Alan Turing and the Limits of Computation

Ray Briggs
Can computers be like human minds?

Josh Landy
Or is the human mind just a kind of computer?

Ray Briggs
Is there anything computers can’t do?

Josh Landy
Welcome to Philosophy Talk, the program that questions everything

Ray Briggs
…except your intelligence. I’m Ray Briggs.

Josh Landy
And I’m Josh Landy. We’re coming to you via the studios of KALW San Francisco Bay Area.

Ray Briggs
Continuing conversations that begin at philosophers corner on the Stanford campus, where Josh teaches philosophy.

Josh Landy
And at the University of Chicago, where Ray teaches philosophy.

Ray Briggs
Today, we’re thinking about Alan Turing and the limits of computation. Limits.

Josh Landy
What limits Ray? I mean, look at everything that’s happening around us. You’ve got cars that drive themselves. You’ve got protein folding software that discovers new medicines. You’ve got great tools to help students cheat on their papers.

Ray Briggs
Sounds like there’s some of that that you’re not totally enthusiastic about, but it does sound like you think computers are pretty powerful.

Josh Landy
Yeah, unfortunately. I mean, the way things are going 10 years from now, we’re all gonna be obsolete.

Ray Briggs
But are we though? There are lots of things that humans can do that computers can’t like. We can feel emotions. We can fall in love. We can hold a real conversation where we say what we actually think.

Josh Landy
Well, some people think computers can do that too. They reckon you can have a totally genuine conversation with a machine. Heck, there even folks out there with AI boyfriends and girlfriends.

Ray Briggs
You cannot have an AI boyfriend or girlfriend. They’ll never understand you because they can’t think, Josh.

Josh Landy
Well, Turing might have disagreed with you. He wrote a famous essay in 1950 at a time when the only computers in existence were those giant behemoths with much more limited powers than the ones we have today. But he was already thinking ahead to what those machines might one day become, and he thought that one day they might be able to think just like human beings.

Ray Briggs
No, you can’t let them fool you. Josh, yeah, sure, computers are really good at mimicking humans these days. They’re better than they were in Turing’s lifetime, but that doesn’t mean that there’s anything interesting going on in there. The lights are on, but there’s nobody home again.

Josh Landy
Turing might disagree. He thought that if a computer can fool you into believing it’s a human, that in itself is proof that it can think that’s what he called The Imitation Game, and what people nowadays call the Turing test, yeah.

Ray Briggs
But what does the Turing test really test for like, suppose a computer tricks you somehow into thinking it’s human. All you can conclude is that you’ve been fooled. If I trick you into thinking my coffee cup is made of pure gold, you shouldn’t just decide that it’s actually made of gold. And if I trick you into thinking a computer has a human level intelligence, you shouldn’t conclude that it actually has human level intelligence?

Josh Landy
Well, I see that argument, but I feel like it’s different when we’re talking about intelligence. Look, you believe other people have minds, don’t you?

Ray Briggs
Well, sure. I mean, even you have a mind. But how do you know that other people, including me, have minds?

Well, I guess I have conversations with them. I ask them questions, and they seem to give me sensible answers, and they say creative and original things that really surprise me, right?

Josh Landy
And Turing’s point is that we should judge computers exactly the same way. You decide that your fellow human beings have minds by talking and listening to them. Well, same for machines. You talk, you listen. If the entity in the other room manages to speak sensibly and say stuff that surprises you. Hey, presto. It’s a mind doesn’t matter whether it’s made of carbon or silicon.

Ray Briggs
Now you’re telling me that humans are no better than machines.

Josh Landy
Well, Turing is telling you that he thought the brain was just a big computer. Imagine a gigantic number of tiny little bots, each doing one incredibly simple thing, add them all up and you get Proust, geometry, bowling leagues,

Ray Briggs
I don’t know that’s a pretty depressing picture of what our inner lives are like. Isn’t there something more to cognition than just zeros and ones? What about beauty? What about love?

Josh Landy
Zeros and ones, Ray—iI’s all zeros and ones all the way.

Ray Briggs
Yeah, I really hope that’s not true.

Josh Landy
Well, maybe our guest will change your mind. It’s Juliet Floyd from Boston University, editor of “Philosophical Explorations of the Legacy of Alan Turing.”

Ray Briggs
But first we sent our Roving Philosophical Reporter, Sarah Lai Stiland, to explore Turing’s legacy as we forge ahead into the era of AI-assisted everything she files this report.

Humans
You’re just a stupid machine aren’t you? Yes, Laura.

Sarah Lai Stirland
Back in 2015 a team of TV producers came out with a sci fi series called “Humans.” The title is a pun for a world where robots and humans have become indistinguishable. Robots are called synths, and they can be household helpers, psychotherapists, even surrogate children.

Humans
I don’t want you touching Sophie, I’m prohibited from initiating physical contact with a human without a clear recorded request to do so. My protocol set currently demands that any such requests from children under 12 must be referred to a parent or guardian before being met, unless I judge the child’s safety or well being to be at immediate risk.

Sarah Lai Stirland
The disclaimer reminds Laura that the synth is just an object, not a sentient being. This might sound futuristic, but we’ve already started to replace human agents with computers,

ABC News
From planning vacations to making dinner reservations. OpenAI is taking its chat GPT chat box to the next level. It’s called Operator.

Sarah Lai Stirland
The new product from San Francisco startup OpenAI clearly isn’t human, but it can be programmed to behave like a human assistant, and this is just the beginning. AI systems can act as teacher, lover, therapist, almost any role a human might fill for another. It seems as if our technology is following an archetype first framed in 1950 when Alan Turing proposed The Imitation Game. His metaphor continues to shape the most influential minds in computer science. Here’s Sir Demis Hassabis, CEO and co founder of alphabets DeepMind, an AI research lab. He’s speaking with tech journalist Alex Kantrowitz in a recent episode of the podcast. Big technology,

Demis Hassabis
The human brain is probably some sort of Turing machine, at least that’s what I believe.

Sarah Lai Stirland
Neither Hassabi, nor many of his peers, believe that any company or project has yet achieved what people are calling artificial general intelligence, or AGI, but they think we are going to get there. AGI implies that computers can do whatever humans can. But what about the coffee test described here by Apple co founder Steve Wozniak?

Steve Wozniak
Could a computer make a cup of coffee? If you analyze all the steps you had to do, to walk, to know what kitchens are. You got that from years of life, of living around kitchens, to know what a coffee machine might be, to kind of look at it and look for clues, like words, to look in drawers for pieces that might fit into you know what a human being is? I like that. AI is so far above anything we’ve ever done.

Celeste Kidd
Yeah, they’re not going through things that haven’t happened, but could.

Sarah Lai Stirland
That’s Celeste Kidd, a cognitive scientist who studies learning at UC Berkeley.

Celeste Kidd
You absolutely could build AI that specialize in doing that sort of thing. And there are efforts to do this kind of thing, but they don’t work that well.

Sarah Lai Stirland
Kidd thinks we’re wrong to stick to Turing’s metaphor.

Celeste Kidd
Because it’s really more of a like psychological test than it is a test of computer intelligence, because it’s dependent upon what people’s expectations are about what AI is like.

Savannah Thais
You know, if you think about human intelligence, we have, we have five senses, and we’re able to integrate those inputs together. You have sort of long term reasoning and planning abilities.

Sarah Lai Stirland
That’s Savannah Thais, a mathematician and research scientist at Columbia’s Data Science Institute. She’s working with experts in different fields to create benchmarks for how AI actually performs. Because putting AI into these intimate decision making roles can be dangerous.

NBC News
Messages from a chat bot may have led to a teenager’s death. His mom is now suing the artificial intelligence chat bot character AI, claiming the chat bot is responsible for her son’s suicide.

Savannah Thais
It’s largely under-tested, so we don’t know exactly how it’s going to behave, the kind of responses it’s going to produce the kind of predictions it’s going to make for people.

Sarah Lai Stirland
Maybe we need to be more like the skeptical mother in the TV series Humans.

Humans
So, have you given that a name? We were waiting for you,Mmum. I think we should call her Anita. Like your friend who moved darling. You know, that’s just a machine. It doesn’t have feelings. It can’t replace Anita.

Sarah Lai Stirland
For Philosophy Talk. I’m Sarah Lai Stirland.

Josh Landy
Thanks so much for that great report, Sarah. I’m Josh Landy. With me is my fellow philosopher, Ray Briggs, and today we’re thinking about Alan Turing and the limits of computation.

Ray Briggs
We’re joined now by Juliet Floyd. She’s professor of philosophy at Boston University and editor of “Philosophical. Explorations of the Legacy of Alan Turing. Juliet, welcome back to Philosophy Talk.

Juliet Floyd
Well, so nice to be here, Ray and Josh, thanks so much for inviting me.

Josh Landy
So Juliet, you’ve been working on Alan Turing for many years now. How did you first get interested in

Juliet Floyd
him? Well, my father was an early computer scientist who worked for the Air Force during the Korean War out in Albuquerque, and he had to model the effects of turbulence created by nuclear weapons on airplanes. And he was very upset by that, so he left and began to work on speech recognition, and later the computerization of hospitals. And he knew that I loved logic problems. We would solve them at the dinner table together. And he used to pay me $1 a board to wire up motherboards for his machines. Motherboards were small, you know, seven inch by five inch metal plates with little posts coming out. And you had a chart you had to follow to literally wire one peg to the other so you could get a machine to boot. And I was down in the basement one day at about the age of nine, and he came down and he saw that I had wired up about 10 boards, so I had $10 and he said, Juliet, this is logic, and it’s going to change the world. And it turns out he was right.

Ray Briggs
What a prescient dad you had. So Juliet, the story of Alan Turing’s life is also fascinating and tragic. Can you give us the one minute version of that?

Juliet Floyd
Yes, there’s a lot to say. Only great drama could do justice to it. I think he’s one of the greatest scientists of the last 500 years because as a mathematician, he gave us a rigorous analysis of the notion of computation, or of algorithm, a very ancient notion, but one that we didn’t have a mathematically robust representation of. He also envisioned how to embody that analysis in machines, and he showed at a very young age, in 1936 that no machine can determine in a finite number of steps whether a theorem is true or false in a given system. So he showed the limits of computability that, in other words, you can algorithmize truth. But then he went on during the war to really, I think, help save Britain in 1940 when u boats were destroying the shipping across the Atlantic, because he went to Bletchley Park and headed up the cryptography division, and they managed to break the German Enigma code, which was impossible to do without the computers that he had actually theorized and designed. And then after the war, very tragically, he remained interested in computers. He did work for the National Physical Laboratory, and he went to Germany late in his life, he was talking to physicists about computers, but he was gay, and he picked someone up in Manchester who robbed him. And being very naive, he reported this to the police, and the police ended up getting him convicted for gross indecency, which is the same law in Britain that convicted Oscar Wilde, and he had a choice between a year in jail or being treated with female hormones and chemically castrated, and he chose the latter. So for about a year, he was very depressed, and he ended up it’s a little unclear how he died. Some people think he committed suicide. His mother never agreed with that. I’m not so sure, but he died by eating a cyanide apple, and as a result of this, years later, the largest petition in the history of Britain was filed with the government in 2009 arguing that he should be pardoned for this. And this moved the British government, so tourings law was passed, the Queen gave her Christmas Eve speech in 2014 and forgave him, and then Turing’s law was extended afterwards to the 40,000 other people who had been convicted of gross indecency since the 1950s so there are over 40,000 people whose lives have been changed by Turing’s law, and now he’s a great celebrity. He’s also on the 50 pound note.

Ray Briggs
That’s so recent, the pardon. So it sounds like Turing did a lot of things, both to serve his country in World War Two and also for computation. Why is it so important to have a rigorous definition of computation?

Juliet Floyd
Well, you see what counts as a computation doesn’t matter which language you’re speaking. It’s kind of a universal concept. There are other concepts that seem to be much more dependent on the particular language. So for example, if I ask, is a theorem provable? Well, it depends upon what the axioms are of that theory and what language you’re using to formulate it. But with computation, what he showed is that the notion of an algorithm is absolute. It doesn’t matter what programming language you use. The other thing is that he has this notion of a Turing Machine, which is really a language game. And. Sense, it’s a comparison between the image of a human being with a paper and pencil calculating according to a step by step rule, unthinkingly, a comparison between a human and a machine. And that’s the reason his analysis of computation persuaded all the greatest logicians immediately that we had the final, ultimate simplification of that notion.

Josh Landy
You’re listening to Philosophy Talk today, we’re exploring the life and thought of Alan Turing with Juliet Floyd from Boston University.

Ray Briggs
Should you trust self driving cars to know what they’re doing? Do you ever find yourself talking to your home sound system? How smart can smart devices be?

Josh Landy
Reasoning with robots—along withyYour comments and questions, when Philosophy Talk continues,

Janelle Monáe
Cause I’ll ever be is your dirty computer.

Josh Landy
Dirty or clean, how special is computer intelligence? I’m Josh Landy, and this is Philosophy Talk, the program that questions everything…

Ray Briggs
…except your intelligence. I’m Ray Briggs, and we’re thinking about Alan Turing and the limits of computation with Juliet Floyd from Boston University.

Josh Landy
Got questions about Turing’s legacy? Email us at comments@philosophytalk.org or comment on our website, and while you’re there, you can become a subscriber and compute the possibilities in our library of more than 600 episodes.

Ray Briggs
So Juliet, I want to pick up on something you said at the end of the last segment, where we were talking about Turing’s idea of computation and why his rigorous definition of computation, or an algorithm, was so important. So if I’ve got this right, you can ask, sort of, Can I do this task in Python, or can I do it in C plus, plus, you can ask for any programming language. What can I do with this programming language? But Turing’s idea was, like, there was a kind of a universal programming language that you can ask just, can I do it in this programming language? And if yes, I can do it with a computer, and if no, I can’t do it with any computer. Is that right?

Juliet Floyd
Yeah, that’s absolutely right. So he thought software was very important, and the particular language you’d use might be very important for that task, but the important point is that there are other languages that could do it, and there’s a limit.

Josh Landy
Okay, now let’s get back to something we were talking about earlier. One of the things that Turing’s most famous for today The Imitation Game, also known as the Turing test. What is that test?

Juliet Floyd
Actually a test of Well, I think it’s an experiment in human phraseology. I think it’s an experiment you set up to explore human human interaction in the presence of machines. So again, it’s a language game in Wittgenstein sense. The question is, what are we supposed to say when we go through this game? And the setup of the game has two parts. When he first presented it, he first asked you to think about a test for gender, which already shows you that we’re dealing with a culturally colored test. So you have a man and a woman on the opposite side of a screen, and the questioner is to determine who is the man and who is the woman, naturally, he assumes the woman should cooperate and the man should not cooperate, and so that’s his control. And then at the second step, you take the man out, and you replace the man with a computer. And the question is, if by posing questions to the two, let’s say A and B on the other side of the screen, can the human determine which is the woman and which is the computer? And so that gives you a certain kind of measure or operationalized procedure. But what I think is very important that philosophers often overlook is that after the game is played, the screen comes down, and then the questioner comes out from behind the screen and begins to chat with the human on the other side of the screen. And if the human on the other side of the screen has been classified by as a computer too often, they might not be willing to have a cup of coffee in a conversation with that human being. So it’s really a social test.

Ray Briggs
So Juliet, I was taught something really different about the Turing test or The Imitation Game, which is that it’s a test of whether the computer, like, is sentient, whether it has an inner life, and that if it’s good enough at imitating a human, then we should think that it’s sentient. But it sounds like you don’t think that that’s the right reading. Like, why do people think that’s the right reading? And why isn’t it the right reading?

Juliet Floyd
Well, it might be. In other words, I can go feed the parking meter. You know, I tell people I’m going to do that, and they understand immediately what I’m going to do. And it’s a nice metaphor, but of course, a parking meter is not an animal and it doesn’t eat but it’s perfectly legitimate. And you know exactly what I mean. So it might be completely right to say the machine seems to be thinking that it seems to be searching for something. My phone seems to be able to find those restaurants nearby. I have no objection to people talking that way, but I think we have to ask what it really means.

Josh Landy
Yeah, that really is the question. What does it actually mean for something, to understand something, or to know something. I mean, Turing himself said, I believe that at the end of the century, or the end of the 20th century, when we’ll be able to speak of machines thinking. So he thought machines could eventually think, and we may hope that machines will eventually compete with man, human beings, in all purely intellectual fields. So that does raise this definitional question, what does it mean to think? What does it mean to know? What does it mean to understand? Well,

Juliet Floyd
I don’t agree that chat G P T hallucinates, for example, because in order to hallucinate, you already have to have the distinction between truth and falsity in place, right? A hallucination is false, but chat G P T doesn’t have that distinction. We have that distinction in relation to one another and the machine. It’s the same with the notion of knowing, if we know, then it has to be true, and with the notion of information, which has received a lot of throwing around in the wake of what Turing did, that’s also misleading. Is a piece of information true or false? We can’t quite figure that out now, so we talk about misinformation or disinformation, and we’re not quite sure. Do we mean the vehicle of the bit of information that’s being carried, or do we mean something that’s true or false? Now, Turing himself thought that intelligence was what he called an emotional concept, that is, it’s colored by human responses in the way that color words are. So yes, he predicted that we probably would be willing to say that a machine thinks. But the question is, what does that mean? And for him, the nearest he gets to talking about what intelligence is, is to say that intelligence consists in understanding the difference between different kinds of searches. That’s a lovely definition of intelligence, because some searches will be algorithmic and some searches will be biological, and some searches, as he insisted, would be cultural.

Ray Briggs
This is fascinating. So there’s a couple of things going on here. One is the idea that, like the computers of Turing’s time were all algorithmic, like you, kind of, you had a programmer who told them what to do, giving them a set of rules, and then sort of let those rules be executed. It seems like the kind of chat AI that we have now isn’t algorithmic in quite that way. You give it a bunch of training data that it’s sort of like, I might say, perceives you might you might dispute whether it’s really perceiving it, and then it kind of tries to generalize in a way that it’s human programmers don’t understand. Does that make it kind of fundamentally different from the computers that Turing was talking about.

Juliet Floyd
Yes, it’s fundamentally different because it’s probabilistic. Now, Turing, already, in 1936 had the idea of an indeterminate machine, one that would act randomly at various points, and he wrote his dissertation on probability. So these are probabilistic machines. They’re not step by step. They’re not working in the same context as an algorithmic system of logic, although we notice now that the makers of chat GPT would like the machines to reason more. And so we’re seeing that the whole control of logic has to be somehow brought into the context of these large data machine learning machines that are capable of predicting so much. But really it’s not surprising to me that chat GPT was created. I think it was surprising that it worked when it worked, but most of what we say is very predictable. This is something ordinary language philosophers saw in the 50s, and in fact, language wouldn’t work unless most of what we say is predictable. So it’s no surprise that if you have a billion parameters to me, you can do it. My father told me they’re going to get it back in the 1960s.

Josh Landy
You’re listening to Philosophy Talk today. We’re thinking about Alan Turing and the limits of computation with Juliet Floyd from Boston University. Juliet, everything you say about your dad reminds me of my dad, computer scientist. And yeah, I was soldering, or soldering that we say in this country, those boards. And it was such a beautiful and exciting time and and it’s fascinating to see so much of those predictions and and hopes and perhaps fears coming true nowadays, I wanted to pick up on something you were just saying about, essentially, about the limits of computation. There’s a really interesting quote from Turing that I’ve I came across recently, if a machine is. Expect it to be infallible, it cannot also be intelligent. So that that goes against, I think one of the ways that people are currently talking about machine intelligence, that we were expecting machines actually, in a way, to be better than human beings at reasoning, not just good at reasoning, but in fact, impeccable at reasoning. What did Turing mean by this? He thought, ultimately, if we do have a thinking machine, if we do have an intelligent, what we would now call computer, it would not be infallible. It would make mistakes. Why is that?

Juliet Floyd
Well he knew very well from the time of 1936, building on work of Kurt Godel, he knew very well that there were limits to what algorithms could do. So, for example, you can’t program a theory with a machine to divide the truths from the falsehoods, the ones that follow by logic and the ones that don’t. You can’t do it. So there’s always room for there to be things that the machine can’t actually prove. And Godel had shown this for, of course, the theory of arithmetic, back in 1931 so in that quote, what Turing is saying is, you can’t expect machines to be infallible, because there are mathematical limits on infallibility. When you reach a certain degree of complexity in your theory, you’re capable of self reference, and at that point, it explodes in complexity, and you cannot expect to be able to control it. So for this reason, he didn’t think that the idea that humans surprised us in conversation was really going to be the thing that would differentiate humans from machines. And he said explicitly, and he was one of the first to say the machines he worked with really surprised him. He said, I’m often incredibly surprised at what they do. And the deeper idea is this, it’s computational irreducibility. For the first time in the history of science, we see that you can have very simple equations. And it used to be that we thought, well, you plug in values to those equations, and you can predict how the world is going to go, but after touring and go, we see, no, it isn’t like that. You can put in values and you’ll be really surprised about the behavior of that machine after 30 or 40 steps. And there’s no way to predict in advance what the shape of the behavior of that machine will be?

Ray Briggs
Yeah, I want to pick up on this. So one of Turing’s Insights is, first of all, you can ask machines questions about themselves. You can describe a computer program in a programming language and then ask another computer, like, what happens if I were to run this computer program? And then another insight, the computer will sometimes not be able to give you the right answer, like, it sometimes won’t be able to give you any answer. It’ll just get stuck. Are humans better than computers in that way? Or are we just as fallible as computers, like, are we doing something fundamentally different, and if so, like, why?

Juliet Floyd
Well, we surely are fallible. I mean, to be a human being is to live with the mistakes you make, and we all make mistakes. On the other hand, I think Turing would agree with this. I’m going out on a bit of a limb here, but we get the distinction between right and wrong and truth and falsity as we speak to one another and as we pursue science. And when Turin was talking about the future of AI, he said he thought there would be three kinds of searches that would be very important in the future. The first he called the intellectual search. That’s the search for algorithms to give determinant and objective answers to things. The second kind of search he called the biological search, because he knew about Watson and Crick, and he could see the future of computational biology, where we use computers to look at the structure of DNA and RNA. But the third search, he called the cultural search, and he said, Look, an average human being doesn’t know very much, and you never have any intellectual power if you just work on your own. What you need is about 20 years of working with other people. And so the last search that he thought would be most important in a world of AI was what he called the cultural search, and that, he said would be a search among the human community as a whole, across the whole world.

Josh Landy
Okay, so we were talking about the difference, if any, between machine intelligence and human intelligence. That brings me to a slightly broader question about Turing’s picture of the human mind, right? So one question is, can a computer be intelligent? Another is, you know, does human intelligence look like a computer? What did he think human mind slash brains look like? Are we just like sort of zillions of tiny bots strung together and somehow, through emergence, consciousness arises and thought arises and creativity arises. Or what was his picture?

Juliet Floyd
Well, he certainly thought about what was called, at the time, electronic brains and neural. Nets. And he could see that with probabilistic use of machines, you could begin to string them together, as you say, and get more and more complex behavior. He was very interested in the point at which life or something looking biologically more sophisticated would emerge. So I think he was fascinated by this whole question of what’s alive. But He also emphasized that the machines would need what he called discipline and initiative. So initiative means that the machine is capable of putting together things that, as it were, knows, and then drawing inferences from them. But also needs discipline and notice today how many human beings are already involved in helping chat GPT to get better and differentiating things that are thrown through the web. So I think it’s really the integration of human and machine intelligence which is what he was looking for. And in that regard, you really have to pay attention to human human interaction in the presence of these machines, and that’s what the cultural search is all about. He said he was very influenced by Wittgenstein’s lectures. What influenced him was the whole exploration of types. Now, a type would be a category, okay? So you have to have various categories in a data set or in any computer programming language. And Turing thought that software and the development of language and phraseology and how we frame things in language was really, really going to be the central force in the development of computing. And actually, he was right about that. He complained about the Americans. He said they have huge hardware and they throw hardware at their problems, but software is the most important thing, and he wrote the first software manual for the Manchester baby.

Ray Briggs
So it seems to me like right now, there’s some kind of like special sauce that humans have that like we’re not predictable. Computers are not predictable, but we’re unpredictable in different ways. Do you have a like summation of what that special sauce is? Does anybody know?

Juliet Floyd
Well, I would call it our forms of life as human beings. You know, this is partly biological and partly cultural. And for Wittgenstein, he only gets to the notion of form of life after he reads Turing, which I think is very important. But forms of life are these local routines that we have, and they’re dynamic. They can change, whether you’re in a context where people greet each other as they walk by or not. And this permeates everything, all the scientific laboratories, all of everyday life, all of politics. What is pedestrian space? It’s something that’s embodied that you walk through. So these are things that we’re very, very familiar with, but we often overlook their complexity. And so there’s a lot of work to do as human beings and philosophers and humanists in trying to articulate what we find valuable in those things. And I think what the Turing test and all of these recent developments is going to do is to get us to talk about what really matters to us, what do we really care about? And that’s a very human thing. And I don’t know where this whole thing is going. I mean, some people already prefer sex robots. I don’t. In my lifetime, I probably won’t. I wouldn’t recommend it. I don’t think everybody will, but we’ll have a lot to talk about, and that’s really what Turing was envisioning.

Josh Landy
You’re listening to Philosophy Talk today. We’re thinking about Alan Turing with Juliet Floyd from Boston University, editor of philosophical explorations of the legacy of Alan Turing.

Ray Briggs
What will AI be doing in the next 10 years or 20? Will you communicate with supercomputers or just strategize around them. How will they affect the way we see our own humanity?

Josh Landy
Machine learning and human limits—plus commentary from Ian Scholes, the Sixty-Second Philosopher, when Philosophy Talk continues.

BBC
The machine’s obviously not in the mood!

Josh Landy
That was the earliest known recording of music produced by a computer. It’s that very machine in Manchester. We were just talking about the Ferranti Mark One. It was captured in 1951 and program by a friend of Alan Turing. I’m Josh Landy, and this is Philosophy Talk the program that questions everything…

Ray Briggs
…except your intelligence. I’m Ray Briggs. Our guest is Juliet Floyd, and we’re thinking about Alan Turing and the limits of computation.

Josh Landy
So Juliet, as you mentioned earlier, Alan shearing is finally receiving the recognition that he deserves. During his lifetime, and the Turing test has become a household phrase. But what’s something Turing said that people should be paying more attention to these days?

Juliet Floyd
Well, we’ve emphasized the social context of the Turing test, and I think it’s important to remember that Turing thought that that was extremely important to the future of AI. He talked a lot about worries about bias, about racism, about treating machines as slaves, and the tendency to treat human beings as slaves. And he also made some very interesting remarks about loss of jobs and job displacement in the face of the increase in the use of algorithms. So he in particular, said the mathematicians are going to be very upset, because a lot of what they do is going to start to be automated. And indeed, this is a theme at most of the major mathematics meetings. Now across the world, will the kind of math done by human beings that we grew up with become artisanal mathematics, and most of it will be done by machines. We don’t know. So Turin predicted that, and he said people will be very upset. The mathematicians are bound to make up all sorts of ways to get in the way of these developments. And he said, I think this is a real danger. So Silicon Valley’s fear about people not liking self driving cars or being upset when people are hit by cars, that is extremely important. And Alan Turing really understood that the way the machines are perceived by society will have everything to do with how we develop the machines in the future.

Ray Briggs
So I’m wondering if there are some opportunities here for using these machines more wisely than maybe the most gung ho of Silicon Valley bros wants, like I’m thinking about mathematics. So I know people who just use computers to aid them in their mathematics as a tool, like computer assisted proof is a thing that that you need, like real, live mathematicians to study and understand and like, there are lots of uses of computers in art that don’t involve just replacing the artist with a cheap algorithm. Do you see opportunities for integrating AI into society instead of replacing people using it as a tool?

Juliet Floyd
Yes, and I think this is what we’re seeing, both with the universities and with the museums in the world and the art world, you’re seeing some effort to cooperate. The universities and the museums have content that, of course, people in Silicon Valley would like, as well as our philosophy papers, the publishing industry. So the question is, can there be smaller regions of chat, gpts, let’s say, for universities and other things, where we integrate and we’re able to sift through this vast flood of information and find out what matters to people? That’s what’s really important. And we can’t rely on the software companies to do that, because they have financial bottom lines, but we do, I think, have to rely on us. So if you imagine a large insurance company where calculators were introduced in the 1950s it’s important that the calculators were introduced and so that certain skills of adding in your head were not so important. But what’s equally important is, how did you organize the people in the insurance office into various activities so that they could each enjoy their work and participate in a creative way? That’s an equally important issue.

Josh Landy
So I guess I still have a couple of worries, because obviously, there are plenty of neutral cases. Recommendation algorithms don’t seem to be doing much damage. There are plenty of positive cases, like alpha fold self driving cars we talked about. But there are also some troubling cases, like the AI that, you know, basically rips off the work of artists, and I believe there’s currently a lawsuit ongoing about that and other, you know, obviously the software that people could use to cheat on their papers and and things like that, it’s and it really, it’s not clear to me how exactly we really, I mean, it doesn’t seem like we have that much say in what goes on. The other thing is that the calculator analogy is often wheeled out as a kind of rhetorical move to say, don’t trouble your pretty little head about this, Josh, it’s just like calculators we all got over that turned out you didn’t need to be able to add numbers in your head. But when it comes to writing right, people want that analogy to extend to writing right in the same way chat GPT, it’s not a problem, just like a calculator, and just as calculators helped us to vault into even more interesting mathematics, so chat GPT will help us to vault into more interesting thinking in writing. I’m not convinced that’s the case. I wanted to ask you if you think that’s the case. And also. If you think, Well, you know what we can do about it, if we’re troubled?

Juliet Floyd
Well, I think it’s an excellent question, and writing is a key issue for us. We who teach philosophy, I have my students writing journals now by hand. They have to write by hand, and at first they resist, and I tell them, I’m very old, but when I was young, people kept journals, and it could be really interesting to see what you were thinking 10 years ago. The students love it. So there’s a case where a certain activity of the hand and the eye are very important for learning how to express yourself and take ownership of what you do. Obviously, if you’re doing boring memos for a large corporation, use chat GPT. I don’t care about that. Arenas I’m more worried about include mental health. So for example, therapists, they are being automated up the wazoo with having to write reports on patients. They spend more time doing that because the hospitals want the data, they spend more time doing that than talking to the patients. And I just heard a story last week about a friend of mine who’s a psychiatric nurse. She was told she was not supposed to be writing down verbatim descriptions of what the patients said. These are very, very sick patients because of privacy issues. So this is something we can do something about. I think we know that it’s better for the therapist to be able to spend half an hour with the patient rather than 15 minutes. And we know that really, many of these institutions under financial pressure are seeking the bottom line they’re seeing. You know, how few copy editors? Can you have? How few therapists can you have? It’s a big experiment to see how quickly you can make people go and I think it’s very important for us to say, no, no. The people who are caregivers care about their patients. They need time to write those memos, and it matters. So that’s why I always come back to what matters. You know, you can’t prove that it matters, but things matter to us, and you don’t need a theory of that, and that’s going to be so. So we need to speak out about what matters to us and be very clear about that and open about all these problems.

Josh Landy
Okay, so let’s switch gears for a second. We’ve talked about Turing’s life. We talked about his influence on computer science, his theory of intelligence, what he thought about the human mind, and also his theories about how these computers could affect society. But he was also working in a very different field, as well, the field of morphogenesis, biological processes by which organisms developed their shape. Can you tell us a little bit about that?

Juliet Floyd
Yes, that again, gets us back to forms of life, doesn’t it? Because, for example, the petals of a flower or the rind of a pineapple have a certain kind of regular structure. And if you represent those with numbers, they follow a Fibonacci series, which suggests something about the mechanisms, biologically and morphogenetically that are going on in the creation of these symbolic forms. I might also say that it says something about what we respond to because we are evolved so that we find certain kinds of patterns to be very pleasing. And that has to do with why someone might want to do mathematics, for example, because it’s beautiful, or you might want to create things as a human being in art because it’s beautiful, or because it’s simple. So I find all of these things very moving that they were connected. And Turing came up with a very few number of equations to try to characterize the phenomena he was working with. I think he said they’ll be these six. Should be just about it. I think they since then, they found one more. But he was very far thinking on that score too, and I’m studying that material now. I’m not an expert on it, but I find it absolutely brilliant to be thinking about how the mathematical structures and the forms of nature actually unfold evolutionarily.

Ray Briggs
Juliet, we’re almost out of time. So I’m going to ask you for sort of one big piece of advice from Turing. So we’ve seen that the way that we organize ourselves as humans, into cultures, and the way that we interact with machines is really important. What’s something that fans of Turing can take into the 21st century?

Juliet Floyd
Well, I think that it’s all local. I think it has to do with speaking to people face to face. Turing loved to talk philosophy with people, and I think politically, we need to organize very locally and focus in that kind of way so it’s human, human interactions, and paying attention to what the machines are doing to us, how they’re changing things, what we care about, in terms of the history, what was important, what’s not important, things will shift. My prediction is that there will be a great variety of responses to these machines. I think that. Cannot treat this as a scientific question where there will be one answer, there will be an increasing variety of forms of life. We’re going to have to learn to live with that in democracies around the world. But the best thing we can do is to be human and talk to one another and work on our phraseology, as he would have said.

Josh Landy
Well, thinking of being human and talking to each other. It has been an absolute joy spending time with you today, Juliet, thank you so much for joining us.

Juliet Floyd
Well, I’ve had an absolute pleasure too. It’s a wonderful show, and I’ve learned a lot.

Josh Landy
Excellent, I’m glad to hear it. We certainly have. Our guest has been Juliet, Floyd, professor of philosophy at Boston University and editor of “Philosophical Explorations of the Legacy of Alan Turing.” So Ray, what are you thinking now?

Ray Briggs
Well, I did just want to name the thing that was done to Turing at the end of his life, because it is still a problem today. It was conversion therapy. Basically the government was trying to make him not be gay anymore. And this still happens to queer people. It happens to trans people, and we should be vigilant about trying to outlaw it and stop it from happening.

Josh Landy
Yeah, I mean getting the pardon very late and getting Touring on a 50 pound note, that’s a good start. But hey, we can still do more. We’re going to put links to everything we mentioned today on our website, philosophytalk.org where you can also become a subscriber and question everything in our library of more than 600 episodes.

Ray Briggs
Now, could a computer imitate this? It’s Ian Shoales, the 62nd philosopher.

Ian Shoales
Ian Shoales… The German mathematician David Hilbertgave the math world Hilbert spaces, via which 2 dimensional math is used to work in multi dimensional spaces.  This is space in the hippie corporate vernacular sense, as in telling people if you join the army you’re in the warrior space, even if it’s a desk job!  Or if you’re a bankrupt investor you’re in the crypto space.  In the early 20th Century Hilbert was worried about the future of math, what he called the Entscheidungsproblem, aka the “decision problem,” is math reliable?  Is there an algorithm that can tell if a proposition is provable?  To a guy like me, who doesn’t even try to balance a checkbook any more, this seems overly complicated.  Rather than solve a problem, you look to acknowledge the existence of a solution, without necessarily finding one.  How would you even know?  Well, there’s all kinds of problems, as Alan Turing saw.  So he devised the Turing Machine, which proved it’s impossible to prove that a proposition is provable.  Anyway he laid out the principles of modern computing, and at the same time revealed its limitations, even when unhindered by time, memory, and actual existence.  The Turing Machine was just another thought experiment, after all.  It existed to factcheck the output of a second imaginary computer, there to find a solution to the decision problem, and failing, so IT couldn’t exist. Today we have the very existential artificial intelligence, which spins its own algorithms out of fairy dust and mortal dreams, and will continue into eternity, or we all return to dust.  Prior to the 20th Century, math geeks had to rely on pencil stubs and scraps of paper.  The invention of the blackboard helped a lot.  The golden age of computability loomed.  Keep in mind, math’s difficulties aside, algorithms as we know them weren’t invented until 1928.  Alan Turing waded into this at the same time perceptions of evolution, of empires and culture, human nature, of the very nature of life itself, were bending reality in ways that had philosophers clapping their hands with delight in anticipation of what the world would come to know as existential dread.  Welcome to the 20th Century, buddy!  Turing wound up helping to preserve what remained of the British Empire against the onslaught of Nazi coded messages during World War II.  What thanks did he get?   Popped in the fifties for homosexual acts, he was castrated chemically, gained weight, grew breasts, and committed suicide via poison, or least was careless with cyanide.  The point being that perceptions of science and homosexuality have changed a lot since I was a kid.  Gay people are going back into hiding I think.  Scientists, who once gave us vaccines and bombs in equal measure, have given way to mistrusted bureaucrats, who will be the source of zombie and robot armies in the not too distant future, if movies are to be believed, which they are.  We still love the computers though.  Beginning with calculators and pong, oh wait that was pretty far along wasn’t it? Now we have AI, and it can write books, but you know so can a thousand monkeys, or self-important transphobes, or at least J.K. Rowling, or ghost writers, or me, as far as that goes.  Of course, much of what we read and see, seems like it was made by computer these days.  Like old time radio, or commercials, or newscasts, or weather alerts, there’s a voice conveying information, but it doesn’t correspond to human voices we know.  It’s an announcer; there’s nobody telling the story, it’s presented.  There’s just you reading or watching or listening.   And now it will all be created, by artificial intelligence, with the proper prompts, for whatever topic we the audience convey algorithmically to the intelligent equations that will eventually lead us consumer citizens to bliss and multiple addictions and never leaving the house again.  I may be too premature.  Right now, AI is just irritating.  Weird hippie art on the internet. Porn with too many fingers.  But AI is incredibly wasteful.   It’s the equivalent of driving a four wheeler up a mountain for six hours just to download a discount for pharmaceutical gummies from a website in Delaware or Serbia.  We can burn down a forest to put boobs on the Pope for a gag on a Facebook post.  Do I blame Alan Turing for that.  The Turing machine, you’ll recall, had to run forever for it to work.  Since it didn’t exist, well yeah, that’s easy.  Computers were fine when they weren’t around, but I’m starting to think it’s too much of a good thing.  The Turing Test was supposed to see if we could be fooled into thinking a machine might be human.  Well, that ship has sailed.  Maybe we don’t think machines are human, but I personally have had conversations with robo callers.  Vacuum cleaners, smart phones, and iPads have all engendered more affection from humans than I ever have.  What does that tell you?  Don’t ask me. sk your girlfriend, Siri. I gotta go.

Ray Briggs
Philosophy Talk is a presentation of KALW San Francisco Bay Area and the trustees of Leland, Stanford Junior University. Copyright 2025.

Josh Landy
Our executive producer is James Kass. The senior producer is Devin Strolovitch. Laura Maguire is the Director of Research.

Ray Briggs
Thanks also to Pedro Jimenez Merle Kessler and Angela Johnston.

Josh Landy
Support for Philosophy Talk comes from various groups at Stanford University and from the members of KALW local public radio San Francisco, where our program originates.

Ray Briggs
The views expressed (or misexpressed) on this program do not necessarily represent the opinions of Stanford University or of our other funders.

Josh Landy
Not even when they’re true and reasonable. The conversation continues on our website, Philosophytalk.org where you can become a subscriber and question everything in our library of more than 600 episodes. I’m Josh Landy.

Ray Briggs
And I’m Ray Briggs. Thank you for listening.

Josh Landy
And thank you for thinking.

The Imitation Game
Can we find a clue here that we can build into Christopher? Who’s Christopher? He’s my machine. You named him. Is that a bad name?