A

Foreign.

B

Welcome to the Report Card with Nat Malkus, the Education Policy Podcast from the American Enterprise Institute. On this podcast we've discussed AI and K12 education a fair amount. But how will and how should a AI change higher education? Should any human instructors be replaced with AI? Should universities change what and how they teach to strengthen their value proposition? And in the age of AI, how should universities prepare their students for an unpredictable job market? To discuss these questions and more, I invited Hollis Robbins onto the podcast. Hollis Robbins is a professor of English at the University of Utah, where she was formerly dean of the College of Humanities. She writes about AI in higher education on her substack Anecdotal Value. Hollis Robbins, welcome to the Report Card.

A

It's a pleasure to be here.

B

So, Hollis, a basic question to start with. How high is the quality of teaching in American higher education?

A

Well, first of all, there's no studies, so I could say it's all over the place. It's really bad. The average, the median, the mode, we just actually don't know. What I've seen as a former dean, as a professor at multiple institutions is that the best is great. I would say most teaching that a student is going to get at a college or university is going to be mediocre to bad, if not very bad. Particularly given the rise of online asynchronous instruction.

B

How much has that online asynchronous instruction invaded? Sort of the market share of sort of a typical higher education program. I mean, certainly we know that in K12, which is kind of my area, we know that there's a lot more, but, but that the modal instruction is still done through class, although lots of assignments go through these portals and so forth. How much is asynchronous instruction in the university? Again, not from necessarily studies, but from your gut. How much of an invasion into higher ed has that had?

A

Huge invasion because it's cheap. I mean, that's the thing. Anytime when you have a new course, for example, you've gotta go through all this governance committees to approve the new course to improve things. I mean, there's really high barriers to entry for new courses. And so every time somebody comes up with something new to be innovative, let's say the pressure comes down is like, can we have some online asynchronous sections of this, right? Or can you record some of your lectures so that we can put it on the web and other people can take this? So anything new, the demands are can we make it online asynchronous to make some money off it? Or to at least lower the costs of new innovations and then existing classes. It's the same thing. Or existing programs. And the rationale is never, oh, this is going to be cheaper, but oh, this allows access. So what becomes a reason to do things because it's either better or worse or becomes let's let more people have access to it. It's a sort of emotional political argument for doing it. Nobody ever says it's better.

B

And what's the unit of analysis here? Is it if I take, you know, Sociology 201 or whatever the case may be, that a chunk of sociology will now be asynchronous? Or is it that, well, there's a number of courses that are just wholly asynchronous. So is it the course or parts of courses?

A

Well, it's courses. I mean, I'm glad you mentioned sociology because sociology is, I think, one of the most egregious examples of content delivery that is being done online asynchronously because there's, you know, it's a matter of reading studies, saying what the studies say. How would you apply that to a new group of individuals? It's not a, it isn't a field or a discipline that requires this kind of conversation, at least at the undergraduate level. So when you look at, and I spent a lot of time criticizing Arizona State and you look at, for example, their big build out partnership with Starbucks, right? Almost all of that, channeling baristas, Starbucks workers into an ASU degree, was almost all sociology, almost all online asynchronous, and almost all of it terrible.

B

So you have said that general education courses may be particularly bad. How bad do you think general education courses are and what do you mean by general education courses?

A

Well, general education courses are what I would call known knowns. Do you remember the big Donald Rubsfeld quadrant? Knowns, unknown knowns, known unknowns, unknown unknowns. For example, the question is, what is the purpose of a university? Is it to give you the dates or French Revolution? Right. Is it to give you, you know, some background on, on Shakespeare? Yes, but in an age of information abundance, which is what we have here in the AI era, what is it that universities are delivering to students that is already available? What you know, the. What is perhaps unknown about Shakespeare or you know, if you're studying Macbeth for the hundredth time, perhaps you're studying with a professor who knows the latest readings of it or perhaps some new scholarship. But that doesn't happen in general education. General education classes are usually taught by lectures by adjuncts Online asynchronous by faculty who might be very good teachers but are far from the cutting edge. And the question is, is this worth

B

it and do professors want to teach those courses? I mean, is there demand for those courses and what's the impact of the answer to that question?

A

Well, two things in major state universities. You went to Covenant College, right? A. I did college, yeah. Tell me about. I mean, can I ask you, what was the gen ed there?

B

You probably there was a core set of courses and they were taught by regular professors, small college. So there's not a lot of outsourcing options. Part, part of the value proposition at, at Covenant was everything's sort of coherent and engaged. And I learned boatloads, 100%.

A

Right. And I think schools like Covenant are actually going to do very well in the AI E. You have engagement, one on one, in person engagement with a person that cares about how you learn, not just delivering instruction. What happens at major state universities? And this was sort of realization to me when I became a dean at Sonoma State in the Cal State system, the largest four year public university system in the country. It is the state legislature in the system that demands, I think it was about 40 credits of gen ed. And they don't care how you deliver it, they just want it delivered. And they have all sorts of reasons for why. And as a dean, I was the one that had to hire the faculty to deliver the courses. And let's just say with enrollments that were going up and down and with a really dominating union, the California Faculty association, let's just say the choices of faculty to teach the classes were not about. About expertise. And we're not always about quality. Sometimes you practically had to drag somebody off the sidewalk to come teach a class so that the demand to teach the class in most states in major public institutions, the state legislature says do it. And this is why I was talking about operationalizing before the job of the dean is to operationalize it. Right. And so I sought for quality. And sometimes you just can't. You can't deliver quality. And the legislature doesn't really care about that. Most states.

B

That's kind of a dismal assessment. How much do you think students are actually learning in general education courses? They're taking a lot of them, it sounds like.

A

Well, I mean, that is a great question. I spent a lot of time as a dean both at Sonoma State and at Utah, reading course evaluations because I was interested in this course question. And let's just say in some classes, and I'm not going to say which university. It's almost universally bad. A waste of time, particularly in first year writing. And I will just say nobody on the planet has ever learned how to write in first year college writing, ever. They either come in with it or they teach themselves. The courses themselves are a complete and utter waste of time, particularly in the AI era. Now when you have a student who maybe comes in to study biology and takes a literature class, sure, that's fun, that's nice, that's great. Could learn things. Sometimes you change a major. These are anecdotes and people love to tell these anecdotes. Oh my goodness. I had this student, she was a biology major and now she's a history major. I did such a great job. Fine. That's an expensive way, right? To support our students is this mandated layer of what is the equivalent of, let's say, $10,000 a year for a student to take a layer of classes just because something might be interesting to them, taught by people who are generally not as qualified as full professors.

B

So setting aside AI for the moment, we're going to get there shortly. If you were the higher education czar and wanted to improve our nation's general ed courses, what would you do?

A

I would cancel them all. I would send out, well, you can't cancel it at the state level, but I would tell states to look at the. And I can't set A.I. aside, really. I mean, it was a problem before A.I. the question is, what is a university for? Is the university to deliver known knowns, or is the university to help students who are going to push the boundary in science, in literature, in knowledge? Higher education should be where students are pursuing the frontiers of knowledge. Now, part of the reason that every state mandates general education today is because high schools are doing such a terrible job. And you probably saw that big dust up at UC San Diego where the faculty blew up and said, people are coming to us with no foundational education. We can't teach them. So we have this layer which allows more and more people to go to college because then they say, okay, we'll teach them this two year layer. This is a departure from the history of general education at a place like Harvard, the, the famous red book of 1945, which was, as with the GI Bill, as people coming back from World War II with different cultural backgrounds and expectations, a way to say, let us have a shared culture. So general education was to deliver a shared culture. That is not the case anymore. It is simply to get everybody up to a certain level and Again, for me, why I asked this question before AI, I ask it now. Operationalized, it's pretty bad across the board. Occasionally you have a great professor and a great experience, but it's not worth the money.

B

So you've made the case before that AI could and maybe should replace human instructors in general education courses. Make the argument, well, here's the thing.

A

I don't know. How much do you use AI? Let me ask you that beforehand.

B

I try and use it sparingly. I find I use it a little bit more and more. Claude code is extremely helpful for me.

A

I mean, these tools are really helpful for aggregating information. We got so used to Googling. Can you Google and find out this? Can you Google and find out that? And again, I'm old enough to remember when before everybody had an atomic clock on their phone or what have you. You'd call the operator to figure out how time, what time it was. Right. I had all the sets of Encyclopedia Britannica on the shelves growing up. Right. There have always been technologies to bring us sort of aggregated information. And we have a new one now which is much easier, much easier and wider to use. I mean, the kinds of informations that. That you can get. So the question is, if you ask, if you ask, what is general education? Almost every state will give you a set of quote, unquote learning outcomes. Okay? These are the things you need to learn about biology. Okay? So plug those into your clod, right? And so you are a student before you show up. You have to learn these things. Well, there are many ways to learn something. You can learn them through storytelling. If you want to learn something about sunspots or tides or photosynthesis. Right. There are many different ways that that information can come to you. And what AI can do is engage with you, with your interests, how, you know, perhaps you grow up on a farm and you want to learn things about the sun, about water, about, you know, insects through that kind of engagement. I think that this is more efficient, more exciting than sitting in a classroom and perhaps working with a teacher in a textbook that is out of date. So, I mean, I can tell you this in a pointillist sort of way. The whole endeavor will bore you to death. But it's possible.

B

Do you think that this is something that should be happening now, already in 2026, or do you think this is something maybe for a few years when we kind of get it together?

A

I think it should have happened already. I would have liked. And a year ago when I published some things on my Substack. I had hoped that by fall of 2026, there would be universities or some schools that would be delivering gen ed through AI. Part of, again, in California. California, where I've spent sort of the most time and attention looking at how they deliver gen ed. They have this thing called calgetc, which lays out the learning outcome for the general education, which they have standardized across community colleges, the Cal State system, and the UC system. So if you get into a uc, let's say you got into Merced or Riverside or what have you, they tell you, you know, you can go take your gen EDS over at the community college and save some money, right. There was a big story in the Washington Post yesterday, excellent story, about how students were zipping through these courses online, asynchronous course, in weeks instead of a semester. We already see opportunities that students, when they are personally engaged in, you know, in getting through something, they can, they can get over these hurdles in ways that our current system is not embracing. So I, you know, written over and over again how California, instead of sending students to spend two years in a community college where they may end up not finishing their degree at all, build some sort of AI so that they can get through things in a semester and start their education.

B

How much of your argument is sort of resting on that AI is better at teaching than humans? And how much of it is that AI is good enough, but it's far cheaper or maybe far more flexible?

A

Well, again, one of the benefits of being a dean is that you get to see from up here, teaching. Not just my teaching. Most professors get to see themselves, maybe a couple of colleagues. Team teaching, you don't really get to see, you know, numbers. Most teaching is really bad. You started out by asking me that, right? Lots of teachers just phone it in. I've written about the fellow I go for, my hair, who was getting a degree in a community college. You know, he's taking a few classes. And so while I sit in the chair, he tells me about this. And his professors or instructors are teaching a 5, 5 load. They're phoning it in. And so he uses AI to pass all his classes. It's a complete and utter waste of time for him, for them, for everybody. Not for me, because I get the anecdotes right. But I would say AI is maybe at 75, 80%. It's not, it's not as good as the very top layer of teachers, but it is better than 75% of the instruction currently going on.

B

Hollis, if you were to teach a general Education course, would you outperform the AI?

A

Oh, God, yes.

B

And what is it like? What's. What's the. What's the clear marker of distinction? I didn't expect you to say no. But I want to know why.

A

Well, because my teaching has always been at. I've always been interested in the edges of knowledge. My scholarship. I have many areas of scholarship. My primary scholarship in the last few years has been the African American sonnet tradition. Very, very narrow slice of the world's knowledge. But I am probably, maybe that there's three people in the world who know as much as I do about this subject. Now, why I say that there's a great line from Jean Piaget, who by age 13 was the world's expert in the mating of freshwater mollusks in Europe. And he said later on, after he'd spent his childhood and through college being a naturalist. He said, when you learn everything about one subject, it trains your brain to understand how we approach knowledge, how we understand things, and where the edges are of what we know, how we know what we know, what is left to be known, right? So this is how I operate in my scholarship. So when I'm teaching students, even if I'm teaching them about something that there's a lot of knowns about, right? So one of the classes that I'm teaching this fall is on American literature, but I'm teaching it from the perspective of how Chinese. These universities read the same literature to learn about us. So we're reading the Godfather, which is a crazy book. I know everybody's seen the movie, but the book is amazing. So what is new for these students would be how to read it, imagining yourself looking at America through this book, right? And there isn't a lot of scholarship on that in America. So I'd be pushing the students. This isn't a gen ed class, but I would be pushing the students to say what is new, what hasn't been written about this, what. What have, what has not been, what has not been thought of about this? And that's why I would do a job that I can't do right now.

B

When you teach, how do you use AI in your classes?

A

Well, I'm teaching for the. I've been a dean. I stepped down as dean. I have had a little leave. So I'm going back to the classroom this fall for the first time since the AI era. So I've been spending so much time on this. I'm teaching a class on AI and literature. I'm teaching this American lit class. I will be Asking students in both classes to get to the edges of what AI knows and to ask them if they're writing papers for me or however we'll be doing some grading and assessment to, to deliver things to me or to understand things beyond what AI knows.

B

So I'm sure many of our listeners are going to be skeptical when they hear you say that AI is as good of an instructor as you say it can be. How do you respond to just the, the, the general skeptic that. That says, I don't know.

A

I think any skeptic should spend some time talking to current college students or recent college students. I think, you know, the other couple of months ago, I was. I had a flat tire and I had to go to my, my local garage in Salt Lake City, and there was a woman behind the counter that I, I could kind of tell was a college student. I said, you know, are you a college student at the University of Utah? And she said, yeah. I said, you know, why are you here? She said, well, I dropped out because I couldn't stand these gen Ed classes. I want to be a neurobiologist. I really wanted to just immediately start neurobiology. And I couldn't believe I was in these classes where people who didn't know what they were talking about were teaching things that I could find out elsewhere. Yes, that's just an anecdote, but it really profoundly moved me because here was an example of somebody who would rather work online on AI, find a way to sort of get through anywhere and work in a garage behind the cash register, working on flat tires than go through that. I talk to students everywhere I go. I just say, like, where do you go to college? What do you do? What do you think? And nobody's happy except at the elite institutions. I'm sure at Covenant they're very happy.

B

Well, my son's attending there now, so I'll catch up with him on that.

A

Well, doesn't he complain or tell you already?

B

He hasn't been complaining.

A

Good. See, this is what I mean. If you're spending time with a senior faculty member, a qualified faculty member, somebody who's been doing it for the long time, somebody with a reasonable teaching load, not a4.4 or5.5 teaching load in small classes, I'm sure you're happy.

B

So today, when professors are trying to deal with AI and not as a replacement, but as including it in their courses, it seems to me that there's sort of three ways they could think about it. They could try and block it. Look, this interferes with my production function. So we're going to just keep it out. They could try and use it sparingly so that they say, you know, we're going to use it here, here's the productive exercises that we're going to model and we're going to try and keep a fence up from other things. Or they can just assume, you know what, there's, there's no holding it back. So I should change my production function in how I do my classes and assume that kids are going to use it. And that's sort of the blue books and the oral exams and these sorts of things. What do you say about those options? How do you respond to professors who ask you from that perspective?

A

Well, the third way is really the only way, though. I think blue books are whatever. I have opinions about that. Like, the real question is, it used to be that faculty had a monopoly over what was being delivered. You went into a class and 100% of what you learned was. Or 88, 98% was from the, from the professor. Right. I'm going to tell you this. Here's your course packet, here's your books, here's the lecture, and then I'm going to grill you on this. The idea that your student would be getting the majority of the material not from the professor was just. That didn't even come into it. Except for somebody, sometimes somebody from France would take a French class for an easy A, and in that case that would be a problem. Now, now, I would guess, I mean, I've heard anecdotes. There's no studying that most students, you know, instead of reading the textbook, they're going to Claude, what does this textbook say? Right. So what, maybe 30, 40, 50% of what is being delivered is from the professor and the rest is from an external source. Right. Number one, the professor has to realize this, that there's no monopoly on the content delivery. So then the question is, if you're spitting something back, you know, how is Bluebook or not Bluebook going to change the fact that they may have gotten their 90% of their course instruction and the information from Claude or from someplace else? Right. The only way to teach is for, as I've been saying, is to have an engagement. Like, what did you learn? Is it real? Is it a known. Known. Who else has written about that? What do you think about that? Do something new. And how do you know what is new? And how do you know what everybody in the world is doing? So, you know, is to get back to the fundamental of how do we Know, how does a teacher know what you know? Like it used to be? It's like, well, because the professor said so. It's in the textbook. I think students today and faculty today need to ask more fundamental questions.

B

Hollis, you say that soon people will start wondering, why pay tuition when they can just learn this from AI instead? But I'm wondering how new of a problem this really is. So for many students, if they are sufficiently dedicated, they can probably learn far more with access to a good library than they can from general education courses. Probably even more from all the free university lecture videos you find online. Brings to mind the goodwill hunting quote you'll realize in a little While you dropped 150 grand on an education you could have gotten for $1.50 in late fees at the library. The problem, of course, is that most students aren't sufficiently dedicated. They need the structure that higher ed provides. Why do you think AI changes that equation?

A

Well, first of all, we're in an era of information abundance. And this idea of who has the monopoly on the best information, right, it, you know, professors don't have it anymore. But then you look at the question of what does a teacher do, right? So once upon a time, right, you would have come to the class and, you know, if you weren't self motivated, like Good Will Hunting, right. If you weren't self motivated, your teacher was a bit of a coach. Your teacher was formidable. You would get punished for not showing up, for writing bad papers. You know, I would be able to stand in front of the class and inspire you to go do your things. That professor role has been diminished, right, for so much to the point where in fact, my field of action as an inspirational, charismatic teacher is severely limited, while at the same time you could go to the gym or get on a team and be screamed at and inspired by a coach to give your best. And that's okay, right? So part of what is happening on college campuses that again gets higher ed to the point that AI can replace us is that structures have kept excellent professors from being truly excellent. It.

B

And so you're getting at something here that I want to, I want to scratch on a little bit more, which isn't that, well, there's an ideal for education and that AI can meet that ideal, but rather that the counterfactual is business as usual. Business as usual for many professors is very constrained and restricted. I'm assuming part of this is because, of course, reviews and so forth and the, the complaints of, of students that constrain how hard educators, college educators, can Push on students. So you're not comparing the ideal to what AI can provide, but you're trying to open the door and say what students are actually getting is not necessarily better than AI. That is 100%.

A

Yep. No, that is 100%. My argument, my argument is that things had attenuated so bad before December 2022, right. When ChatGPT dropped. Right. You started out asking me about online asynchronous, all of this stuff, all of

B

that

A

which has said that the teacher doesn't matter. Online asynchronous classes mean the teacher doesn't matter. Enter a technology. Right. I don't have to make the argument that the teacher doesn't matter. Higher education has already made that case. Right. California has already made the case that it doesn't matter whether you, whether you're taking the place at a community college or at a UC system. So AI, I see, is taking taking advantage rationally of the weaknesses in the system.

B

Let me push on this a bit. AI is a tool and students know it's a tool. There's a transactional that you don't have to be nice. A lot of people say please. You don't have to say please. It's a tool. I think fundamentally, for the most part professors are not tools. And I don't mean that in the obnoxious way. Doesn't that still make a difference? The fact that, well, I have a class and I have a professional professor. Let's assume this is not an online class. Does the fact that AI is a tool add more problems that we don't have right now?

A

Well, the problem is that most students at most public universities won't even get to a tenured professor, an expert in their field until their junior year and even then. So really we're still back to this first two year gen ed problem. Right? What is this sort of substrate level and how does it matter? I mean it's institutions who have said to students, your professor doesn't matter. It's the delivery of content that matters. It's your learning outcomes that matters. It's meeting requirements that matter. Right. I didn't put that into operation.

B

A lot of classes in college, I hope he hopes, without having an evidence base for it, are still discussion based. And obviously students can discuss subjects with AI. But I think many professors would argue it's important that students discuss what they are learning with each other. Do you think discussion based classes can exist with AI or is there an essential human element there?

A

100% look, places that like Covenant, like I spend A lot of time talking to zena hitz at St. John's College about or other higher ed innovators who want to come up with some hybrid St. John's type AI delivery classes. I spent a lot of time with Brendan McCord at Cosmos and he's got a team that's really doing some innovative thought on what the AI university of the future would look like, which is some sort of combination of St. John's and NAI discussion based information elsewhere. And we see this, I mean, you're a K through 12 person, you've probably heard of and talked about the Alpha School, right? Two hours of delivery of AI and then sort of enrichment around that. I actually think that that is the future in higher education. The problem is dismantling the current bureaucracy in higher education. The legislative demands for general education would have to be dismantled. Ask any public college student how many small seminars you're in. Right. Or the pressure of legislatures to raise the caps on those small seminars. What is a small seminar? Is 35 people a small seminar? Right. State legislators want to see efficiency. A 12 person small seminar is not efficient. It never has been. It's effective, it's not efficient.

B

So it can be easy to forget now. But it wasn't that long ago that we experienced a pandemic and school was virtual. Many parents and many students hated it. And not just because the educational quality was poor, but because it was impersonal. Students like being among other students. They like being face to face with an instructor, even if it's sort of inconvenient. Do you think our recent experience with the pandemic will make parents and students more averse to the kinds of courses that you're describing?

A

Well, you know, I would have thought so. Right? I'm going to sigh tragically. I would have thought so. Right? And yet. Right. And yet. You see, I was just listening to, before I got on this morning, a podcast at Freakonomics with Michael Crow who was talking about how awesome it is to have online asynchronous education because it's innovative and it has access and it's better than students in an old fashioned university studying great things. And I'm listening to this thing and I was like, do you remember the pandemic? Do you understand that, you know, that our, our universities are now, or have been until they've started to crumble, the best in the world because we honor the past, because we transmit the best of what we know and what humans have done and discovered in small groups to students hungry for that knowledge. You know, I thought that that was the level. The lesson of COVID is that people want to come back to small groups and, you know, interpersonal discussions. But the realization is. And Michael Crow kept saying the word scale, Scale, scale. We've got to scale. We've got to scale. Right. Coventry doesn't scale, does it? Part of its excellence is that it doesn't scale. Right. And I think the pressure on universities to scale are why they are uniquely vulnerable to AI.

B

All right, Hollis, it's time for Grade It. I'm going to give you some topics. You give me a grade and a brief explanation. Are you ready?

A

Yep.

B

The Advanced Placement African American Studies course.

A

I don't even know what to grade it. Let's just say I don't believe that there should be Advanced placement courses. So what grade is that?

B

An incomplete, I don't think.

A

I mean, neither of my kids did Advanced placement in high school, so it doesn't matter to me.

B

Being a dean,

A

I enjoyed it. It's a very hard job. I'll give it a C. Using AI

B

to write academic articles.

A

A. I think everybody should because it's the best way to figure out who has. I mean, you could spend a lot of time, as we all used to, on databases, on Google Scholar, in the library, to find out who has written on something before and to position your article against that, against what has been written. I think one of the sort of most. So much scholarship people have skipped that. That's like, I've got something to say. And they don't go look to see who has said it before. This is before AI. Now, AI's ability to give you that information means that if you don't do that, you're doubly stupid. Right? So if. If. If people are writing papers without doing that step with AI, they shouldn't be writing papers. Everybody should be doing that. Now, once you've done that, you should kind of write and think your way through your paper yourself. But everybody should use AI.

B

Robert Hayden.

A

Oh, A. He's probably the greatest poet that not enough people have read or heard of. He has a great poem called American Journal, which is a poem in the voice of alien visitors to America reporting back to the mothership, all of our foibles. And it's a brilliant poem.

B

Efforts to achieve viewpoint diversity on college campuses.

A

I think a waste of time. Deck chairs on the Titanic. F. I think universities do this because it's cheap. It has nothing to do with expertise. None of the viewpoint diversities ever asked me. Viewpoints on Robert Hayden. Poems and that is actually what I have the most to say about. So I think it is an F. I think it is a waste of time. I think it is a window dressing. And I will say one more time, deck chairs on the Titanic.

B

Having Aphantasia A.

A

It's so exciting. I thought it was when I discovered that I had the inability to voluntarily call up visual images. When I learned this about four or five years ago, I was sad at first, and then it made sense to me that, you know. Do you know how in your phone, your big photo bank takes up so much memory? Well, apparently that's your. Do you have it?

B

I do not.

A

Right. So your brain is filled with pictures. Right. Mine is nice and clean and empty to have really interesting, innovative thoughts. So I'm happy about it.

B

Does having aphantasia change how you read and study poetry?

A

Well, I realized, you know, I don't understand, you know, like William Wordsworth's daffodils poem. Like, I don't see daffodils. But the word daffodil is really interesting with Fs and Ls and, you know, it looks interesting on the page, like a field of daffodils. So it's fun to me to read things in words and letters that maybe other people are filling their mind with pictures. I think, look, the differences that we have inside our heads are so much more interesting than the differences outside on our faces. And I think spending more time on that would be wholesome for everybody.

B

Civic centers in American universities currently structured

A

with the new legislative. The new legislative mandate to build them. I would say I would give them a provisional F at midterm with the possibility of getting an A if they recognize that black studies programs on their very campuses are asking the same questions about individuals and their responsibilities and ability to participate in the government and defining citizenship. If civic centers begin to embrace black studies programs on their own campuses, I will give them an A.

B

The sonnet as a form.

A

Oh, the sonnet is the very best form. I give it an A. Everything you would want a poem to do in 14 lines. You got a problem, you got a solution. It's handy, rhymes in some nice ways. Flexible enough that you can kick it and throw it on the ground. It'll bounce right up and still be a sonnet. It's. It's the best. It is the best poetic form.

B

Give me a grade for how big a problem grade inflation is in universities.

A

Well, the problem of grade inflation. I was just interviewed by. By this lovely reporter at the College Fix in Fact was out today on the real problem is not the B to the A. The real problem is the F to the D to the C. So universities do not like students to fail or get Ds because it brings down their college rankings, even though we want colleges to be firm and educate students. So I would urge, I mean, it's an A problem that I give an F to the attention being paid to it.

B

The films of Quentin Tarantino.

A

Oh my God, the best. How many times have you. What's your favorite one?

B

Pulp Fiction's just the classic, I guess.

A

Oh, it's the classic. But these Revenge ones, Hateful Eight, Django Unchained, Inglourious Basterds, that revenge trilogy is some of the best cinema ever made in America. So absolute A.

B

What is the aspect that makes it so great?

A

So I study the music that Tarantino uses and he often uses music that have appeared in other films that he imports into these, into these films which you don't know. But if you're a film going person in the back of your brain, like multiple narratives of what is going on in the film are, are should be rushing into your subconscious. And I think a filmmaker that has that much attention to craft is worthy of attention.

B

Last one, Sports betting.

A

I don't understand this. Let me just say I don't understand it. I give it, I mean, I don't want to even give it a grade because I don't understand it. If people want to spend money that way, who am I to say no as a libertarian?

B

All right, thanks for playing great at Hollis. It seems to me that when it comes to AI in universities, many universities have their heads sort of buried in the sand. Do you think that's true? And if so, why do you think that is?

A

100%. Look, I was lucky enough to have met Sam Altman at a small dinner in 2019. Six person dinner. I was sort of sitting at the table. I didn't know. Know what OpenAI was. I didn't know what the future was. But I, I asked for and got early access to GPT3 in 2020 and I started using it and I, it was, you know, it was a little slow to, to realize, but you know, maybe a couple of months in, I was like, this is going to change everything. This is going to change everything. And when Chat GT dropped in the fall of 2022, I immediately made all the department chairs get subscriptions. And I said, find out what it knows about your field. Find out. Get to the edge. I said that to everybody. Get to the edge. So that we know and keep trying to see where the edge is. And as these models have been getting more and more sophisticated, better training data, it is absolutely the case that universities have lost their monopoly on knowledge. Lost it. I mean, we have a tiny sliver, there are unprocessed papers and books and research in universities that we have a monopoly on the edges. And the edges are where the most valuable knowledge is. Right now I hear no university presidents talking about this. None. And so this is what I talk about on my substack to try to get attention to it.

B

Do you think professors feel threatened by AI or not threatened?

A

The burden should not be on faculty to figure this out for themselves. The burden should be on university presidents to say, what is our role in this new era? And this is why I was looking at this. Michael Crow has this big platform. It's supposed to be the most innovative university in the country. He hasn't talked about this. He keeps talking about filial pietism and the problems of universities in the past. Everything has changed. University presidents need to be saying to their faculty, let's solve this problem as an institution. It should not be on you and you alone.

B

You're a professor in the humanities. In the humanities specifically. It seems many professors are particularly skeptical of AI. Why do you think that is, if you think it's true?

A

It's absolutely true. And it's fear. Fear. I mean, again, I don't think, I mean, all of half of my substack is yelling at higher ed leaders. The other half is addressed to colleagues saying, get over it. Right. Understand what you know that AI doesn't know and spend your energy there. Do you know things that AI does not know? And if they say no, they shouldn't be in the classroom anyway. Right. And. And if they say they do know more, I say then you're fine.

B

Do you think AI will put many professors out of a job?

A

Yes, AI will put the professors out of job, out of a job who cannot articulate what they know and what they can do that AI cannot deliver.

B

How much do you think AI should change the administrative work that goes on at universities?

A

Well, I think it should change a lot. So we were having, we were having this conversation, if I could say so, before we got on the podcast about, you know, these demands on universities for return on investment. Right. And legislatures are wanting to say, we want to say, we want you to show university that your history graduate, the salary of an average history graduate from your university makes more, more than X amount. Okay, now who should do that work a right. This is what legislatures are asking, which is I think a ridiculous question to be asking, especially without operationalizing it, who should be collecting that work? Is it the job of the educator to monitor what you do after graduation with your degree? Now can AI help with that? Probably. But right now the burden is on professors to make sure their students get good jobs. At the same time they're trying to deal with AI. At the same time they're trying to deal with a million other things. When I say that it is the university president that needs to address these questions, it's because nobody else can do it.

B

Even before AI, there were plenty of people that would argue that there's many more administrators in universities than there needed to be. Are you sympathetic with that position?

A

100%. 100%. And I think again and back to the grade inflation question. If you allow students to fail, then you can get rid of a whole offices of people whose existence is keeping them from failing. Right now the amount of infrastructure that is deployed to keep students from failing is huge. Also, if I may talk about something briefly, which is really, really boring, so I'm just warning you, is what I call the transfer industrial complex. Okay, so what percentage of students do you think transfer at some point in time during their college years?

B

If we include all Title iv institutions, probably 30%.

A

Yeah, it's about 40%. Right. 40% of students transfer. The bureaucracy that is required so that courses seamlessly transfer across institutions is vast and incredibly expensive. So you have to have this whole group of people looking at every class, right. To say, does it meet these things? Does it do this? Does it show this? Is there this kind of learning outcome? Right. So that there's just a entire bureaucratic layer in existence invisible to you? Not invisible to me when I was a dean. So you are paying, when you pay for your kids education, you are also paying for this transferability. Now why, as I said, this keeps individuality, this keeps. This says that the professor doesn't matter who taught the class, the course matters more than the professor who taught the class. Like everything is getting flattened into this sort of instructional. You know what, what Michael Crow calls Cheesecake factory, right? It's a cheesecake factory. That once again, what you're paying for is the kind of flattening that says AI could deliver this just as well because the system has already taken the professor out of the equation.

B

So Hollis, if universities were willing to employ far more administrators than they needed even before AI, that suggests to me, at least to some extent universities aren't the most efficient organizations out there. And if that's the case, why do you think universities will end up cutting human professors and administrators now that they have AI? I mean, will there be enough competition to actually turn this into job cuts?

A

Well, frankly, I think more state legislatures should listen to me. I mean, I have a small platform that I've got a bunch of educators, college presidents, faculty, economists, tech people who listen to me, who invite me to come speak. Going off to South Korea in a couple of weeks to have a big conference on AI and higher education. The audience I don't have, the audience that I should have, is state legislatures. They are the ones with the power. They are the ones that are accreting responsibility onto institutions to measure efficiency. Well, asking universities to measure efficiency is going to make them more inefficient.

B

Hollis, you're going to South Korea. Do other countries have a different take on this? Are they viewing AI and higher education in the same way? Sort of a slow or maybe head buried in the sand fashion that we're

A

doing in the US What I'm seeing in other countries is, is institutions that are not burdened by state legislatures, that are not burdened by a whole bunch of committees of people. What will happen is some state legislature from some county someplace will have a niece that moved universities and a credit didn't transfer. Right. He's like, oh my God, this is a problem. We have to make sure every single class taught at every place transfers to every other class because that was will make higher ed more efficient. The amount of bureaucratic labor required to make a class transfer is labor that is not done to instruction, that is administrative. So you know, it's like, I listen to these things. It's like if you want a more instruct, if you want a more efficient university, stop demanding things that are not instruction. Focus on instruction, small class sizes and faculty that understand the edges of their field.

B

So echoing that, I want to ask if you think the future still holds a place for human led instruction in universities.

A

100%. I think that is the future. That's why I keep asking you about Covenant and I keep thinking about places like St. John's I think that AI in some form, with some sort of guides perhaps like Alpha School, should deliver whatever general education is required and everything else should be small classes with subject matter experts.

B

You've used the term last mile education before. What do you mean by that?

A

That means, you know, the things that I know that aren't in AI like the AI can tell you up to a certain thing about Robert Hayden, for example. And I test these things all the time. What do you know about Robert Hayden? And I'll spit some stuff out. And then there's what I know about some unpublished letters that he had or some correspondence or some ways that he was reading Robert Frost, and Robert Frost really matters to his poetry, for example. So what is known that the AI algorithm won't tell you is what I call last mile. So if you were going to use, you know, if you were going to use your AI to decide where to go to dinner tonight, right? Like it could say all of your preferences in this and that and the other, but you happen to know that this particular restaurant that it might suggest is next to a place where a grunge band practices and it's going to be really loud. You have that last mile knowledge and you're going to say, well, I'm not going to go to that restaurant on that night.

B

What should we make of the fact that AI's output is just generally truncated? It's just very, it's usually sort of summarizing, giving sort of like the lowest common denominator. One would think that when you go into a human LED course, sort of a traditional course, there's a, there's some drudgery, there's some pain in having to read the whole thing and having to kind of go through this. I, I experienced this undergrad and grad. Why am I doing this when I get on the back end? I understand that I've been steeped in a wide array of things and not just been given, I don't want to say like the CliffsNotes version, but nonetheless, when you ask AI for an explanation, generally it's sort of summary is, is that a fundamental problem in the way that AI interacts with students at the university level, that it must be reductive.

A

So, okay, this is a really interesting question and what I think most people don't understand about AI, and this fellow Gary Tan runs Y Combinator, had this really excellent tweet the other day about what's called a Skills Fund file. Right. Most people log onto their AI and ask a question, pro model, what have you and get exactly what you're talking about, like just synthesis, compression, tops of waves. Right now what a Skills file will do is say, look it, I want complexity. I want to know how many people disagree with this. I'd like you to do a consensus on this subject and I'd like you to deliver the 10 different ways to think about this. And you know, you just plug sort of this is what I want alongside every query. And it will give you a hundred percent different output, it will give you complexity, it will give you arguments, it will give you, like, things that you didn't think about. Give me 10 things that I didn't think about when I asked you about, you know, the history of the guillotine and the French Revolution or something. Something, right? So the problem is that most users don't approach their pro models demandingly enough, right? I get in kind of fights with my pro models all the time. I'm like, don't give me tops of waves. Let me reenter the skills prompt, right? I always want you to be telling me something that I forgot to ask or that some really obscure person published about, but that has very low weight in your trading data, so it's hard to see.

B

So if you were talking to professors about this, you, I'm assuming I'm just translating from what you've just said, would say, don't get on play around with Claude or ChatGPT and then say, well, I understand this technology, but with greater exploration, you find that the use cases are much better than they might see with just a casual tour of these tools.

A

100%. And again, let me go back to this question. I had a, I had a big piece a couple of weeks ago to say, what should an AI czar do? Right? This is part of what an AI czar should do at a major university or a small university is to say, you know, you actually, you know, you have to be, you have to be on this thing for a couple hours every day before breakfast, preferably, right? So that you understand the tool and that you understand the new thinking of the tool as it emerges, right? So I probably am on my pro model a couple hours a day, and on top of that, another couple hours reading edge cases of how this technology can be useful that I didn't know about, right? And so when I just ran across this Gary Tan tweet the other day, you know, I'm thinking like, this is April 2026, and this is kind of use 101. But that tells me that even in his world, right. Users are not being sophisticated. And I think he made some sort of example. It's like, you know, having a car engine on a bench and saying, now drive me somewhere. Well, you kind of have to have the wheels attached and the steering wheel and the whole thing. You, you have to give it where it wants, where you want it to go, then it will go there.

B

And so what is the, the way that this Filters down through universities. It's not simple. You could say, well, my dean did it. But that doesn't really solve the problem for the university. The university president probably got to do some fundraising rather than tackle this. I mean, what is the mechanism by which. And I gotta admit, a little personal reflection here, I'm sort of loathe to add yet another administrative layer to most universities, but what's the answer here? If it's not just plug and play, how do we get from a powerful tool that needs to be not only sort of adjusted and understood, but then you have to teach that to a bunch of folks before this can be used productively. Any thoughts on how that should work?

A

Well, so first of all, thank you. That is the nut that I'm trying to crack that people are trying to crack how? And I saw, I went to a talk at Brown University on Saturday. You know, I went down to Brown to, went over to Brown to see this talk on how Chinese universities are doing this. Right? Because, you know, getting ready to sort of think about. And because, you know, China has a different form of governments and can mandate, but not mandate. You must learn this, that, you know, in communities to say, what is this problem that you as a community want to solve? Let's say it's cleaning up a river, or let's say it's building a new bridge over a gorge or what have you, is to have the community understand how AI can help solve that problem as a community. Right. So that people are coming together with a personal incentive to learn the tool outside the classroom. We have nothing like that going on in the United States right now. And, you know, that's okay. You know, we're a different kind of company country. But, you know, I was doing a certain kind of work as dean. I stepped down as dean to focus on this. I don't, I'm, I haven't, I don't know what is, what is being happened, what is happening at Utah now. But I know that at most universities, this conversation, that is the central question to the survival of higher education in the AI era, this conversation is not being had.

B

We've talked a lot about content delivery and course structures and the role of the professor, but not a lot about judgment. And judgment is different. And it's something that we expect from higher education, the development of judgment. How do you see judgment fitting into this constellation?

A

That's also a great question. Right. You know, judgment should have been always central to higher education before this. Right. You know, and, you know, part of the great critique of higher Education since William F. Buckley through Peter Thiel. All the things the coddled mind fire heterodox is that judgment has been sort of co opted as political judgment. Right. When you take away that kind of question, you realize that nothing has replaced that question of where is the centrality of judgment? And for a lot of different reasons. What I would consider mostly the bureaucratization and the metricization. You know, judgment has stopped being a conversation. So I'm so glad you asked that question because it's human judgment. Last mile judgment, Thoughtful critique can once again be a central part of our conversation about higher ed.

B

Do you worry that if AI is implemented not in an ideal way, not in a terrible way, but just in sort of your modal expectation that it will crowd out a focus on judgment in higher education?

A

Well, I guess the real question is, you know, what does. I think judgment will come in the back end. So, for example, I talked about Brendan McCord and Cosmos and the work they're doing, and I had a great meeting with some Liberty Fund people and a seminar on these questions a couple weeks ago back in Austin. And the conversation was about returning to the idea of formation. Now, Covenant is a Catholic school.

B

Is that Presbyterian? Presbyterian.

A

Okay. But interested in formation. Right. Moral and intellectual formation, which is a form of judgment. Right? Right. The entire curriculum and the institution was to form you as a person, which is again, what institutions of higher ed should do. Either form you morally or form you intellectually or some combination thereof. This isn't happening right now. You know this, I know this. I think if we are able to rid the bureaucracy, come up with new institutions that are a combination of us, St. John's in a covenant with. With AI. The human part of this will be the judgment, and AI will be in service of strictly information delivery. I think it's a. It's a beautiful model.

B

Hollis. I hate to go from the. The vaunted to the pedestrian, but actually many go to college because they think it'll help them get a job.

A

Sure.

B

It may be less easy to understand now what jobs of the future will actually be available given the changes that AI is producing. But how do you think universities should deal with this uncertainty of preparing students for what's probably going to be a pretty unpredictable job market?

A

Well, I would separate out first the question of the knowledge economy from other kinds of the economy. So the knowledge economy, if you're just moving information from one screen to another and preparing people to move from one screen to another, there's no future in that. Right. Or anything that is What I would consider sort of, you know, HR or diversity or like, you know, sociology, a lot of sociology. This group of people thinks about this in this way because one of the secrets of AI is it can do sociology better than most people, which is interesting. Again, I don't think there's a future there. So then, you know, again, I don't see this conversation happening at the very top in ways that filter down. I am telling anybody who's on the fence about going to college to go become a plumber, an electrician or an elevator repair person unless they know exactly what they want to do in college. Right. Because we need electricians, we need builders, we need H Vac people, and we really need elevator repair person. Now, people who are worried about class. Right. And this is what, this is, the, the sort of quiet conversation about somebody who doesn't want their kid not to go to college is that there are really blossoming a lot of humanities, you know, groups, reading groups, things like Xena Hits's, the Catherine project, you know, wonderful Internet reading, humanities, history, kinds of ecosystems that you can belong to and become a wonderfully well read person without going to college. And I think getting rid of the stigma of not going to college is something that we can all work on.

B

Hollis, if a student is coming to you, they're 17, they're rounding out their high school career and they're determined to go to college. It's not a question of if, it's where. What's the advice you give them in this context?

A

Get into the best, most affordable small liberal arts college that you can get into.

B

And just when they say, well, why would I do that? There's these large, better known colleges, the exchange value of their degrees may be much better.

A

Sure. And if they're in Utah, I would say go to University of Utah and get into the honors program. Right. Get into the honors program that are small classes, high touch. Do not take any online asynchronous classes because I think very quickly employers are going to say, I don't trust it.

B

So final question, especially in that context, in the long run, do you think AI is going to improve American higher education 100%?

A

I think it absolutely will. And you know, let me also say I should, will tell a student, if you, if you, you know, if you can go to a flagship that has a football team, go to the flagship that has a football team because it will be most likely to survive, because football will make sure it survives. And the big flagships are going to be picking up the lifeboats of all the colleges that are going to close in the next few years, you know, for the better or worse. There's a lot to be said that football schools will likely stay the same, even in the AI era. However, in most places in higher end, AI is going to be extremely disruptive and what will come out the other side is going to be so much of an improvement.

B

Thanks for listening to the Report Card with Nat Malkus. And special thanks to today's guest, Hollis Robbins. Also, I want to thank our producer, Ellie Lucas. He makes this podcast possible. That's all for this episode. I'm Nat Malkus.

A

Sam.