Paul Ford: Hello, I’m Paul Ford.

Rich Ziade: And I’m Rich Ziade.

Paul: And this is Reqless, R-E-Q-L-E-S-S, the podcast about how AI is changing the entire world of software. Top to bottom. No stop. What can you do?

Rich: It’s not just changing software. It’s changing the world.

Paul: Changing the way that people talk and think and educate, which is what we’re going to talk about today with our very, very special guest, Clay Shirky. I’m going to introduce him in a minute, but we gotta play our theme song first.

Rich: Let’s do it.

[intro music]

Paul: Clay.

Clay Shirky: Paul.

Paul: Welcome to Reqless.

Clay: Thank you.

Paul: I’m going to tell the people your job, and you tell me if I get it right.

Clay: Okay.

Paul: You are the vice provost—we’re going to talk about what a provost is—of educational technology and AI, artificial intelligence, in case people don’t know, at New York University, NYU, in New York City.

Clay: Yes. Correct.

Paul: And other—there’s many places where NYU is. It’s a brand.

Clay: Right.

Paul: So, okay, welcome. Thank you for coming on.

Clay: Thanks for having me.

Rich: Clay, take a minute and tell everyone about your life and world before NYU. I actually ran into your names because I enjoyed your writing a lot before NYU, but I think it’s worth it to take a minute to do that.

Clay: So I had one of those came across the web in the early 90s kind of random careers. I moved to New York to go into the theater. I was a lighting designer. I worked with a company company called the Wooster Group downtown. Did a bunch of work with theater and dance companies, and then directing my own theater company, we were doing a bunch of research, and I was telling my mother I was doing this research, she said, “Oh, you ought to learn about this thing we’re learning about in library school”—where she was enrolled at the time—”it’s called the internet.” And I was like, “Okay, ma. I’ll check it out.” So this is like ’92.

Rich: Okay. That’s early.

Clay: And… It was early.

Rich: Very early.

Clay: And I was like, oh, this is really interesting. And after a few more weeks, I was like, this is more interesting than the thing I was doing previously. [laughter] So I then jumped into, you know, whatever, teaching myself enough Perl to be dangerous.

Paul: It’s the opposite of everybody—everybody else had to explain their, the internet to their parents.

Clay: Right. Exactly.

Paul: Yeah. [laughter]

Clay: Right. So I never fell for the kind of mirror-shade, cyber-cool thing because the internet was the thing my mom knew about.

Paul: Yeah.

Clay: And was like, and told me about. It’s like, “Okay, I get this.”

Paul: “And clean your room!”

Clay: Exactly. And by the way, literally the first email I got was from my mother, and she wanted to know, were my shirts clean? [laughter] So I’ve never had a sense that the internet was this kind of thing apart from real life, which kind of helped, you know, thinking about it.

Paul: I have heard over 2 billion internet origin stories, and this is actually complete—I’ve never heard this one before. [laughter]

Clay: Yeah, it was an unusual one. And look, it saved me because, I mean, you may remember, I don’t know, Rich, where you were in the early ’90s, but there were two internet providers in town. There was one that was sort of devoted to cyberpunk aesthetics. And there was another that was cranky UNIX system administrators hanging out when they weren’t working for banks.

Paul: Choose wisely.

Clay: I ended up with the UNIX systems administrators.

Paul: There we go.

Clay: Those are some cranky motherfuckers. But they taught me so much. It was Panix, was the service. I learned so much in just a couple of years. Worked in the web industry. One of the people I worked with, Sam Ewen, his dad ran Media Studies at Hunter College. They were looking for somebody to teach web technologies. So in ’98, I end up at Hunter, which I loved. The undergraduate program in particular was amazing. Got recruited by Red Burns to NYU in early 2001. Red was the founder of the Interactive Telecommunications Program.

Paul: I don’t think most people will know who she was.

Clay: I know.

Paul: Just take, like, one minute for the audience to understand this human being.

Clay: So Red was somebody who made things happen. And she, at a time when NYU had lots of real estate and no money, went to NYU and said, “You need someplace that concentrates on media. Not television and film, but media.” She had seen the Sony Handycam and closed-circuit and cable television and thought, something’s coming. And so she started a thing called the Alternative Media Center that was later rebranded ITP, Interactive Telecommunications Program, which is still going at NYU. And Red’s genius was to understand that the technology was enabling stuff, but she was never interested in the technology herself. And so it kept…

Rich: She saw disruption.

Clay: Yeah.

Rich: Change happening.

Clay: Kept the department focused on, you know, we jokingly later called it the Center for the Study of the Recently Possible, right? [laughter] It kept the department focused on, what’s the new stuff that just showed up? So they invested hugely in CD-ROMs. That was the first time I’d seen them. And then they pivoted to the web within two years, because a lot of places were like, “Oh, but we have all this expensive equipment for burning CD-ROMs.” Red’s like, “I get it. It’s over. We’re done.” And they—

Rich: Yup.

Clay: —started teaching web classes. It was incredible.

Paul: This is a woman in her 60s, too.

Clay: Yeah, yeah, no, no, she—

Paul: This was wild. Like, this…

Rich: Interesting.

Clay: This was her second—like, this is her, actually, may have been her fifth career for all I know, but this is a late career change for her, and she was a force of nature. She was incredible. And she passed away in, I think, 2013. But I started there in 20—so I worked for her for a dozen years. And that was, that was amazing. So then, you know, juggled work in the web industry and at ITP, got interested in social media early on. We did a big, big conference on social media in 2002. We called it social computing in those days. But ended up writing a couple of books about it, and then moved to Shanghai to help NYU open the campus there, because that seemed like an adventure. And when I came back, I joined the provost’s office. And the provost is just, it’s a fancy way of saying chief academic officer. And so I’m a vice provost, which is an academic vice president. I work on any place where anything digital touches the classroom.

Paul: Tell me about a normal day in a vice provost of educational technology—

Clay: Well, we have to go back to before January 20th to find a normal day.

Paul: Yeah, okay.

Clay: I will tell you, given what the last three weeks have been like. But, you know, some of it is very advanced planning. We’re looking at sort of are there new online education efforts we can invest in? Some of it is short-term stuff, right? We’re working with New York state on trying to get some things approved. And then some of it is very short term. Like, somebody’s got either a research request or some kind of emergency and they want to talk about how they fix it. So I spend a lot of time talking to faculty, and I spend a lot of time talking to IT, and I spend a lot of time translating between those two groups.

Paul: So meetings and email.

Clay: Meetings and emails. I have a fake job.

Paul: Okay. No, I mean, we all do. It’s okay.

Rich: Things sound normal. Big institute. NYU. It’s a big organization.

Clay: Yep.

Rich: Big, big university.

Paul: It’s about 100,000 grad and undergrad—

Clay: Not quite. It’s 65—

Paul: Ah, okay.

Clay: And then there’s probably another 10,000 employees. So.

Rich: Okay.

Clay: Medium-sized Midwestern city.

Paul: Yes.

Rich: Yeah, but big operation.

Clay: Yep.

Rich: You’re seeing the swirl of education and IT and technology kind of dance. And then a new kid shows up. Tell us when you realized that, oh boy, this isn’t just like another line item, but something different.

Clay: Right. So, I mean, look, it was right after Thanksgiving in ’22 when ChatGPT drops. I’d already been looking at some earlier uses of large language models. There was, there were some students at ITP who were working with them, like, it was kind of in the air. But it was all very much, it was all very much in the kind of, this is a toy. Maybe we could do something interesting with it. People were using it to, you know, DM D&D games, because if, you know, something appears out of nowhere, it isn’t that much of a surprise in that context. You would never use it to write a business memo. And then GPT 3.5 drops and suddenly it’s producing competent college-level paragraphs on all kinds of information.

Paul: Well, and NYU students would never, but other schools. [laughter]

Clay: Well, now look, I’ll tell you. In fact, most of my time in the last three months has been trying to figure out under what circumstances can students be persuaded not to use the tools because it’s bad for learning.

Paul: Yeah.

Clay: And the answer is that the upper limit so far of that persuasion is lower than faculty had hoped. I’ve been to a number of meetings in the last three months where the dominant emotion was sadness.

Paul: I mean, I don’t think NYU—we’ve talked to people from other orgs, and NYU is in no way alone in this.

Clay: Right.

Paul: Students like to use digital tools.

Clay: Yup.

Paul: In some ways, there’s a couple analogies from the past that… I’m actually going to, let me throw you just a tiny thought experiment to see what you make of it, which is, wait a minute, we did this with Wikipedia.

Clay: Yup.

Paul: Wikipedia was edited by amateurs. And then many products came around that were going to make Wikipedia smarter. There were more scholarly versions. Google came out with KNOL, which no one remembers, but Google literally—

Clay: Leave the gun, take the KNOL, yeah.

Paul: Yeah, we’re going to make our— [laughing] Exactly, take the KNOL. Google thought they were going to have a more verified Wikipedia, etc.

Clay: Yup.

Paul: And it just turned out that everybody just kind of used the thing and we all got used to it and it was all okay.

Clay: Yup.

Paul: I don’t actually buy this analogy, but I’m going to throw it at you, because we did have the exact same panic in an academic context with that.

Clay: Yep. Yep. Yep. And we had the, we had the K-12 panic with the personal calculator. There are a number of these things. I’d say the difference is between inputs and outputs.

Paul: Okay.

Clay: If I was asked for a bunch of facts, I could go to Wikipedia, get those facts, put them in a paper, and send it off to the professor and get it graded. And I’d done less work to get the facts, but I integrated it into something. Here I’m actually able to take ChatGPT and make it look like a version of me.

Paul: Yeah.

Clay: Right? So less goes through the student’s head with ChatGPT than went through the student’s head using Wikipedia. Now, if you cut-and-pasted from Wikipedia, that was just straight plagiarism. That was a different, different problem. Here the issue is that there’s no plagiarism detector, because every bit of output is unique, and so you don’t get that, don’t get that signature. There are a bunch of people trying to sell plagiarism detectors, but they don’t work well enough to use.

And so we’re in a world where we’ve got this weird binary now that we didn’t have with Wikipedia, which is if the professor doesn’t see the student doing the work, they can assume it’s a human-machine hybrid. And if the professor wants an assessment of what the student has learned rather than what the student is capable of producing, that work has to come back into the classroom. Just today, Harvard announced that its math classes are going to in-class tests rather than take home tests.

Paul: Harvard! I’m shocked.

Clay: So—people have, whatever, it has nothing to do with, I mean, it’s not about the students. It’s about if an institution with that many resources is like, “Yeah, we got nothing. Come back in the class and get out your pencils,” then the sort of store of adaptive answers is less than we had hoped.

Paul: I am going to derail in a very interesting way. Get ready. I think the problem is the outputs and the product work that’s being done by the AI companies more than it is the students. Let me give you an example. OpenAI just came out with a new researcher that does an amazing job.

Clay: Yeah.

Paul: Puts citations in. It’s essentially your faculty’s worst nightmare.

Clay: Yup.

Paul: It’s better than the students. It’s better than an undergrad. You can say, “Write me a paper on blah blah blah,” and it does a great job—with footnotes. It’s everything we’ve said.

Rich: I’ve seen it.

Paul: Okay? So my problem with that is it keeps doing research papers, which is the wrong form for a computer to produce. A computer should say, “I went and I found… I did the summary for you. I went and helped you with your research.” Okay, yeah, maybe they should go to the library and pull books, but that’s not how the future is going to work. Let it go. Gather a bunch of stuff. Summarize and point out where connections might occur. Let them be visualized and let people explore the data space around the problem using all the things a computer can do. And then let them have to make something new out of that.

The problem is we’re in a world where everyone’s decided that this particular kind of paper is the artifact of learning. I don’t think that’s true for the future. I say this as someone who likes papers, was good at them, and built a career around essentially writing research papers. So, like, I don’t, it doesn’t give me joy. It’s just, like, I don’t think that form is the right focus and we’re all oriented around it.

Clay: Right. And look, from our point of view, we have this battle between the intrinsic motivation, like, “I took this class because I like this subject and I want to learn some stuff,” and the extrinsic motivation of, “You’re going to get a grade and it’s going to affect what you can major in and your GPA and blah blah blah.” And the extrinsic stuff has been ramped up by both anxiety about employment after college and by the fact that the students are often working nearly full time jobs, or full time jobs. And the trade-off from the student’s point of view is I can use this tool that’s, as you say, optimized to produce something that basically looks like the output the faculty member is asking for.

Paul: And it’s never going to go away. It’s always going to be here in my future.

Clay: Right, right.

Paul: So if I ever need to do this again, I can.

Clay: There is a Reddit thread that we all went to town on about a month ago, month and a half ago, an NYU student said, “Help, I can’t stop using AI.” They were a school—

Rich: A cry for help.

Clay: They were a student enrolled, they said, in their senior year in our engineering school, saying, “I’m using this thing all the time. I recognize that in my own major I’m learning less than I would if I wasn’t using it. But I can’t stop.” It was that sense of addiction that really started to change—and make me more pessimistic, frankly, about the kind of progressive, like, we’ll just appeal to the student’s natural love of learning.

Rich: Mmm. Mmm.

Clay: It’s like, if this is a student who is in the part of her education where she’s only taking classes in her major and by her, you know, at least by what she said in the, in the Reddit thread was interested in them.

Rich: Yeah.

Clay: And then was saying, “I actually can’t stop using this thing that I recognize is interfering with what I’m learning.”

Rich: Let’s actually probe that statement for a second.

Clay: Mmm hmm.

Rich: I can read it a few different ways. One way I can read it is, “I can’t stop using it because it’s easier, and I get my work done and I get to go out with my friends.”

Clay: Mmm hmm.

Rich: So it’s “addictive to me,” because what usually takes four hours has taken me an hour, and then I went and had dinner with friends. Okay? That’s one lens to see it through. Another lens to see it through is, “I don’t understand—I need it. I don’t understand the work. And I don’t guess I don’t have to because I could just punch it in and submit it through from whatever this thing is spitting out.” A third is, “I can’t stop because I love its output and I’m learning five times as much. It’s so glorious, because it writes really well, and I’m learning even more than I used to in the past.” [laughter]

Paul: I can simplify this for you. You ever asked a developer to write documentation?

Rich: Why are you going there, Paul Ford?

Paul: You ever seen their, like, their face, when you say, “Hey—”

Clay: You’re bumming out Rich’s party head, you can see it.

Rich: [laughing] Yeah, yeah.

Paul: “The docs aren’t done here, buddy.”

Rich: Yup, yup.

Paul: And they go, “Oh—” And then you know what, you know what happens when you as a manager tell a developer to write documentation?

Rich: Yeah. Yeah.

Paul: You hire a documentation professional because they’ll never do it.

Rich: Yeah.

Paul: Or they’ll do a really cursory job, because it just feels like moving backwards.

Clay: Don’t you understand that literate code is what we need? Yeah, yeah.

Rich: Yeah.

Paul: Humans cannot move backwards once they have felt a taste of the future.

Clay: This is—well, I don’t even know if it’s the future. But what, it’s difficult to make things hard on yourself when you could make things easy on yourself.

Paul: Have you ever seen a millennial use a phone? You might as well ask them to, like, strangle a puppy.

Rich: Or an email write. Like, writing emails is strange because messaging—

Clay: No, a lot of people, a lot of students that we hear from, the number-one use that they rush to isn’t papers, which are relatively infrequent in their life. It’s things like resumes, cover letters, writing emails.

Rich: Sure.

Clay: It’s semi-structured stuff where they’re anxious about, “Do I have the full format right?”

Rich: Oh, interesting.

Clay: I was, I was talking to—so we have, we have a 15 campuses, we’re operating on six continents. So I go to other campuses and talk to students. I was at the LA campus and a bunch of students get internships, unsurprisingly, with film producers. And there’s this kind of standard format for an idea for a film. It’s basically a structured PowerPoint deck. Everybody in LA knows how to use it. You roll up, you’re 20 years old, you get hired as an intern, and somebody just says, “Oh, yeah, make me a pitch deck for X,” or, “Take a look at this pitch deck,” or, “Find clip art for this pitch deck.”

Rich: Yeah, yeah.

Clay: And you’re like, “What the fuck is a pitch deck?”

Rich: Yeah.

Paul: Right.

Clay: And you were definitely not going to the person you’re trying to impress to say, “I’m sorry, I’ve never heard this word before.”

Paul: A huge part of socializing yourself into an industry is knowing the forms.

Clay: Right, right, right. Exactly.

Paul: Lawyers have to do this, everybody has to do this.

Clay: Exactly. And so those students just go to ChatGPT and are like, “What’s a pitch deck? Make me a pitch deck. Find me some clip art for a pitch deck. Or make some art for a pitch deck.” And all of a sudden they bring this in, they hand it off, and the producer goes, “Oh yeah, this is really good. Thanks.”

Rich: Yeah.

Clay: Like, the transaction with AI is faster, better, and lower anxiety.

Rich: Yeah.

Clay: The thing, though, that you said, Rich, I want to go back to your third option, which is, I’m learning so much. I want to say my thesis is 3A, and I got this from Joss Fong, the documentarian. People love this because it feels like learning. It’s not actually learning, but it feels like it. And so you think, “I have a problem, and I’m gonna think about the problem for a little bit, and then all of a sudden I have a great answer,” that feels like you learned something. But what you learned was, “This is how I asked ChatGPT about this problem.” Now, if what you need to be learning is just, “How do I use this tool effectively?” That’s okay.

Rich: Yeah.

Clay: But if what you need to be learning is, “What does a good idea for a movie feel like?” This is less effective.

Rich: Yeah.

Paul: You know, a funny thing, as someone who’s looked at probably thousands of different manuscripts written by people at this point, and blog posts and keep going, right? Like just, I can kind of tell just by the texture, by squinting without reading, if something involves a certain amount of thought or not. There’s a little jaggedness to it. And everything that ChatGPT puts out is very symmetrical. It’s very smooth. And it actually reads as, like, it’s just sort of glossy. It’s like an 80s synth sound. Like, just kind of—

Rich: Yeah.

Paul: And I think, like—

Rich: Oh, don’t worry, the jagged filter for output is coming. [laughing]

Paul: I don’t I don’t think—I swear to God, I don’t think anybody wants it.

Rich: You click on the number of beers.

Paul: Yeah. Nobody wants that.

Rich: And it revises the text.

Paul: I don’t think you’re going to get to a point where humans—nobody knows to ask for that. That’s an artifact of humans.

Rich: You just said it on a podcast.

Clay: Yeah, I was going to say Rich is going to, Rich is going to make it.

Rich: [laughing] Yeah.

Paul: Well, let’s—

Clay: So you’ve got a dial from three espressos up to three beers and any place in between. Like, give me a writing style—

Rich: Yeah, it’s a spectrum.

Paul: Let me draw a line down the middle of this, right? So you’ve got faculty who are figuring it out, students who really want to use it, and we’re learning from Harvard and other second-rate institutes of knowledge that this is everywhere. It’s in every community college. I’m assuming it’s in every—

Clay: It’s everywhere, yeah.

Paul: High school, every big college, every small college. I’m not asking you to even make a prediction, but I’m just, like, there’s gonna just be more and more. Do everybody a favor. Sit down the frustrated professor and tell them what to do for the next couple years. Here I am. I teach anthropology, Clay.

Clay: Yup, yup.

Paul: And I’m hitting my wall here. They’re using Claude, and I don’t even know what that is. I went in and used it. It does write pretty good. I don’t know why I talk like this. I have a PhD. [laughter] But could you just help me out here, man? You’re good with computers. And my printer doesn’t work. [laughing] So could you just—

Clay: Yeah, right. And by the way, my printer doesn’t work. Yes. Always part of any conversation with people, understand—

Rich: I mean, this has got to be happening.

Clay: Yeah, no, no, look, I mean, when you asked earlier about what’s, what’s the average day, so much of my day is framing exactly these communications.

Paul: All right, let’s bundle this up. We can put it in the podcast and then you can just say, “Listen to this instead of bugging me all day.” [laughter]

Clay: Yeah, you’ve caught me at a moment where I, I can’t answer that question easily because our old theory collapsed last fall and we’re slowly trying to express the new theory. The old theory was uses of AI are on a spectrum from lazy to engaged. Right? Learning inhibiting to learning, learning enhancing. And we thought the good uses will crowd out the bad uses. Right? If you eat carrots, you will eat fewer potato chips. So we’ll introduce the students, we’ll coach the faculty to talk to the students about the kinds of effort that actually you learn from. And at the end of the semester—and this is where I was, you know, the aura of sadness came into meetings with the faculty—faculty said, “Some of my best students were in a class where I talked to them about this, and they didn’t stop. They didn’t say, okay, if I’m going to learn, then I shouldn’t use these tools.”

Paul: It’s that Reddit thread again.

Clay: And I got—exactly. “And I got AI slop.” One faculty member said, “I get AI slop. I did what you said. I didn’t, you know, I didn’t yell at the student. I just sent it back and said, this isn’t good.”

Rich: Yeah.

Clay: “Please rewrite this. But, you know, rewrite it in your own words. Here are the mistakes.” And the faculty member got back a point-by-point email rebutting some of the things that they said.

Rich: Oh no!

Paul: [deep sigh]

Clay: You see where this is going?

Rich: Yeah.

Clay: And she looked at the email and said, “You just used artificial intelligence to respond to me about a mail where I said, this is not good for you.”

Paul: Clay, as a large language model, I reject that implication.

Clay: Yes. [laughing] As a large language model, this is out of my corpus. But she wrote back and she said, “Look, this conversation’s over. I’m not talking to a machine.”

Rich: Yeah.

Clay: And so what we’ve realized now is the good uses and the bad uses are completely orthogonal.

Rich: Yeah.

Clay: A student can do either or both.

Rich: Yeah.

Clay: So what we’d hoped is that if you introduce the good uses, you’d lower the bad uses. But there’s this class of, I shouldn’t say students. It’s this class of reactions to assignments that we’re starting to call omnivorous, in which you do the learning enhancing stuff, I generate some ideas, whatever. And you still use AI to write the paper.

Rich: Mmm.

Clay: And it is very difficult to students to say, “This assignment is only good if it’s hard for you to do.” Right? Something that makes it easier to do your work is less useful to you. Because our output is the paper. Our output is the student work. But our product is the student.

Rich: Yeah.

Clay: And we don’t say that a lot. It’s been kind of implicit, and now we’re having to be more explicit about it.

Paul: You know, what’s wild here is really what we’re talking about is risk. Okay? Like, you are asking them, when you’re asking them to do that, you’re asking them to take an exponentially greater risk of failure in an expensive academic program that defines their future.

Clay: Yup.

Paul: And they have every reason to be scared.

Clay: Yup.

Paul: Right? Which is actually part of the process. Like, we’ve all been through it. We all have degrees.

Clay: Yeah. Yeah.

Paul: Right? And so we, we see the value of that—and we’re parents.

Clay: Right.

Paul: But I think when you’re 18, 19, you’re just, like, “Just tell me what I gotta do, man.”

Clay: Yep. Yep. Yep.

Paul: And so that tension is gonna be really, really hard to unlock when you have the ultimate risk reducer available in your pocket.

Clay: This is—yeah, this is exactly right.

Rich: When you say “risk reducer,” you mean the student’s anxiety can go way down because he can rely on this thing that’s pretty good at whatever.

Paul: It’s gonna do what the man asked you to do.

Rich: Can I, can we have a different roleplay? Can I be a different professor?

Clay: Sure.

Rich: All right. I’m coming into your office. I’m wearing weird shoes. I’m eccentric. Clay, what’s up, man?

Clay: Hey.

Rich: Everybody’s such a bummer. I think this thing is great.

Clay: Yup.

Rich: I think it’s absolutely great. Look, it’s put a little more work on me. I still got to challenge these kids somehow and figure that part out. But they’ve got an incredible tool. They’re learning stuff. My assignments have changed. I’ve tweaked them so that if they’re more exploratory, and I make them come and present, and share new ideas found in the corners where I’m less directing them down a curriculum and more telling them to go explore. Use whatever tools you want. I don’t really care. But when you’re coming back into the classroom and when they communicate with me—

Paul: Man, you really are a professor because you’re not asking your question, you’re just talking.

Clay: More of a comment than a question. I’m used to it. [laughter] This is how higher ed works. Go on.

Rich: I’m telling them to come back and present and be creative and show me what you learned.

Clay: Yep.

Rich: You got to talk to me. Sometimes I’ll ask you to write stuff. Isn’t this great?

Clay: So it’s great for professors for whom it’s great. Right? I mean, so, yes, there are professors, and we have people who, except for the weird shoes, act exactly the way you just did, which is to say—

Paul: Yeah, but you’ve got a big battleship to turn around.

Clay: That’s the thing. So I’ve got a journalism professor colleague of mine in the journalism department, where I also have an appointment, who has had this attitude from day one of, like, “Look, my students are producing written output. They’re always going to do the last edit, but I’m going to teach them how to use these tools from the beginning.” And he’s been advocating everybody should transform and teach this way.

Rich: Yeah, yeah.

Clay: I’ve got biology professors who are saying, “The students need to understand the two kinds of cell division. Like, they don’t need to go to a machine that tells them about meiosis and mitosis. They need to have it in their heads.”

Rich: Yeah.

Clay: And so there’s absolutely a spectrum, and there are some faculty who do what you’re suggesting. But I will also say that you’re, like, “You got to come talk to me. You got to come present.” That is the medieval turn that we’re all going through right now. Higher education was, for hundreds of years, a principally oral culture.

Rich: Yeah.

Clay: Right? I would read—the “lecture” was really literally me reading from Aristotle or whatever.

Rich: Out loud.

Clay: Out loud. And the students were sitting. And when the printing press came along, hilariously, at the University of Paris, they told professors, “You should speak fast enough that the students can’t write down what you’re saying so that it doesn’t get printed.” [laughter] Right? There was this terror of the printing press sucking the life out of the lecture.

Paul: I mean, the lecture system still goes on in the UK, too.

Clay: Yeah. And look, it still goes on at NYU. But the thing about the turn to writing as the standard college output takes a few hundred years to kind of congeal after the invention of the printing press, but really gets a boost in the United States after the Second World War, when everything scales up from first the GI Bill and the Baby Boom.

Rich: It’s just raw numbers.

Clay: And so you need some asynchronous way of assessing students.

Rich: Yeah, yeah.

Clay: It used to be much more oral presentation. Let’s, you know, see what, you know.

Rich: Sure.

Paul: Big middle class.

Clay: Right, right. And so part of the problem we have with doubling back on, “You got to come talk to me. You got to come present in class,” which does tell you whether the student knows what they’re talking about or not, is it gives up the principal way we scaled up between 1950 and 1980, which is the hyper-growth of American higher education.

Rich: Yeah.

Clay: So small liberal arts colleges, right, if you’re Swarthmore, if you’re Haverford, right, you’re laughing. Your classes are already set up that way. If you were the University of Missouri and you’ve got a lecture hall that holds 500 people, and you’re suddenly thinking, “I can’t talk to 500 students.”

Rich: Sure.

Clay: “How do I assess what they know?”

Rich: Right, right.

Clay: So it’s, there are, there are some faculty who have adapted the way you just said, Rich. There are some faculty who could adapt that way but haven’t. But there are also some faculty who can’t adapt that way.

Rich: Sure.

Clay: Either because of their subject or because of scale.

Rich: Yeah.

Paul: Bio 101.

Clay: Right, right.

Paul: At Penn State.

Clay: The opening—right. The opening maw of pre-med and calculus are the two, like, “We just need everybody to learn the same thing.” I think our calculus program at NYU enrolls more people than are enrolled at Dartmouth, full stop. Right? It’s just this juggernaut.

Rich: Giant.

Clay: Right.

Rich: Giant.

Clay: Right. Because so many other majors rely on those students knowing calculus. It’s a service department for the—it’s a service class for the math department, because nobody who’s studying math at NYU needs calculus when they come in. Right? So this is entirely for others schools, and it’s massive.

Rich: Interesting.

Clay: And so the question of can you have a take-home exam? And the answer has gone from yes to no in two years.

Rich: Is that right? Is that, that’s dead.

Clay: It’s so—

Rich: Kind of impossible.

Clay: It’s kind of impossible. And it’s, like, think about programming. Right? I mean, the programming people, as often when a new technology comes along, the people who teach CS are the furthest along. They look at this thing and they think, by the time someone is graduating with a degree in CS they need to know how to use a copilot. Right? Anyone who’s going to hire them is going to expect this. But when they’re learning for loops and variables, they can’t have these tools, because they have to integrate it.

Rich: Yeah.

Clay: And it’s that—

Rich: They have to file it away, instinctively.

Clay: It’s that changeover from—

Rich: Yeah.

Clay: We don’t, just because of the calculator, we don’t stop teaching kids times tables anymore. But we do stop teaching them how to multiply three-digit numbers in their head or do long division on paper.

Rich: Right. You want them to get the fundamentals internalized.

Clay: Right. Most fields have not yet had to identify “these are the fundamentals.” Because we haven’t had tools that can do this kind of heavy lift.

Rich: You haven’t had to.

Clay: Right.

Rich: You’d had to learn them.

Clay: Right, exactly.

Rich: And the learning process—

Clay: Right.

Rich: The gift is inside of the learning process.

Clay: So there’s an interesting test case—I was just actually talking with a student about it this morning. We have a, my colleague De Angela’s put together this student advisory council which is phenomenal, just hearing from students about AI. And we were talking about this. I was trading mail with one of these students this morning and there was a class at Georgetown in the School of Foreign Studies called “Map of the World.” Map—singular—of the world. And the test is, identify all of the countries and their capital cities and the major geologic features, rivers, mountains. And if I tell you that I’m trying to get something from Kinshasa to Brazzaville. If you’ve taken that class, you know that that is a short physical distance and a long political distance, right? And it’s just like that. [snaps] You don’t have to—anyone could get out their phone and look it up and say, “Oh, huh, huh. Two cities on opposite banks of a river, but in different countries.”

Paul: But it’s Georgetown. You’re probably going to work in the State Department.

Clay: Bingo.

Paul: And if you’re going to be acculturated, you got to have that in your head.

Clay: You have to have it in your head, not in your phone.

Paul: Yeah. Otherwise, if you got your—if somebody said, you know, “Kinsa to—” And you get out your phone, you’ve humiliated yourself.

Clay: Right, exactly right. If I hear “Kinshasa” and I’m my hand’s in my pocket, I’m, I’m already like—

Paul: Kinshasa. See, I humiliated myself just trying to pronounce it.

Clay: But it’s like, I’m not going to put you in front of any diplomats if you’re like that.

Paul: Anyone who’s tried to do that with me has really regretted it. All right, so let me, let me bring this all back around. Let’s summarize. We got—we kind of have. We have two situations. We have, “Lean in. You’re gonna have to figure it out because it’s not going anywhere. And whatever you do, you need to adapt and just, like, get to use these new tools,” which is the sort of like, positivist, Silicon Valley, here-we-go kind of mindset. The other side is, “You basically only can look at the students in the eye while their mouths move to make sure that they’re doing it. And God help you when smart goggles show up.”

Clay: Right.

Paul: So, like, it’s a sort of surveillance thing, and it’s, it’s kind of, it’s getting really perplexing for places where it’s a big broadcast style—

Rich: Well, it’s backwards.

Paul: What you’re saying, and I think this is an important thing to say, because everybody loves conclusions: There’s no conclusion between these poles.

Clay: Both of those things are true.

Paul: Yeah.

Rich: Yeah. It also sounds like you’re in the middle of it.

Clay: Right.

Rich: You’re still processing this seismic change.

Clay: Yep.

Paul: Well, and I actually genuinely appreciate you not having the answer. I think it’s really good.

Clay: [laughing] Bring me on any time then. If you want somebody who doesn’t have the answer…

Paul: No, this is such a know-it-all field.

Clay: Oh God, yeah.

Paul: Silicon Valley is a know-it-all place.

Clay: Right, right. 100%.

Paul: And, like, you know, we know a lot of people and a lot of really smart people, and no one has a truly credible theory of everything that’s changing, including us.

Clay: Right.

Paul: And we’re building a business on top of this stuff. So I think that it’s just important to get that out there, like, to the list—the listeners are sitting there going, like, “What am I going to do for my career?” Clay’s telling you, like, “Whew! Good luck!”

Clay: Yeah.

Paul: It’s still kind of coming together.

Rich: Clay, let me ask you one closing question.

Clay: Yep.

Rich: Does this leave you frustrated, optimistic, pessimistic? From a sentiment perspective, as you look ahead on the work here?

Clay: Frustrated described the first year of this. Described 2023, basically, where we could see things we needed to do and everything moved slowly. We’re past that. We’ve got a large language model deployed for everybody. We’re using Gemini. We’ve got NotebookLM deployed, which academics really like.

Paul: So every student has access to these tools as part of their…

Clay: Right. And this, I should say, that’s just this semester. So the future, the use of these things, we’re still observing. But we’re past the kind of, like, misfit between what the tools look like they could do for us and what we’re able to do inside the institution. I would say that what I am right now is I’m less frustrated because the answers seem.. The options seem clearer. But I’m also having to deliver news to more news to faculty that they don’t want to hear.

Rich: Mmm hmm.

Clay: Which is the ability to give a student an assignment, have them go off on their own, and be confident that the work required to do that assignment caused them to learn the subject matter is now deeply in question at best.

Rich: Yeah.

Clay: And in many cases, you can just write it off.

Rich: Yeah.

Clay: That just X’s out a whole range of, you know, assignment plus assessment. Faculty, you know, people who study learning talk a lot about assignments and assessments, but faculty often didn’t even think of those as distinct categories. Like, you did a paper, and I reacted to it. It was kind of one thing.

Rich: Yeah, yeah.

Clay: And having to say to faculty—I remember a conversation with a philosopher who came to a meeting with me, literally with his arms folded across his chest, like, “Oh yeah, this guy’s just come in from New York, and he’s gonna tell me I have to get with the program.” And he said, “I’m a philosopher. I’m trying to get these students to go from having a partial idea and actually work out the ramifications until they can express it. I don’t want them using this tool at any step. I don’t want them brainstorming. I don’t want them ideating. I don’t want them to use it at all.”

Rich: Yeah.

Clay: “We did what you said. You know, I talked to, I talked to the students at the beginning of class about this, and it didn’t work.”

Rich: Yeah.

Clay: And I think he was ready for me to give him a tongue lashing or tell him he wasn’t trying hard enough or whatever. And I was just like, “I get it. I’m sorry. I hear you.”

Rich: Yeah.

Clay: And—

Rich: You have been submitted the worst support ticket in the history of everything.

Clay: [laughing] And it’s—

Rich: What’s crazy is, this is a—

Clay: How do I fix this? Yeah, exactly.

Rich: Global problem.

Paul: We—

Rich: And people are going to Clay’s office.

Clay: Yeah. [laughter]

Paul: No, that’s right. And they’re like, “Hey, you destabilized the entire foundation of my multi-thousand year—”

Rich: “Fix the WiFi.”

Clay: And there are some people for whom the fact that I am simultaneously the person saying, “We need to put these tools in people’s hands to see how they use them,” mean that I become the proxy for the problem. I am the messenger—

Rich: They shoot the messenger.

Paul: You are technology.

Clay: I am technology. Including your busted printer.

Paul: No—

Rich: It’s the worst IT problem in history.

Paul: And your general positivity and enthusiasm only work against you. [laughter]

Clay: Yeah, that’s—

Paul: If you could just be like, “Yeah, it really sucks, man.”

Clay: Yeah, that is… Yeah.

Paul: But, you know, there’s an important thing I think to close on here is, like, we all know that it’s just going to be real hard to put this back in Pandora’s big box/jar. Like, it’s just, we’re not going to be able to escape this change. But you sort of do have to approach it with empathy. It’s a lot.

Clay: Exactly. Yeah. Yeah, that’s exactly right.

Rich: Clay, when you do have the breakthrough?

Clay: Yep?

Rich: Please come back.

Clay: Absolutely. [laughing] Thanks for having me.

Rich: Thank you, Clay.

Paul: How could people get in touch? How can they learn about you and your work? And, like, what would you have them go?

Clay: I mean, I’m such an inside dog now. I’m really like, I’m on LinkedIn as Clay Shirky. I’m on BlueSky as Clay Shirky. But mostly I’m in meetings and answering email at NYU, so.

Rich: [laughing] Okay.

Paul: Awesome. Thank you. This is really helpful.

Rich: This is great, Clay. Thanks again.

Clay: Great. Yeah, no, it’s fun.

Paul: So, Rich, you and I are building a product that makes it really easy to build just about any software solution using guess what foundational technology?

Rich: AI.

Paul: Artificial—

Rich: Sourced ethically. [laughter]

Paul: Very, very good stuff. Secure, private, all those—

Rich: All the nice things.

Paul: And we’re going to have a lot of big news really soon, but in the meantime, we’re going to keep telling the story and talking to really interesting people like Clay.

Rich: Yeah. Check us out at aboard.com, and if you’ve got topic suggestions or any questions at all, hello@aboard.com.

Paul: Yeah, we’re ready to Car Talk this up. So if you have any—if you want to talk about how to build something using this stuff, you let us know because we’re all in it. Hello@aboard.com.

Rich: Have a great week.

Paul: Back to work.

[outro music]