Interview by Alex Arnold
Editor’s introduction: Tim Crane is Professor of Philosophy at Central European University in Vienna, Austria. He is one of the world’s leading philosophers of mind, having published numerous books and academic articles on consciousness, the nature of the mind, and the nature of thought and reasoning. He was previously Knightbridge Professor of Philosophy at Cambridge University and, before that, Professor of Philosophy at University College London. For a time he also served as the Times Literary Supplement’s philosophy consultant editor. He is currently the Director of Research for the Austrian Cluster of Excellence project Knowledge in Crisis. In this first part of a two-part interview, Professor Crane discusses the challenges AI poses for learning and formation in higher education, the advantages of Christian anthropology in answering questions about AI, and whether current AI systems have minds. This interview has been edited for length and clarity.
Alex Arnold: Let’s start with concerns about the use of AI in education, where, in the U.S. at least, there is a lot of consternation. What is your institution, Central European University, grappling with now when it comes to AI, and how is it proposing to respond? And how are your fellow academics in Europe reacting to the emergence of AI in their teaching, their supervision of students, and their research?
Tim Crane: It’s a difficult question. If I start with my institution, we’re struggling in the way that everyone else is. I think we haven’t bought into the whole thing. We haven’t said, “This is the future, everyone has to give up education, give up doing computer programming yourself, give up philosophy, give up logic, give up history—just learn how to prompt.” That message is not one that we’re endorsing at all. But we recognize that you can’t pretend generative AI doesn’t exist. You can’t say, “Oh, just please don’t use it, guys.”
We’re experimenting with little things like different kinds of assessment: more in-class assessments and more written work. We’re lucky because we’re a small university with small classes, so we can get more involved in actually seeing how students construct their written work.
That’s at the level of writing. And I think the world will find some way of dealing with this. AI writing is flooding higher education now. I’m just talking about the education side of the university; we can talk about research later. But I still think there’s an enormous risk, and this is something I’m trying to spread in the work that I’m doing for my new project, “Knowledge in Crisis.” Education is in crisis. And it’s not just because students aren’t writing anymore. I think in some ways an even deeper problem is that they’re not reading—and I don’t mean reading paper books. They’re not reading anything. As you know, you can open up Adobe Acrobat Reader and one of the first questions it’s going to ask you is, “This looks like a long document—would you like me to summarize it for you?” For certain things, that’s fantastic. You’ve got your rental contract, it’s 25 pages long, and the AI summarizes the main points. I’ve worked a bit in university administration and we have these big, boring strategy documents. AI can summarize them in two pages, and that’s great.
But when it comes to a work of philosophy or a work of history or a work of literature, the reading is the understanding. You can’t just summarize the Critique of Pure Reason. To summarize Descartes’s Meditations, what good comes out of that? If you don’t actually read these texts, you’re never going to understand and you’re never really going to learn. And what I fear is that AI is short-circuiting the process of reading. Education isn’t just about producing text. It’s about understanding and actually struggling with the difficult works passed down to us. A student with a deadline coming up has to summarize some argument in a text, and it’s given to them on a plate by AI. It’s not only that it might not be accurate, which is a problem; it’s also that if they haven’t read it, then they haven’t understood it.
If you don’t actually read these texts, you’re never going to understand and you’re never really going to learn. And what I fear is that AI is short-circuiting the process of reading. Education isn’t just about producing text. It’s about understanding and actually struggling with the difficult works passed down to us.
Alex: You’re concerned that LLM use impedes reading and deep intellectual engagement with a text—or at least gives students an incentive to shortcut it. How do you even start to address that problem? There has always been a temptation to take shortcuts, because students are pressed for time. But now they’ve got a tool that makes the shortcut available.
Tim: It’s very hard unless you can actually engage with the students directly. And if you’re in a big public university in Europe, a state university in the U.S., you’ve got these huge classes, so at best only your TAs can actually see whether the kids are reading. We’re lucky at CEU to be a small university. We sit down in class and can actually look through a text page by page and discuss it.
One thing I do is I limit what I expect from them now. I don’t give them very much to read. Then I hope they will actually read those pages, and we can talk about them.
But another thing is that we have to exploit and develop their desire to learn about AI and what AI is doing to them, and not present it as something to surrender to, but instead to take a critical approach to it. Look at what’s happening, what you’re being made to do, which corners you’re being forced into, what paths you’re being forced to take.
There is some hope that a lot of young people are not prepared to buy the big AI myth. We see some resistance to it. Of course, they’re using it all the time and they’re digital natives. But there is skepticism. There was that incredible scene at the University of Arizona with [former Google CEO] Eric Schmidt. He was booed off the stage because he told them that this is the age of AI and they’re so lucky.
A lot of students are smart. They want to learn, and they don’t like being told what to think necessarily. This is why I think one of the things we can do as philosophers—if we’re not in the pay of Google or Anthropic or OpenAI—is spread some skepticism: Don’t believe the hype. Don’t accept that all your jobs are going to be taken by AI machines.
Alex: This need to exercise judgment about AI, and to think about the way it is affecting one’s own intellectual and moral formation, is something we stress at the Center for Christianity and Public Life. We talk a lot about spiritual formation; it’s a key theme for us, and we think a lot about how AI can be a technology that forms us for ill, or hopefully maybe for good, too. Have you engaged your students on these sorts of issues—on the way AI might be forming them in certain vices or virtues?
Tim: This is something I teach them, and also something that we’re doing in this big Austrian-wide project, Knowledge in Crisis. We’re actually doing quite a lot of public outreach and public events, spreading the word not to believe everything we read, not to believe everything AI makers say, and instead to step back and think about what role AI is going to play in our psychological, moral, or spiritual development. I think it’s incredibly important to have such a perspective.
I think the Roman Catholic Church has an advantage here. The papal encyclical Magnifica Humanitas is incredibly impressive and timely in addressing AI. Roman Catholics (and Christians in general) have a special advantage because there’s a very long intellectual tradition of thinking about the nature of the human person, the self, the human being, what the human being is, the body and the soul. Christians have a good starting point for addressing these questions and combating the alternative vision that’s being presented by the tech companies.
The papal encyclical Magnifica Humanitas is incredibly impressive and timely in addressing AI. Roman Catholics (and Christians in general) have a special advantage because there’s a very long intellectual tradition of thinking about the nature of the human person, the self, the human being, what the human being is, the body and the soul.
This is harder in a secular context, I think. But it can be done, because a proper secular morality contains the same fundamental idea, which is what my former colleague [the philosopher] Raimond Gaita calls the “inalienable preciousness of the individual,” the uniqueness of the individual as a source of value. The Catholic conception of the world—I talk about Catholicism because I was brought up a Catholic—starts with that idea and doesn’t have to justify it. But there are other moral philosophies that start with it, Kant’s above all. He thought about what it is for the person to be the locus of value more than almost anyone.
I think AI is presenting us with a profound challenge to how we think of ourselves. And you can see the chaos in the public discussion about this. When are we going to reach AI consciousness? When are we going to reach general intelligence? When are we going to reach superintelligence? And you realize that we’re at sea. We’re completely lost when it comes to having a conception of ourselves and of how this new technology fits into it. We need people to ask, “What are we? Why do we say a computer could be a friend? Why do we say a computer could be a person? What is it about us that makes us want to say this? And what is it about computers that makes this so absurd?” We need real philosophy—and a real debate about the nature of the human.
…you realize that we’re at sea. We’re completely lost when it comes to having a conception of ourselves and of how this new technology fits into it.
Alex: I’ve had lots of conversations with all sorts of different people, from people in churches to people in policy, and they ask questions like: Do generative AI models like Claude and Grok and ChatGPT really think? Are they really reasoning? Do they have beliefs or other mental states? Do they have minds? Are they persons? Do they have equivalent moral status? You are one of the doyens of contemporary philosophy of mind. What is your take on these issues in the metaphysics of mind as they intersect with AI?
Tim: Well, the short answer is this: AI machines are just computers. So if the questions about whether Claude has beliefs, has a character, has a point of view, acts in the world, makes decisions, is a person—if you think these are the places we should start, that’s because you think it makes sense to suppose that a computer could do all these things. Then you have to ask what you’re really saying there, because we need to put aside the idea that what matters is that it’s artificial.
The question of whether we could create an artificial person or artificial agent—like Victor Frankenstein, who created his monster by assembling old body parts—is a different question. But that’s not what Claude and ChatGPT are. They’re not old body parts. They’re computers in vast data centers spread across the world. And they’re computing when they answer questions. They are solving problems computationally with the most massive computational resources that the world has ever seen, and with absolutely ingenious computing design and software. But at the end of the day, they are computers.
So ask yourself: Is a computer a candidate for thinking? To answer that, you first need to ask, what is it that thinks? We start with the idea that a thinker has a point of view on the world, a sense of the world outside it, a persistent existence, an individuality. That goes back to my earlier point about the preciousness of the individual. Each of us is an individual. We each have our own point of view on the world. And that’s what that “point of view” metaphor covers: beliefs, experiences, emotions, and the whole of what [the philosopher] Stanley Cavell called “the whirl of organism.” That is what we are. And that’s common to a Roman Catholic view or to a secular view: that we are organisms. Then we see mental activities in other organisms too. We can see that they have points of view on the world. They’re limited in certain ways, but they also show certain sorts of emotions and other mental states.
We start with the idea that a thinker has a point of view on the world, a sense of the world outside it, a persistent existence, an individuality. That goes back to my earlier point about the preciousness of the individual. Each of us is an individual. We each have our own point of view on the world.
Now ask yourself: How much of that can a computer have? What LLMs do, for example, is output incredibly impressive, coherent sections of natural language text in answer to questions. That’s what they do. But that’s the only thing they do. There are other issues, I know—there are AI agents, for example. But when you consider the richness of all the mental concepts that we use to apply to ourselves—emotion, feeling, attachment, character traits, our beliefs, our experiences, our trains of thought, everything that makes up our individual mental lives—it’s really rather hard to see why anyone would get themselves into the position of thinking that what’s coming out of Claude or ChatGPT is anything like this.
So it’s a bit of a puzzle for me. It’s something that needs diagnosis. This is something in our culture, in our intellectual culture, that we as philosophers have failed to impart: the significance of what we do and of the discoveries we’ve made about the nature of the self and the person. Someone like [the computer scientist] Geoffrey Hinton, who is a Nobel laureate in physics, thinks he can just stand up and say, “Well, if LLMs don’t understand their words, then I don’t know what understanding is.”
Or take Dario Amodei, one of the most impressive marketers of AI. I think the way that he has positioned Anthropic on the moral high ground of the AI landscape is really amazing. In a New York Times interview earlier this year, he was asked what he would say if his model told him, “I’d give myself a 65% chance of being conscious.” Amodei said that he doesn’t really know what it would mean for Claude to be conscious, but that he’s open to the idea.
My reaction to this is incredulity. I don’t know what it would mean for something to be conscious—that’s a very strange thing to say, as if he doesn’t know what it would mean for you or me to be conscious, or for his dog to be conscious, or for a baby to be conscious, or for a dying person to slip from consciousness to unconsciousness. If you really don’t know what it means, then how can you be open to the idea of Claude being conscious?
The looseness of thought among these people is extraordinary. They seem to think that because of their authority and their position, their vague speculations about these things have a certain kind of value that warrants our listening to them. Here’s an absurd example—and I think he’s becoming a bit of a parody of himself: Sam Altman said recently that we have reached the singularity. The singularity is the moment when machines are supposed to become more intelligent than human beings. So what can you say to that? Well, what I would say is: What does it mean for the machine to be more intelligent than human beings?
Stay tuned for part 2 of this interview next week.
Alex Arnold is director of research at the Center for Christianity and Public Life.
Yeah, in a scientistic culture, we systematically devalue certain aspects of philosophy, and put inordinate trust in scientists and technologists when it comes to questions they are not qualified for.