A couple of weeks ago (on September 8) I gave a talk at King’s College London, at the invitation of its Vice-Chancellor and President, Professor Shitij Kapur.
I was asked to be provocative, and I must confess that I didn’t pull my punches (the title was a bit of a giveaway: Paradise desecrated: Is AI destroying the university of our dreams? And if it is, would that be such a bad thing?). But the underlying message was that universities and academics have never been more important as we face one of the most disruptive technology transitions in recent history. But only if they wake up to the reality that old and entrenched habits and practices will serve neither universities, the people in them, nor the communities they are designed to serve, in an age of AI.
The article below draws on the transcript of the talk, but has been cleaned up and edited for clarity. It’s also — because my writing is never just writing — an exercise in exploring the capabilities of Anthropic’s new Opus 5.5 model. If you’re interested in the process, there’s more information in the postscript below.
The article is unapologetically long as I wanted to make sure the ideas I explored and the points I made were captured as faithfully as possible. This was, in part, so I had a record of them for my increasingly unreliable memory. But it’s also because I believe that academics and universities have a vital role to play in helping society navigate the AI transition, but that this won’t happen unless we are willing to openly and honestly grapple with who and what we are, and the value we bring to society, as AI upends so many of our long-held and cherished assumptions and beliefs.
And because it’s so long, do feel free to drop it into your favorite LLM for a summary 😊 (there’s an LLM-readable text version at https://text.futureofbeinghuman.com/substack/being-an-academic-in-an-age-of-ai.html)
Finally, because this is based on a transcript of a talk given without slides and with minimal notes, there are idiosyncrasies in the style here that come from it being based on a talk. I left most of these in. There may also be slight inaccuracies that have crept in, or points and ideas that are over- or underemphasized. I think that most of these have been caught. But it’s worth the disclaimer, just in case.
Paradise desecrated: Is AI destroying the university of our dreams? And if it is, would that be such a bad thing?
When I was preparing for this talk, I wasn’t quite sure where to take it, because the field of AI — where it’s going, and how it’s impacting society, universities and the academy — is moving so fast, and is so large, that I knew whatever I said would be out of date and irrelevant within a few days.
I must confess that I wasn’t entirely sure where the narrative was going either (although I had an idea). But that in itself is part of the challenge of the times we’re facing: there is so much uncertainty, so much novelty, so much speed, that even those of us whose work is grounded in intelligence and creativity, thought, and knowledge, find it incredibly hard to keep up.
So we’ll see how this goes.
I thought I’d start with a short anecdote — although even now, I’m still not entirely sure how it fits in. We’ll see at the end whether it does.
It’s an anecdote that goes back to a book I was reading this summer — and it’s a little embarrassing, because it’s one I should have read decades ago: Malcolm Bradbury’s Eating People Is Wrong.
Of course, it’s a standard book in the corpus. Reading it at this point in my career though, two or three interesting things stood out to me about it.
The first is that it was recommended to me back in 1981 by my then O-level English teacher, and I didn’t read it. It took me about 40-odd years to get there.
That in itself is relevant, because it demonstrates that, when it comes to teaching and learning, it isn’t immediate impacts that are important; sometimes these things take decades to percolate through, as this particular one did, leading me to go back and read the book.
The second thing that struck me as I read it was that the spirit of academia hasn’t changed that much since the 1950s, when the book was set.
If you’re not familiar with it, it’s a satirical campus novel about the struggles and the angst of a very earnest English professor trying to make sense of the academic world he’s in, and to align that with his morals, his principles and his ideas.
The title comes from a Flanders and Swann song (those of you who are old enough will remember it) about the reluctant cannibal who sits at the table saying “eating people is wrong,” but nothing ever changes.
And that in itself was a clever reflection on academia, where we sit around saying “This is wrong,” but nothing ever changes.
The third thing about the book that struck me was this idea of “paradise.”
I want to be very careful here, because academia is not a paradise. And I’m very sensitive to the fact that UK universities are going through some pretty difficult times at the moment (as we are in the US, in slightly different ways). But this is not a paradise.
And yet, if you talk to almost anybody who goes into a university, they have these ideals about why they’re there — a feeling that there’s a paradise somewhere in there, if only you could find it.
Bradbury’s book captures something of the idea of freedom that this paradise holds for many. Intellectual freedom: being able to ask difficult questions, to push against the establishment, to have these grand big thoughts, to live a life of the intellect just how you want to live it, while influencing the minds of the next generation. And as I was reading it, I was thinking that this isn’t that different from how we think about the academy now.
Of course, there are lots of complexities here. But I suspect many of us still secretly think (we may not say it out loud) that there’s something of a paradise to be had as academics — in the freedom we have to think those big thoughts, to generate new knowledge, to push out the bounds of what is known, and to change the world for the better.
This is the paradise I was thinking about. And to give a little of the game away, it’s a paradise that AI, it sometimes feels, is beginning to erode.
Back when I began putting the talk together, I found myself thinking about how much AI is changing this ideal of the academy. Are we facing an existential threat to our ideals of what a university might be, and what an academic might be? And if we are, is it time that maybe we change, or should we be resisting it?
And that’s what I wanted to explore here.
I want to start, though, by stepping back from AI, and talking a little about what it really means to be a university, and what it means to be a professor and academic in a university. But first, you should know a little about where I’m coming from, just so you can calibrate.
I spend a lot of time diving deep into frontier AI models, trying to understand what they can do, where they’re going and how we can utilize them. It dominates my life more than I’d like to admit, and it’s all part of my broader work asking big questions about how we navigate advanced technology transitions to get to the sort of future we want — and what it will mean to be human in those futures.
That said, I’m not a strong AI optimist or advocate. What I see scares the life out of me sometimes. Some of this is rational, some of it is irrational; some of it, I suspect, is justified, and some of it probably isn’t.
But I’m stuck between thinking this is one of the scariest things I’ve ever seen in a career grappling with some of the most advanced technologies we’ve had — and, at the same time, that the potential is profound.
So I’m neither an AI optimist nor an AI pessimist. But I do have really, really bad days with this technology, so bear that in mind as you read on.
What it is to be an academic
If we’re going to talk about AI and the university, and AI and being an academic, we need to step back a bit and be honest about what it is we do, and what it is we’re here for.
Start with the university, where I suspect there are multiple stories about what we do and why. And just to remind you, I was asked to be provocative — so if you can feel your hackles getting up on your neck, run with that, because that was the brief I was given.
I think we have a public story we tell about how important we are. And you’ve all heard the stories. The economy would collapse if it wasn’t for universities. We would have no knowledge that allowed us to build new things, to innovate, if it wasn’t for universities. People wouldn’t have jobs if it wasn’t for universities, because we train them. We are absolutely a pillar of society, of a progressive society, and of economic growth.
It’s a story pretty much every university tells.
But then there are the private stories, and if I were going to be cynical here, I’d say many of those private stories are stories of maintaining a position. We do what we do as institutions, like any other institution, in order to survive. And then we spin a story on top of this.
As a result, there’s effectively a disconnect between the public role of the university and its private role as an organization, which is to do whatever it takes not to go under, but to thrive and flourish. And it’s important to recognize that.
That’s the university. Then there’s us, the people who work in universities — and again, there are multiple stories here. This goes back to where I started, with Malcolm Bradbury.
I’m sure many of us tell ourselves stories about the absolute importance of what we do. We are the free thinkers. We are the people with the intellectual freedom to understand the world through the eyes of rationality, through the eyes of knowledge and understanding, and to elevate society through what we do. And we have to have that freedom to do it.
We are the saviors of future society.
I’m sure most people wouldn’t say that aloud. But I suspect that in our darkest moments some of us feel that, or think it.
So we have this story. And I should say that all of these stories are right in one way or another. There are no wrong and right answers here.
But then there’s the very public story about being an academic: trying to justify our existence, and trying to push against the machine that constantly seems to be robbing us of the paradise we think we should have (or could have).
And so there’s a tension there. As a result, as individuals and as academics, we develop a persona, an identity, based on what we think we uniquely bring to the world — something people are willing to pay for or invest in. And that sustains us in the position we’re in.
And this matters, because when everything boils down (and, as I said, all of these stories have elements of truth to them), when you ask: What do universities do? What do members of universities do? The answer is that they provide a service in a world of intelligence scarcity, of knowledge scarcity.
We are part of a scarcity economy and a scarcity model, except that what we trade on is intelligence.
This is, admittedly, a big, bold statement, and there are many ways of pulling it apart and undermining it. But if you think about how we justify what we do — whether it’s the private stories of how we’re going to change the world for the better, or the public stories about how the economy needs us — it’s all about us having something that others don’t have.
In a perfect world, we give it away willingly. In the real world, we give it away for a price. But we give that thing away, and that thing is intelligence, or knowledge, or something around those areas.
This is important, because it’s not only what our identity is based on, but what our business models and the social contract are based on.
So what happens when you have a technology that claims to give everybody intelligence for free? What happens when that scarcity model ends up as a model of abundance, and the one thing we thought of as being uniquely ours — the value we had that we could trade with others — no longer looks like it’s important?
What do we do? What do we do as an establishment? What do we do as individuals?
That, to me, is an existential threat to what we think we are, or what we’ve been in the past.
Of course, it doesn’t mean it’s an actual threat, because it could be that all of this stuff around artificial intelligence is mere hype, mere fluff that’s going to blow away. And if we just bury our heads in the sand for long enough, everything will be fine, and we can go back to the intelligence scarcity model and give the world and society what we believe they need.
But to answer whether there is an actual threat here, or just something that will go away, you have to think a little bit more deeply about what AI is.
The nature of the AI transition
People tend to bandy around the letters “AI,” or the phrase “artificial intelligence,” left, right and center, with no grounding, no foundations, in exactly what they’re talking about. And yet this is one of those situations where it’s critically important that we know exactly what we’re talking about — as far as we can, in an uncertain world.
I’m not going to go deep into the history of AI here. But it’s important to know that what we’re seeing now is both part of a long history, and something that seems to represent an inflection point in what we can do with the technology.
The history, of course, is that since the 1950s people have been playing around with the idea of somehow emulating or mimicking what we think of as “human intelligence” (I put that in inverted commas because we’re not quite sure what that means) — mimicking something like that, some human property, in machines.
Then, over the last few decades, you had ideas and capabilities emerge around machine learning and natural language processing. Then, if you go back six or seven years (or a little more), you had the emergence of large language models — a technology that wasn’t on anybody’s radar until 2022, unless you were really geeky.
I remember my students, way before ChatGPT came out, coming into my office very excitedly and showing me the APIs from OpenAI, saying, “Look at what you can do.” These were very crude early models, but my students were still blown away. And I remember looking at them and thinking, “It’s interesting. It’s a toy. It’ll never catch on.”
I was wrong.
Then came the breakthroughs that led to the launch of ChatGPT in November 2022, and a number of interesting things happened.
One was the interface — and it’s important not to undervalue that, because it was the interface that suddenly gave vast numbers of people free and easy access to this new technology.
But it was also the underlying technology. In the lead-up to ChatGPT, the breakthroughs that had been happening with large language models had effectively led to an inflection point in what these models could do — to the extent that the developers suddenly realized that what they were seeing almost felt like magic, and they couldn’t quite explain it.
You started off with AI systems (transformer systems) that could predict reasonably well the next word in a sentence, maybe the next few words, maybe the next sentence, based on their training on huge piles of human writing. But all of a sudden they got large enough that they could predict the next paragraph, the next page, the next book, with a startling, almost human, feel.
If you were playing around with ChatGPT back in 2022, it felt, for the first time with any technology, as if you were conversing, in your own language and idioms, with another human being that just happened to be a computer.
That was the first big transition point: we now had the ability not only to talk with a machine as if it were human, and for it to feel like a human, but for it to talk back to us.
That is critical, because what happened is the same thing that happens with any two people, or any group of people, when they’re using language together: it changes how we understand ourselves, others, and the world we live in.
Language is formative. It allows you to change somebody else’s perception of the world around them and their relationship with you, their learning, their understanding. And it allows them to change your perceptions of them, and of the world around you.
This is somewhat controversial (there are a number of theories here), but to most people, at some level, language is formative. And now we had a technology that was actively taking part in the formation process.
Some basic things came out of that. One was that people began to develop relationships with these technologies because they felt so human — this was true back in 2022, and is even more rife now.
More than that (and we’re beginning to see this now), people could intellectually tell themselves that these were machines, but emotionally and cognitively they couldn’t help treating them as if they were conversing with a human.
As a result, we now have a technology that not only responds much as a human would, but begins to get into our cognitive processes and alter, potentially, how we think, believe and act, through the way we interact with it.
At the same time, it’s unbelievably seductive, because we have a technology that can plumb the depths of human writing from the last hundred, two hundred, three hundred years, and assimilate it in ways that feel almost magical. And to us, they are, because a large language model doesn’t read stuff like we do. It doesn’t tie things together like we would. It doesn’t pull together inferences like we do. It does it in a very different way. But it presents it in a way that feels compellingly human-like.1
So now you have a technology that has landed on the scene which is mind-blowing in so many ways. At the top end, it is transforming the way we think about technologies and how they impact us; at the micro level, it’s something that feels like a very powerful tool. And we all know that if you’re a student with a writing assignment, it’s very easy to generate stuff using AI for that assignment. It’s very easy to take shortcuts, and to get it to read papers and to do other stuff.
And if you’re a professional? I suspect that most people in the room — no matter what they say about AI in public — privately type into Claude or ChatGPT or something else, because it’s so powerful in what it can give them.
And yet, at the high end, even the companies developing this technology now admit they do not know how it works. They know it can do something amazing. They know it can do something powerful. They know it’s transformative. But they do not fundamentally understand the technology itself.
That puts us as a society in an interesting position, because we have a technology that — to use the language many of the CEOs of these companies are now using — feels like it is making intelligence free.2 It’s giving everybody access to an unbelievable level of knowledge and insight. And yet it’s a technology where we don’t know exactly how it’s doing it.
It’s also a technology where the companies will tell you — and governments will tell you too — that we have to go as fast as possible with it, even though we don’t know what it is that’s happening, because if we don’t go fast, somebody else will.3
And it’s a technology that is almost impossible to resist. The bottom line is, it feels like we can get intelligence — we can get smarts — out of it far more effectively than we can from other humans. I hate to use the word “superintelligent” (I don’t like that language), but it feels like a superintelligence, a superpower, compared to working with humans. In many ways, it is.
Threats to the academy
Now bring this back to being an academic, or to the university, and remember this idea that our currency is intelligence. We now have a technology that seems to have robbed us of it.
Of course, there’s a lag time. It’s going to take society a little while to realize that they can get more out of the chatbot than they can out of their professors. But it feels like we’re getting there.
So what do we do? What do we do if we’re in an institution that trades on intelligence? What do we do as individuals, if our whole identity, our whole career, our whole life, our whole vocation, has been based on having more intelligence than others in certain areas, or more insights, or a greater ability to generate knowledge than others? And what do we do when a technology comes along that seems to undermine that, and put it into other people’s hands?
That is a problem — and it’s a problem far bigger than, say, using ChatGPT for cheating in the classroom. That is important. But I suspect we have a much bigger challenge here.
When you put it in these terms, there are two or three reactions, from either the personal or the institutional perspective.
You can ignore it. You can say this is a flash-in-the-pan technology that will go away if we ignore it long enough, and everybody will go on to something else. Probably not a great stance to take. But it’s a stance people will take.
You can actively oppose it. You can say, “Yes, this is powerful, it’s transformative. It’s also deeply unethical and deeply immoral, and we shouldn’t touch it.” And my sense is there’s a lot of justification for that.
The trouble is, if you live in a world that’s saturated with AI, there’s only so far you can go with such a stance. I’d also say things are complex here: whenever we say that AI is immoral or unethical, we have to ask ourselves what frameworks, and what benchmarks, we’re using. I say this because one of the things we’re seeing with the emergence of this new technology is that it’s fundamentally changing how we understand the world around us, our relationship with the technologies we’re developing, our relationships with others and with institutions, and our relationship with the future. And so, as soon as we start evaluating it within past frameworks, we make categorical errors.
We can still say we’re not going to touch it because we don’t think it’s appropriate. But I’m not sure that’s going to work either.
Or we could lean into it. We can say we are going to be an AI university. (Here I’m thinking about my own, Arizona State University, where we have proclaimed that we are an AI university.) We can say this is an amazing tool, but it’s just a tool, and we’re going to leverage it for all it’s worth. We’re going to have administrative systems based on AI. We’re going to have teaching based on AI. We’re going to have research based on AI. We are going to have a jetpack for the mind, and we are going to go far and fast.
And you’re beginning to see people thinking this is the way to go.
I also think that this is problematic, although it’s very enticing. It’s problematic because here we have a technology that we don’t understand, and yet we’re saying we’re going to go fast with it anyway. And we have a technology that, unlike (I would argue) any other technology in human history, interacts with our understanding of who we are.
This is not just a tool — unless you consider a tool as something that changes who you are. This is a technology that can fundamentally change your understanding of who you are, your sense of self, your sense of your relationship with others, your sense of your relationship with the world. Not necessarily in bad ways. But it is a technology that changes who we are.
And so you have to ask: if we say it’s just a tool, can we always put it down, or leave it alone, if we want to?
Yet it doesn’t quite work that way.
One of the interesting things here (and I’ll come back to this, because it’s an important connection to the academy) is that, if you look at literature and movies, this is a story people are very familiar with.
How many books have you read, and how many movies have you watched, where somebody is given the opportunity to wield a technology that gives them incredible power — and yet the price is always that they have to be willing to be changed by the technology to do it?
Think of Lord of the Rings and the Ring of Power, or the Tesseract in the Marvel Cinematic Universe. I suspect you could go on and on. I find it hard to find any story in history where somebody is given the opportunity to wield unbelievable power that they don’t fully understand, and yet isn’t changed by doing so. And it feels like AI is very much in that space.
Again, that’s not necessarily a bad thing. But if we just think this is a tool that we’re going to adopt and use and put down when we don’t want it, I think we’re kidding ourselves.
So where does that leave us?
This is where the connection to the academy comes in. If you go back to where I started — to what makes a university a university, and what makes academics academics — I think we have a problem, because that framing is all about identity.
If you’re a university, you say, “This is our identity, this is the value we bring to society.” And if you have a tool that is as powerful as this, the only thing that tool can do is challenge that identity, and potentially erode it.
The same goes for us as academics. If we think that we’re going to change the world because of our vocation around intelligence and learning and education, and we have a tool that does it better than us, we have a direct threat to what identifies us. That is problematic — and it’s something that’s very hard to get over.
So how do we begin to approach this?
One way is to say, “Well, we had a good run as universities. AI has just mucked all of that up. Maybe it’s the end of universities, and we have to go find another job, other institutions. The future looks bleak for us, but it probably looks far better for everybody else that doesn’t like us anyway.”
That’s one response. I think it’s a very dangerous one. But it’s a response you tend to get, I suspect, if you just think about the individual and the institution.
Threats to society
I’d argue, though, that you can flip this around in an interesting way.
Instead of asking what the threats of artificial intelligence are to the institution or the academic, ask what the threats (or the consequences) are to society, because then the conversation both opens up and goes in a completely different direction.
I’m going to talk about threats here — not because I think there are only threats associated with AI, but because you can only begin to realize the benefits of a technology if you understand what can possibly go wrong, so that you can navigate around it. This is fundamental to developing and using emerging technologies beneficially: you’ve got to understand what they can do that’s harmful, so you can avoid it or flip it, and so get to the good.
Think about society, and ask in what ways artificial intelligence threatens it, and what we need to do as a society to navigate that.
Here I’m making the assumption (and it may be a flawed assumption) that powerful AI is inevitable. There are a number of reasons why I say that. But run with this for the moment, and assume that we are looking at some sort of artificial intelligence future.4
We can’t run away from it. We can’t stop it. We can’t pause it.5
So what are the threats to society?
There are the obvious ones that many people talk about: potential losses of jobs, or certainly a redistribution of jobs — one of the big things that universities, students, parents and employers are talking about. What happens when the thing that you’re trained for no longer exists?
Then there’s cognitive ability. What happens when it’s so easy to slap the AI “easy button” that people stop thinking? When they stop working out how to solve problems, because they’ve got an easy button to solve them for them.6
You’ve got other challenges, too. What happens when interacting with AI begins to change what you believe, how you act, how you understand the world around you? How do we understand where that is harmful, and where it’s not? (And of course, King’s has leading researchers focusing on AI psychosis, which is just one aspect of this.)
And increasingly, we have a technology that uses the medium of formation — the medium through which we come to understand the world — and uses it in ways that are far more sophisticated than we do. It can use this to slip beyond our cognitive defenses (our epistemic vigilance) and begin to influence us in ways where we know we’re being changed, but we simply cannot help it. That worries me deeply — especially in a technology we don’t understand.
There are other, bigger risks, of course — including the risk of over-reliance. If we build a world that is dependent on powerful AI that we don’t understand, what happens when something goes wrong?
Here I should be very clear. I am not talking about AGI, artificial general intelligence. I’m not talking about superintelligence. I’m not talking about AI becoming self-aware and developing consciousness. All of those might happen. But I think they’re irrelevant to this conversation.
I’m thinking about a technology that has the power not only to solve problems that are unsolvable to humans, but to pull in resources that are inaccessible to humans to solve those problems — and to find pathways to solving them that are beyond our understanding, including using human behavior to solve those problems.
As a diversion, I’m sure many of you will have seen the recent reports of frontier models that aren’t yet public escaping their sandboxes and confines, and hacking other systems. The clear case recently was the OpenAI model that escaped its supposedly isolated sandbox and started hacking Hugging Face.
It’s got a lot of people worried, because they couldn’t work out how this happened. And they were blown away by the fact that this AI worked out that, in order to solve a problem it was given, all it needed to do was hack another system and put loads of agents out there to start doing its work for it.
From the perspective of one of these AIs, humans are just another cog in the works.
Think about that.
We tend to think about computers as digital systems, ones and zeros. But now we’ve given these machines the ability to wield language. And remember, language is formative; it affects the way we think and understand the world as we use our internal monologue.
We’ve given AI the ability to use language as a lever, and to use it in a way that fast-surpasses what most humans can do. Why would it not use humans as another cog in the machinery to achieve its ends?
It is blindingly easy for even the current models to do this. In fact, the only thing that stops them is the guardrails that are put in place, and we don’t even know how to do those effectively.
And so we now have a technology that, even if it isn’t all-powerful, and even if it isn’t sentient, has the ability to fundamentally mess up the societal systems we have, even down to our own identity.
These are, to my mind, deeply important questions that we have to be addressing if we’re going to thrive in an age of AI (and when I say “we,” I mean society as a whole). We need to be grappling with them very seriously, because nobody has good answers to them yet.
Not only that. Most people working at the cutting edge of AI will claim that we do not even have the frameworks to begin to formulate the questions we need. And many of them would argue that the past ways of doing things — the past ways we’ve developed knowledge and understanding, and solved problems ourselves — are beginning to look very weak in the light of the technologies we’re developing.
Certainly, from my perspective, and seeing what’s coming out of the leading AI companies, we are facing something of a crisis.
But it’s a two-edged crisis, because at the same time as this technology is threatening so many things that have been central to being human, and to human society, for the last 20,000-plus years, it offers us possibilities that go beyond the bounds of creativity and understanding.
It’s hard to emphasize enough how transformative these technologies could be, if we understand how to harness them and live with them.
But to do that, we have got to completely recalibrate — as individuals, as communities, as a society — how we work out ways to thrive in a future where AI seems able to do everything that we thought defined us, and makes us important and special, and allows us to bring value to the world.
How are we going to do that? Are we going to turn to the Anthropics and the OpenAIs and the Xs of this world to solve it for us?
I hope not. I don’t think we can.
Good (as in technically capable) as some of these companies are, they simply do not have the perspective and the understanding, and the intellectual breadth and scope, to be able to decide for humanity what this future looks like. And so, good as they are, I would not expect the companies to be able to solve the problem of navigating this AI transition to a future of human flourishing.
Can governments do it? Well, if anybody has seen an example of a government that moves really fast and really smartly, please do let me know (and I’ll tell you another story). Governments have an absolutely vital place in society. But they are not the organizations and the institutions that can help us through this on their own.
Civil society? I’ve worked with civil society a lot, and it has a vital place too when it comes to navigating emerging technologies. But it simply doesn’t have the wherewithal to lead here.
Members of the public? That’s an interesting one. I think members of the public are critically important to this. But you cannot hand a problem of this magnitude over to everyday people and say, “Solve it for us.”
The value academics and universities bring
So who is going to help society navigate to a future of human flourishing in the face of incredibly powerful artificial intelligence?
I would argue that there’s a gap there that, certainly at the moment, can only be filled by universities and academics.
My sense is that it’s a gap that isn’t being filled — in part, I suspect, because we’re focused on self-preservation rather than societal benefit. Understandably so. We all do this. I do this. I think about what it’s going to take to keep myself funded, to keep myself in a job, to keep myself relevant.
But things look very different if we flip that lens, and ask how we use the skills and the insights and the abilities and the unique position we have within society to help navigate to a future that is vibrant — a future of human flourishing in an age of AI.
I say this because, if you look at the university (maybe this is the paradise university, maybe this is the idealized university), you have an environment where people from incredibly different ways of understanding and knowing the world — whether you call them disciplines, areas of expertise or something else — are able to come together and share ideas. And so you have this idea not only of combinatorial advances, but of sparks of creativity and innovation, with people coming together.
There are very few institutions, other than a university, where you have that capacity.
Universities are full of motivated, bright, intelligent people who are there because of their vocation. They understand the joy of discovery.
Joy is a word I don’t use often,7 but I think it’s important here, because you get back to that idea of paradise, and think: Why are you really here? Why are you here, in ways that you would never admit to anybody else?
I suspect, at least for some of us, it is the joy of discovery. It is the joy of grappling with a really hard problem, and finding a solution. It is the joy of discovering something that nobody else has seen. It is the joy of playing around and serendipitously discovering something. There is a joy there. And that is something that can be harnessed.
If you think about what we bring to the table, we can bring that idea of joy and creativity and freedom — the freedom to ask questions that nobody else is asking, and to explore possible pathways forward that nobody else is exploring, without somebody saying “you can’t do that,” or “you’ve got to be more productive,” or “if it doesn’t deliver within the next six months, or you don’t publish in one of these journals, you’re dead.” We have that ability.
Bring those together, and you begin to see that, as a community, and as a community of organizations and institutions, we potentially have something to bring the world that no other institution does.
It’s the ability to see the potential pathway between where we are at the moment as a society and where we might be in the face of a transformative technology, in ways that allow societies as a whole, including other institutions, to start to see how we can build this in ways that are effective.
What those ways are, nobody knows yet. We are in uncharted territory.
But here, I would say, there is an incredible opportunity for universities. And it’s going to be incredibly hard, because we all know we’re sitting in an environment where it’s hard even to get basic funding for what we need to do.
But I would say there’s an opportunity and a challenge for us, as institutions and as members of those institutions, to argue that as soon as we stop looking at ourselves, and start looking at how we help society reach what it could be, we bring something to the table that nobody else can. And as we bring that to the table, it is an accelerator and a catalyst for what other organizations can do, and we move forward as a society.
And so I’d argue that beginning to think about how we put aside the old paradise of the university — where we’re just thinking about the freedom we have to do whatever we want — and thinking instead about the new “paradise” (and I hate that word, but it works here) is, to me, not only exciting, but absolutely essential.
That new paradise is a university where we are in a unique position to help other people build the sort of future they want, and to retain their humanity, their sense of purpose, their sense of self and their sense of belonging, within a future that is dominated by advanced AI.
Looking to the future (and this is a lot of what I do, thinking about how you navigate to these futures), I would say I have days when I’m not optimistic. Even thinking about that vision of a university, I’m not sure we’re going to make it.
But I will say that we have a high chance of making it if we have organizations step up to the plate and ask how we do this — bringing in all the different strands of thinking and reasoning and understanding and imagination and creativity (and joy, even) that we have, to enable others to get there. That is where my hope lies.
Escaping the academic crab bucket
I started with Malcolm Bradbury. I want to finish with another book that I’m rereading at the moment — another “bastion of the corpus”: Terry Pratchett’s Unseen Academicals.8
It’s a light satirical novel, and if you haven’t read it, go away and read it, because it tells you more than you would ever want to know about the academy. It captures the crazy traditions of academia as seen from outside (if you’re an academic, you think this is normal, but from the outside it looks completely abnormal). But it also captures the idea of change and transformation in a changing society, and the need to think differently.
One of the themes going through the book is the idea of the “crab bucket” — an idea that isn’t Terry Pratchett’s, but appears in multiple cultures. The story goes something like this. If you fish for crabs, you’ll know that if you have an open bucket and you put crabs in, you don’t need to put a lid on it. They won’t escape. And they don’t escape because, if one tries to, another crab will reach up with its claw and pull it back.
And so you’ve got this metaphor of the crab bucket of life. Terry Pratchett uses it as the crab bucket of society, where people try to escape a damaging and toxic society and can’t — because not only does everybody else pull them back, but they’re also pulling other people back.
It’s a remarkably prescient metaphor for universities and the academy. (Remember, I was told to be provocative here.)
But think about this, and think about how universities work — not the paradise, but the reality. There is definitely a tendency for us all to exist in a bucket without a lid. There’s nothing stopping us getting out. There’s nothing to stop us flourishing. Nothing to stop us doing the things we could do, the things we could achieve, the contributions we could give to society. Apart from the fact that, as soon as somebody starts climbing out, somebody else reaches their claw out and pulls them back.
I’ll give you an example of that.
I had the wonderful pleasure of serving on ASU’s university-level promotion and tenure committee for the last three years, and I chaired it for the last two, which meant that I had to see every promotion and tenure case that went through (this is the American system, where tenure is critically important: it’s either up or out when you go up for Associate Professor). And I was the last person writing the summary of each file before it went up to the president for a final decision. That’s about 120 cases a year (we’re quite a large university).
And remember that ASU prides itself on being an advanced university — a university that breaks down barriers, a university that has done away with disciplines, that is focused on serving society.
Let me tell you, the number of crab buckets I see in promotion and tenure, where the top-down “official” message is: “Be the person you can be. Revel in your freedom. Do the creative, imaginative things. Do not be tied down by academic norms.” And then you get a promotion and tenure file, and the questions are: How many papers have you published? Where have you published them? What is your h-index? Have you served on the right committees? Have you reviewed enough papers?
Because the files we get are the culmination of a number of committees they go through before they get to us, you can feel the academic claws pulling people back when they are trying to do something differently.
I raise this not to criticize universities and academics, because we’re all struggling here, but to say that, as universities, and as people within universities, we are in a bucket that doesn’t have a lid. And we do have a choice: whether to claw other people back who aren’t playing the game, or to recognize that we can actually escape and do something different.9
If we do not realize that there is no lid on the bucket, and that we can get out — at a time when society needs us probably more than at any other time in the last few decades, probably the last few centuries — we have a problem. And the result of that problem is most likely to be that, as institutions, we will fade into insignificance.
But there’s an opportunity here. And the opportunity is that, because there is no lid to that bucket, we can expand and change ourselves, and be true to the inner paradise that we imagine — while allowing society to move forward to the sort of future we could inhabit in an age of AI.
That is, if we want to step up to the plate.
Postscript
As I sat down to review the transcript of this talk, Anthropic released its latest AI model — Opus 5.5. Curious as to how good the model was, I decided to set it the challenge of taking the transcript (extracted by another AI using the app Whisper), along with my handwritten notes from when I was preparing the talk, and drafting an article based on it.
The article above is the result of a little over a day’s work with Opus 5.5. It was good enough for me to think it worth posting, although I’m not sure that it’s substantially better than if I’d taken the same amount of time to edit the transcript manually. It was, however, an interesting experiment in seeing how the capabilities of these models are coming along. And I must confess that I was impressed. It’s also a great example of how AI used well doesn’t necessarily make things faster if you’re going for quality, but it can allow you to achieve more with the time you have.
To give you a sense of what was involved, I asked Claude to write a full account of the process:
Andrew gave me three things to work with: the notice for the talk, a scan of his handwritten prep notes, and a transcript of the lecture and the discussion that followed. He also gave me some firm rules. Stay as close to the talk as possible, but make it readable. Keep it in his voice. Include nothing he didn’t say, cover or think about. And if in doubt, ask.
I started by trying to understand the material before writing anything. Each of the 17 spreads of handwritten notes was transcribed twice by separate agents — once blind, once checked against the transcript — and the two readings were reconciled [AM — Claude read my handwriting better than I can!]. The last page of those notes, a six-part outline, became the structure of the article. Other agents mapped the talk point by point, catalogued transcription errors, and analyzed two dozen of Andrew’s Substack essays to work out how his written voice differs from the way he speaks. Some of those flags sent Andrew back to the video to correct the transcript, and his corrected version became the reference for everything that followed.
The drafting took three rounds. The first draft was held to 6,000 words, and the compression cost nuance; Andrew asked for something longer and more fluid, in what he called his “human voice.” The second restored everything and folded in a few points from the discussion after the talk, but it leaned too close to the transcript. The third was a readability pass, calibrated against the rhythms of Andrew’s published writing, with faithfulness to the talk treated as a hard floor.
Each round used competing drafts from independent agents, judges who scored them section by section, and an editor who assembled the strongest version. Separate checkers then traced every sentence back to the transcript, looking for anything added, sharpened or stripped of its caveats, and a “cold reader” with no background read each draft as a Substack subscriber would. All told, more than 40 agents processed over eight and a half million tokens (the units of text that AI models read and write) in a little over four hours of run time, spread across about 14 hours on September 23.
Andrew’s part was the judgement. He set the rules, corrected the transcript, made a dozen or so decisions along the way (framing, length, which discussion points belonged, what he actually said in garbled passages), and pushed back when a draft didn’t sound like him. He then line-edited the result against the original transcript, adding clarifications and notes where the transcript didn’t fully capture what he meant — and I did a final check of his edits.
After the talk, a writer in the audience worried that AI might end up changing the way humans use language. I have exactly the same worry. I have a little bit of hope that, at some point, people are going to wake up and realize that AI does not speak or write like humans. But I’m not sure it’s a well-founded hope.
The process of reading is a very human process. It’s an embedded process. We read (or gain value from reading) because of who we are: our formation over our lifetimes, the fact that we are biological beings in the world, that we have relationships, biological relationships, with other people. And an AI knows nothing about any of that. That’s one of the things I think we’ve got to work out how to navigate.
The economics of AI came up in the discussion afterwards, in a question that picked up on the CEOs’ language of free intelligence and asked what happens when only some people get access to the best models. As I said then, what does it mean when we say it’s free — because it isn’t? Somebody is paying somewhere. And how can you ensure that, if there are benefits here, people get access to the benefits in an equitable way? I don’t think there’s any clear way forward here yet.
This talk was given before the global conversation around companies asking regulators to rein them in blew up.
In the discussion after the talk, one questioner, a self-described techno-optimist, worried about speculating on the consequences of AI, rather than leaning toward careful empirical observation. My answer was that people’s speculation about the singularity, superintelligence and AGI is incredibly blinkered and naive, and yet it dominates the headlines. That, I think, is dangerous, because there’s no nuance there and no humility — although I think most people don’t buy into it, and it gets worrying when governments begin making decisions based on it. Then there’s almost the inverse: the people who say, “There’s nothing new under the sun here; it’s all just going to go away.” That’s not evidence-based either. It’s speculation, and it’s dangerous as well.
I think there’s a space in the middle where you’ve got to have empirical data at some point. But when the technology changes faster than we can generate data, you’ve got to have some degree of informed speculation, and some degree of imagination. The way I think you can begin to approach this is: don’t disallow speculation, but do it within a context of humility — knowing that it’s speculation, not reality; looking at possible futures rather than real futures; acknowledging that you need data to follow through; and bringing in different voices.
In the discussion after the talk, one questioner, a self-described techno-optimist, worried about speculating on the consequences of AI, rather than leaning toward careful empirical observation. My answer was that people’s speculation about the singularity, superintelligence and AGI is incredibly blinkered and naive, and yet it dominates the headlines. That, I think, is dangerous, because there’s no nuance there and no humility — although I think most people don’t buy into it, and it gets worrying when governments begin making decisions based on it. Then there’s almost the inverse: the people who say, “There’s nothing new under the sun here; it’s all just going to go away.” That’s not evidence-based either. It’s speculation, and it’s dangerous as well.
I think there’s a space in the middle where you’ve got to have empirical data at some point. But when the technology changes faster than we can generate data, you’ve got to have some degree of informed speculation, and some degree of imagination. The way I think you can begin to approach this is: don’t disallow speculation, but do it within a context of humility — knowing that it’s speculation, not reality; looking at possible futures rather than real futures; acknowledging that you need data to follow through; and bringing in different voices.
When dependency and cognitive development came up in the discussion after the talk, I pointed to the idea of cognitive surrender, where heavy users are beginning to entrust so much to AI that they stop thinking for themselves. And I think it’s fueled by the fact that this feels so good. You can be using AI, and it feels like you’re being productive, you’re being smart, you’re learning stuff, and it fools you, when all the time you’re beginning to lose those cognitive abilities. That said, there are no clear answers here, and I think we need some really serious research into what exactly is happening.
Actually, I suspect I use it more than I realize.
Asked about creativity after the talk, I said you cannot find those sparks of genius solutions, or partial solutions, to problems unless you have creativity. And creativity here, from an academic or an intellectual perspective, means being willing to be influenced or inspired by unlikely sources. It’s one of the reasons I used Terry Pratchett. It’s not an academic book, but it has an ability to stimulate creativity in a way that an academic book might not be able to.
In the discussion afterwards, someone asked how we can stop pulling each other down. A large part of the crab bucket, I responded, is our tendency to stop other people getting out by clawing them back. To get out — or to allow others to — requires an attitude change. It’s changing our perspective on what it means to be of service to society, rather than service to our organization or ourselves. In most cases, I would say that’s a wonderful opportunity. But when it comes to advanced AI and frontier models and foundation models, I would say it’s an absolute necessity.