ongoing learning things are not necessarily gonna take place through the traditional institution of universities. 

[00:25:37] Matthew Rascoff: Well, let's talk a little bit about that because you predicted disaggregation of those institutions and it coming a part of the multiversity into research centers, coming-of-age residential communities, and then kind of separate trustable teaching systems.

So how does that play out? Like, are we gonna go to college or not? Like, what, what happens to Berkeley, let's say, in that future? And I wonder if you might respond to Ian Bogost, who wrote this piece in the spring about the accidental winners of the war on higher ed. Who are they? The liberal arts colleges.

[00:26:14] Nils Gilman: Yeah. 

[00:26:14] Matthew Rascoff: Who were, you know, the original opponents of the research university in the 19th century, who may, he thinks, you know, be the beneficiaries of the federal assault on research universities because they don't take that much money from the feds, and they don't have grad students, so they don't need to worry about dependency on federal funding.

So do you think that's the future, and how does this play out in terms of, like, the actual structure of, like, what we go to and when? 

[00:26:39] Nils Gilman: Yeah. So I, I read that Bogost piece, and I definitely-- I mean, it, it was sort of a, a slightly perverse piece, but I basically agreed with it. So a couple of things. First of all, like, what is it that's gonna be disaggregated?

It's-- When Clark Kerr talked about a multiversity, I mean, he had a, he had a joke that, you know, the three great problems for university administrators are football for the alumnis, sex for the students, and parking for the faculty. And I know this is still kind of the case, right? In the sense that, you know, when I worked as the chief of staff to the chancellor at Berkeley, y- it's just an unbelievable array of things that, you know, you have aggregated within, you know, a complex university like Berkeley.

You have museums, you have libraries, you have residential facilities, you have dining facilities, you have sports teams, you have obviously all the educational things. You have classrooms, you have laboratories, you have research facilities, you have all these different departments. Like, it's just a vastly complex system and array of different things, not all of which necessarily-- There's always been tensions between them.

Now, some people have argued that those tensions are productive or useful. That's a whole de- other debate we can get into. But certainly it's an extremely complex structure, and it's a very expensive structure to hold together, and it's a very hard structure to manage because anytime you have that much complexity, it's hard to focus on core competencies, right?

The term core competencies emerged in the 1970s and '80s in management literature specifically to criticize the conglomerate form in corporations. The idea that, you know, if you're a corporation who's working in 20 different industries, how are you gonna be able to be the best at any of them, right?

Instead, you know, you'd wanna break these things up and focus on individual things. So I think that when you think about universities or colleges, let's just say initially, what is it that they're doing? So there's a couple of different things. I mean, the first thing I wanna say is the vast majority of people in the United States, and certainly abroad, who go to college don't really go to college.

They usually s- live at home, and they go down the street to the university that's nearby them, right? They don't move into a dormitory or a fraternity. That's actually a relatively unusual experience, right? Most people are commuter colleges and so on. But in the image, in the mind's eye, most people think of going to college importantly as a milestone in one's maturation into adulthood, right?

And so a big part of what universities do that have residential facilities is they, you know, run dormitories, residential facilities overseeing, you know, 18 to 22-year-olds who are going from, you know, late adolescence into early adulthood basically during those years. And so there's a whole coming-of-age process.

But there's no reason necessarily why that has to be the case. And in fact, most colleges, as we just said, don't do that, right? That's, that's a function you don't have to have. And the process of coming of age and moving out of the house is managed by lots of other people in lots of other ways. Most people don't go to college, even in the US.

In other countries, almost nobody has a residential college experience. And yet somehow they eventually move out of the house and they get their own homes and so on. And insofar as they don't, it's not because they didn't go to college, it's because housing has become too expensive, right? Which is a different problem.

So I think you can break these things apart, right? And you don't necessarily have to have the curricular aspects of things institutionally co-located with the residential aspect of things, the residential coming-of-age aspect of things, right? So you certainly don't need to have it co-located with minor league sports franchises as, you know, most American big universities do also, right?

I mean, these are, these are weird things that we put together for historical path-dependent reasons. If you were starting from scratch and you just wanted to provide the best possible education to people, you certainly wouldn't adopt the model of a multiversity today, right? And so part of what I try to do is say, you know, if we're gonna try to create the minimum viable product for a higher education institution that actually did do these two things at once, it was both a coming-of-age experience and a curricular training experience in light of AI, how would you organize that, right?

And, you know, I did a little bit of penciling back of the envelope calculation and, you know, I imagined, uh, you know, a university that had, you know, and I, I say this in the article, imagine sort of Deep Springs on, you know, Deep Springs, which is a small college in the White Mountains in California. It only teaches people for two years.

They get an associate's degree, and then they transfer to some higher education place. There's only 25 students a year who go there, but the alumni, the alumni network, if you go to Wikipedia, is just incredible. And, you know, given how small of a, of a college this is. And they have a very intense experience.

They create... The students create their own curriculum. There are teachers who come in and guide them, but it's a co-created experience, right? By the students and the teachers who come in usually for one semester or one year at a time to teach them. And so I think that something like that could work, where you, you know, you live on campus You, or you li-- you know, you basically you would have to buy a building.

There could be, you know, dining hall on the ground floor, there could be classrooms on floors two and three, and there could be residences for the students on floors four, five, and six, right? And that would be about half the cost, right, is actually providing the b- building the physical plant. And the other half would go to having a group of teachers who are basically there facilitating the kinds of conversations we were discussing earlier.

And I think a model like that could be really attractive, and it would actually be significantly cheaper than attending us. I mean, I don't know what the sticker price is now for the high-end schools. Bennington is usually the most expensive college. Is it over $100,000 now? If not, it probably will be pretty soon.

[00:31:56] Matthew Rascoff: Yeah. So what happens to research then? Because, you know, the premise of the research university is that research and teaching were commingled for a reason, and that there was some synergy between them, and, and that there was a core competency around the combination of those two things in a hybrid. It sounds like you've basically given up on, on that hybridity completely in this model.

D-does research just move to independent Arc Institute, that CZI, that sort of thing? And what are the implications of that for the structure? 

[00:32:26] Nils Gilman: You know, look, again, there's three different functions from a ideational perspective that universities do, right? They teach undergraduates, they teach graduate students, and they do research.

They perform research. You know, in a lot of countries, the research Functions are largely happening in other kinds of organizations. In France, they have something called the CNRS, which is a network of researchers. Some of them also teach at universities, but a lot of them don't. They just do pure research.

You have in Germany, you have the Max Planck Institutes. Obviously, there's been a huge growth all over, you know, in the United States. You have some of the most active research labs are happening in corporate settings, right? And that's been true for a long time. I mean, one of the most innovative research centers in the post-war period, two of them were not universities, right?

One of them was Bell Labs, the other was the RAND Corporation. So it is possible to imagine research happening in non-university settings. It would require a really different model, but you know, it wouldn't be that hard actually for the Nas- National Science Foundation and the National Institutes of Health to just start funding independent laboratories, right?

Instead of having them, having them happen to be at universities. Again, there's all sorts of structural reasons why they most, you know, the, the lion's share of that research funding does go to those institutions. But there's no, you know, theoretical reason why it couldn't happen in a different way through a different kind of institutional edifice.

[00:33:46] Matthew Rascoff: And there is a move in Washington, they're calling it X Labs. It's an effort to move research out of higher education institutions into standalone laboratories that are independent. 

[00:33:56] Nils Gilman: Exactly. Now- 

[00:33:57] Matthew Rascoff: So that sounds like you're supportive of that. 

[00:33:59] Nils Gilman: I'm not necessarily supportive of that, uh, just in its current form because I think the things that are motivating that in Washington are, are kind of nefarious.

But in theory, it's not a bad thing. I mean, as so often with things that Trump proposes, they're often responding to something that actually is dysfunctional, but then they're executed in ways that are not necessarily very, very smart or effective. So I'm not sure that the current leadership that's promoting those ideas is necessarily gonna b- do a good job of it.

I think they regard-- I mean, JD Vance has said explicitly that, you know, universities are the enemy, so this is a way of performing a cachectomy on a cultural adversary. So I think that's what's really motivating them right now. With that said, I don't think that that necessarily has to mean that science will be destroyed by this process, and if we go down that path and we get people who actually believe in science running it, then it might actually be an effective do.

Now, with that said, I wanna go back to something else, Jed, because I think it is important to acknowledge this point about synergies. And I'm gonna tell a story that I don't think many people know about Clark Kerr and his master plan, which is that when Clark Kerr proposed the master plan for higher education in California, as I said earlier, he had this idea of three tiers.

There would be the super elites, which would be the UCs, general education happening at the Cal State system, and then mass open access education through the community college system. Now, originally, what he actually proposed in, in an early draft of the master plan was that the UCs would only be for upper division and graduate students.

They-- Everybody would go to a CSU or a community college for the first two years And take their general education classes 

[00:35:35] Matthew Rascoff: Like your core competency idea. That would be their focus. 

[00:35:38] Nils Gilman: That would be their focus. And the logic here that Clark Kerr said is, "Why should we have these professors who are basically being recruited for, you know, their research prowess teaching, you know, freshman seminars or even big freshman lecture classes, right?

Where there's a, a thousand kids." That's not a good use of Jennifer Doudna's time. Jennifer Doudna, you know, the Nobel Prize winning biologist at Berkeley, she often teaches Bio One. Why does she teach Bio One? Is it because she's the best teacher of Bio One? Is it because only she can, like, teach undergraduates basic biology?

No, it's because she's a superstar, and it's a way of attracting students to the major, right? So that's sort of a, not a particularly great use of her Nobel Prize winning brain and time. But, you know, it makes sense within the internal logic of the system. 

[00:36:18] Matthew Rascoff: So you wanna go back to that, the original model- Right

the, the first master plan. 

[00:36:23] Nils Gilman: The funny thing about this is, Matthew, do you know why this plan was defeated? Can you guess? 

[00:36:27] Matthew Rascoff: I can guess based on other models because I know Daniel Coit Gilman wanted to do this at Johns Hopkins too, but the worthies of Baltimore persuaded him that he needed to train undergraduates, and if they were gonna support him philanthropically, then he needed to enroll their sons, and they weren't gonna get PhDs, and they basically forced him to do a bachelor's plus PhD program.

I would guess something like that happened in California too. 

[00:36:50] Nils Gilman: It's even dumber than that, I'm afraid, Matthew. The reason is the football boosters at UCLA and Berkeley said, "We can't have a football team if we don't have a four-year undergraduate program And at the time, by the way, in the late 1950s, Berkeley had a national championship winning-- won several national championships in the football arena, so they had a lot of sway back then.

But, you know, the logic here was, okay, if you've identified this aristocracy of talent in their first two years going to community colleges or Cal States, then it makes sense to bring them for upper division purposes. They can sit at the feet of graduate students and so on. So I think there is for sure, if you're trying to train people eventually to become the master scientists of the next generation, then, you know, getting them exposed when they're, you know, pretty wet behind the ears is a good idea.

Not necessarily as freshmen, but I think there is a loss, and I, I can testify this to my own experience, right? My first y- year and a half at Berkeley, I took general ed classes, and some of them were from big name professors, but I barely ever talked to them because they were huge classes. And it was only actually in the spring of my sophomore year that I took an upper division class, and it was my first intellectual history class with Martin Jay, and it totally seduced me.

I thought I was gonna be an econ major, but as soon as I took his class and I saw what his brain could do, I was like, "I wanna do that," right? And I knew almost from the get-go that this was what I wanted to do professionally, and I wanted to, you know, went, went on and eventually got a PhD, worked with him as grad- in graduate school as well.

So I think some exposure of undergraduates to big name professors, research, and so on, I think is valuable, but primarily because you're gonna wanna recruit them into graduate school, right? So the question is, is that the only way you can expose people to stuff like that? I think the answer is clearly no.

You know, you could imagine a situation where in your last year, I'm, you know, I'm imagining my, you know, my one building, minimum viable product, AI-enabled new style college, right? You could have a program where you go and, you know, the particularly precocious students will get a chance to take a graduate class at the University of California that's just down the street or what have you.

So you can imagine programs that would allow students who want to go down that path towards graduate research-focused education, getting exposed to that in the last year or two that they're at the university, right? But the truth is that the vast majority of undergraduates, they say they're doing research, but it's kind of play research.

It's not really original research. How many times do undergraduates write a paper that goes on to change the world in some way? I mean, it does happen from time to time, but it's a pretty rare phenomenon. I can only really think of one offhand. Ronald Coase won a Nobel Prize in economics for where the primary citation was for the thing he wrote as a senior in college at the University of Chicago 60 years earlier.

But you know, that kind of thing is super rare. Basically, you're getting trained in the process as an undergraduate. You're not actually doing core original research. That really only happens when you get to graduate school. 

[00:39:35] Matthew Rascoff: What about the undergrads who are solving unsolved problems in math, like open Erdos problems with a single ChatGPT prompt?

[00:39:43] Nils Gilman: What's really interesting about that is those people, first of all, they weren't even students, and second of all, they weren't even mathematicians. They were just people who work at OpenAI who pointed the tool at these kinds of things. You know, look, it's, it's a crisis for mathematics in particular, I think is undergoing, I don't think there's anybody would dispute this, the biggest revolution in the history of mathematical research is happening right now as a result of these things.

And what the mathematics business, if you wanna call it that, the profession and what mathematics research is gonna look like going forward, I think is anyone's guess. I personally think this is, would be a very exciting time to be in math because of that, but certainly it's also a, a super big disruption.

[00:40:20] Matthew Rascoff: Before I let you go, I wanted to ask a little bit about your review of Chad Wellmon's new book, After the University. You gave it a very positive review in your Substack, and we'll link to that in the show notes. But it sort of names this human capital theory as the bête noire and as the source of the divestment from higher education.

Can you just help us understand maybe in brief the argument of the book and, and why you thought it got it right? 

[00:40:45] Nils Gilman: Yeah. So Chad Wellmon is a intellectual historian at the University of Virginia. And this is his new book. It just came out, I think six weeks ago or something like that, two months ago maybe.

It's called After the University, and I, I can't recommend it warmly enough. It's beautifully written. You know, and he, he writes in a sort of elegiacal register about what he feels the university has lost. And what he, he describes that as disciplined learning, disciplined study, right? That, that, that core value is what universities seem to have lost as they become very expensive places and people, undergrad-- students and parents who are paying for them are very, you know, understandably, if you're gonna spend hundreds of thousands of, of dollars on going to the university, you get to be very focused on the instrumental value of your degree.

Is this actually going to get you to where you wanna go professionally? And Wellmon feels like that's a real loss. You know, the things that originally motivated people to go to universities back in the nineteenth and up through the middle of the twentieth century before we started to have these escalating cost spirals, was that people wanted to actually learn and study, and study in a disciplined way, right?

Then he asks, "Why did we lose that?" And the culprit, and I think, you know, I hadn't heard this theory before. I think it's pretty original, and I think it's, uh, he-- I found it extremely convincing, is the rise of human capital theory. So human capital theory emerges in the 1950s among economists initially as an explanation for what was a puzzle.

Generally speaking, you know, most economists assume that the rate of productivity growth, it has a very linear relationship with the amount of investment in capital and tools. But when they looked at the history of American productivity growth over the previous five or six or seven decades, they saw that productivity growth was actually increasing more rapidly than the rate of capital investment in physical capital.

And the explanation they came up with was, it's not just physical capital that increases productivity, it's actually the quality of the workforce, right? So if you train people to be better workers, then you will actually have a higher rate of growth. So when this idea is... And, and by the way, Clark Kerr, who we talked about earlier, was a big believer in human capital theory.

He saw the multiversity as investing in the human capital for society as a whole And he said, "This is great. Like, the reason why the state of California should invest all this money in the college system," and by the way, it was free back then, not just low cost, but free. You had to pay for the dorms, but you didn't have to pay for anyth- anything for the education.

And as recently as when I was going to Berkeley, my first semester at Berkeley, it only cost $800 a semester to go to Berkeley. That's, uh, I'm dating myself by saying that, but you know, it used to be even pretty cheap. Why should the state spend all this money on the University of California? It's because it would be, it was investing in the human capital of society as a whole, and it would improve the economy as a whole, which we all would benefit from.

That was kind of the dominant theory. Then what happened was two things simultaneously and in parallel. One is that a lot of people, specifically at Berkeley, really hated undergraduates, really hated this view. You know, they thought, "What are we, just cogs in the machine? Are we just like, you know, you're investing in putting education in us so that we can go work at a, you know, chemical company?"

And this is where, you know, Mario Savio, the leader of the Free Speech Movement in 1964, he gives a very specific image. He says, "You have to throw yourself on the gears because the operation of the machine becomes so odious." That's the line, the famous line he uses in one of his most fiery speeches in Sproul Plaza in 1964, and he used that metaphor really specifically because he felt like the University of California had become a completely instrumental place where instead of promoting human values and helping people realize their own life philosophies and to have a life of disciplined study, was just actually trying to push them through on a machine-like basis to run through a machine, eventually to become cogs in the, cogs in the overall economy.

So on the one hand, there was a student revolt against that instrumental view At the same time, Wellman also points out there was a transformation in the understanding of human capital that evacuated the idea of social benefit and the enhancement of the economy as a whole, and instead focused on the way in which educational investment actually improved you as an individual.

And they, they, you know, this was, to use the jargon, this was methodological individualism, right? The idea is we're not investing in human capital of society, we're investing in the human capital of this student and this other student, and this other student, and this other student. 

[00:45:02] Matthew Rascoff: You could say the consumer model.

[00:45:03] Nils Gilman: The consumer model. 

[00:45:04] Matthew Rascoff: It's the consumer model. Exactly. It's, it's a lot of, a lot of the education industry is organized around this idea, too. 

[00:45:09] Nils Gilman: Right. And by the way, yes, the overall economy grows, but who's the primary beneficiary if you get a degree and you're able to make twice as much money? It's you, right?

And so that became a justification for saying, A, if this in- investment in your human capital by going to college is gonna enhance your earning power, well, then you should pay it forward, right? And why not, you know, why shouldn't the university itself try to capture some of that benefit? And so they started trying to charge.

I mean, Ronald Reagan, when he fires Clark Kerr, and one of the very first things he insists on is that the University of California absolutely has to charge something to go there, partly 'cause he wants to like, you know, have student... And he says it explicitly. Ronald Reagan says, "Students need to have skin in the game."

And the idea is they're gonna benefit from this education. They should pay for it. Why should we pay for it when they're the- Right ... individual beneficiary? And so this then set off the spiral of increasing costs, and then students, as they're paying more, demand more services, and then s- then, you know, universities have to provide them more services, 'cause if you're gonna charge people $50,000, $75,000, $90,000 a year to go to this university, they ha- they come to have a set of expectations about the quality of the physical plant, what kind of mental health services and what kind of health services, et cetera, et cetera, et cetera, and training and tutoring and so on.

So there's this inflation in the number of things, and this only enhances and increases the number of functions that the multiversity is supposed to be having. So it creates this metastasizing problem of the multiversity because the individual student becomes seen less as somebody that society is investing in and instead of somebody who's an individual intellectual entrepreneur 

[00:46:39] Matthew Rascoff: And you can see in that some of the populist, you know, backlash, or at least the origins of it, when only, you know, forty percent of the population has a bachelor's degree, why should everybody subsidize that?

And there is a certain logic to that critique if, if most of those benefits will flow to the forty percent who have it. And we know that divide has gotten bigger around health outcomes, around economic outcomes. Like all, all these deaths of despair can be tracked to the bachelor's degree. So there is some rationale there.

It's not totally illogical. 

[00:47:09] Nils Gilman: Yeah. And you know, and Well- and Wellman, Wellman talks about this, I thought, quite movingly in his book. I mean, he's, uh, he comes from a big family. His parents didn't go to university. Only one of his siblings, one of his, I think, seven siblings, in addition to him, went to university.

And you know, he points out that even though higher education is no longer about disciplined study, it is still the gateway to up-- middle class and upper middle class access, right? If you don't have a c- university degree, it's hard to get ahead in America, and that sucks, especially if this government is subsidizing it.

So this brings us all the way back to Ian Bogost's argument about are the liberal arts colleges maybe gonna be the winners, and is that kind of the model? And one of the things I conclude is that, yeah, a liberal arts college may well be the model, not just for the financial reasons that we're talking about, 'cause they don't rely on a lot of state and government money, or any really usually.

It's also because their model of education and pedagogy is the one that's closest to the one that I laid out original- uh, originally. Small seminars, really tutorial work, those kinds of models already exist. And you know, we have, we have thousands of those kinds of colleges across the country. Now, a lot of them are in trouble.

Uh, Hampshire College just closed, for example, this fall. Yeah. So there's no bed of roses, and it's not gonna bend the cost curve to have all of a sudden a lot of bespoke teaching. I mean, obviously, that's an expensive thing and an expensive model. So you know, one of the questions I land on at the end of the essay for persuasion in "The Chronicle" is I don't know if this scales.

And are there possibilities for using technology to bend the cost curve? I think there are. But at the end of the day, a lot of custom education is gonna be an expensive thing. But then again, if everything else becomes cheap, then investing in disciplined study and making people really learn really, really well the skills that are gonna continue to be valuable in a world where AI is commodifying most other things, that may well be the world we wanna go to anyways.

[00:48:58] Matthew Rascoff: Well, I think that's a great note to end on, and I do commend you for thinking through the scale implications of this piece. I mean, that's part of what distinguished the essay for me, and I think it spoke to your technological orientation that we started with, that you're willing to, to contend with the Baumol's cost disease question and, and how this might play out n-not just for the select few, but for many.

I really love the essay. I hope many more people will read it after hearing your summary and our discussion of it. I encourage you to check it out. In persuasion, where should people go to find your writing? I know you publish in Noema, you've got your Substack now. What are your outlets? And maybe you can just close with some reading recommendations of other folks that you enjoy that you'd recommend to, to listeners for this.

[00:49:43] Nils Gilman: I publish in a variety of different venues. I've had a couple of pieces in foreign policy this last year. Had a piece in Dissent. I had a piece in Salmagundi that just came out. I don't generally publish in academic peer-reviewed forums anymore because I'm so far away from that. But yeah, set, set up a Google alert and you can get my stuff any- wherever it may appear.

[00:50:04] Matthew Rascoff: And you've got the Substack too. It sounds like you're doing that independently. 

[00:50:08] Nils Gilman: I am, I am. I, I, you know, I was part of the first wave of blogging back, uh, twenty years ago, and then I dropped it. Basically, Twitter ate my, uh-- I'm not on Twitter anymore. I am on Blue Sky and I am on Substack, and those are probably the main ways you can, you can get my stuff.

The main thing I use Blue Sky for, I don't really sort of get in, get into the back and forth with all the sort of annoying lefties on there. What I do is I actually use it in the sense as a log of the stuff I'm reading, where I often will have links to article. Part of the reason I left Twitter is that one of the things that Elon did is he started downranking people who had links to articles 'cause he didn't want people leaving the Twitter platform.

And since one of the main things that I had used Twitter for was both to share content and to get content from other people, it basically degraded the quality of what I was there for. So yeah. In terms of other people's writing, I read pretty omnivorously, so it's a little bit hard for me to answer those kinds of questions generally.

I'm reading a fun book right now called Goliath's Curse, which is about societal collapse in historical and futurological perspective, and I do a lot of reading on those kinds of things. 

[00:51:11] Matthew Rascoff: Fantastic. Thanks so much for being with us. It was a great conversation, and I look forward to the feedback that we get from our listeners.

[00:51:18] Nils Gilman: Thanks a lot, Matthew. 

[00:51:20] Alex Sarlin: Thanks for listening to this episode of Edtech Insiders. If you like the podcast, remember to rate it and share it with others in the Edtech community. For those who want even more Edtech Insider, subscribe to the free Edtech Insiders newsletter on Substack.

This season of Edtech Insiders is brought to you by Starbridge. Every year, K-12 districts and higher ed institutions spend over half a trillion dollars, but most sales teams miss the signals. Starbridge tracks early signs like board minutes, budget drafts, and strategic plans, and then helps you turn them into personalized outreach fast.

Win the deal before it hits the RFP stage. That's how top Edtech teams stay ahead.