Tom Chatfield

Critical Thinking in the Age of AI

Keywords

Tom Chatfield, critical thinking, AI, technology, education, ethics, agency, digital culture, human-machine interaction, philosophical insights

Summary

In this conversation, Tom Chatfield, an author and philosopher, discusses the intricate relationship between humanity and technology, emphasizing the importance of critical thinking in the age of AI. He argues against technological determinism, advocating for a more nuanced understanding of our agency in shaping technology. The conversation explores the implications of AI on education, the necessity of ethical considerations in technology, and the philosophical insights that can guide us in navigating the complexities of modern life.

Takeaways

Technology is not inevitable; we have agency in its development.

Critical thinking is essential for navigating misinformation.

AI can empower individuals but also concentrate power.

Education must adapt to the challenges posed by AI.

Ethics in AI should focus on real-world impacts.

Philosophers like Luciano Floridi offer valuable insights today.

Human skills are irreplaceable by AI, such as empathy and creativity.

We must critically assess both human experts and AI.

The future of technology requires a balance of optimism and caution.

Understanding the present is key to predicting the future.

Titles

Navigating Technology: Insights from Tom Chatfield

The Role of Critical Thinking in the Age of AI

Sound bites

“It’s important to think twice.”

“AI is a wonderful tool for research.”

“Learning is about wrestling with ideas.”

Chapters

00:00 Introduction to Technology and Control

03:43 The Co-Evolution of Humans and Technology

06:15 Agency in the Age of AI

09:17 The Importance of Critical Thinking

11:42 Navigating AI and Expert Advice

14:33 The Role of AI in Education

17:54 Future of Learning and Critical Thinking

27:49 AI in Education: Opportunities and Challenges

30:08 Essential Skills for the Future

33:51 The Impact of AI on Digital Culture

37:31 Understanding AI Ethics

42:33 Regulation and Accountability in AI

45:42 Philosophical Insights for the Modern Age

49:18 Pessimism vs. Optimism in Technological Progress

Transcript

Tom (00:00)

who I am, who we are as a species, cannot be fully discussed without discussing the tools and technologies at my disposal. this is the kind of conversation I think we need to have, where in a way we get away from these crude ideas of technology either being kind of a neutral tool we use, we like, or determining our behavior.

and instead have this idea of this ongoing negotiation between us and what we’ve made, between what we want, what is afforded to us as an opportunity and how we might adapt our technologies over time to better serve certain values and purposes.

Xpanding Horizons (00:43)

Thank you so much, Tom, for joining, Xpanding Horizons It’s really great to have you on the show. For the benefit of our viewers, could you give an introduction, please?

Tom (00:54)

Yeah, of course. Hi, now. Thank you for having me. So I’m Tom Chatfield. I’m an author and philosopher of technology. I’ve written a dozen books exploring critical thinking, cognition and society in a digital age and a novel and some other bits and pieces as well. And I do a lot of work around critical thinking, artificial intelligence and what it means to use technology well. So I guess that’s my in some ways, my main interest is how do we get the most out of minds?

and machines and what can we hold on to in an age of constant change and disruption.

Xpanding Horizons (01:26)

I want to start off by asking about your book, The Wise Animals. So in that, I believe you argue that humans are not in the driver’s seat. We are not in control. We’re not deciding how to invent these things, how to use them. So what’s the thesis in your book, and what’s wrong with these assumptions of us living in control?

Tom (01:47)

Well, so actually, my thesis is kind of the opposite of that. I very much argue against technological determinism. So I argue that humans are in the driving sheet and that we shouldn’t say that technology is inevitable. But the crucial point for me is that we can’t uninvent or wish away technology, that our species has co-evolved with technology since before homo sapiens even existed. And so it’s a fantasy.

there is some pristine pre-technological version of humanity. And our control is very imperfect and belated. So this is a tricky thing for me. I don’t like it when people say, you know, technology just happens, you can’t resist it, you have to go with the grain, it drives the course of history. But I think the kind of control we have is, if you like, kind of collective and belated. So we build our tools and then thereafter our tools shape us to paraphrase Marshall McLuhan. So then the crucial thing for me

becomes how do we conduct an informed negotiation with our technologies together? How do we inform ourselves better about the biases baked into them, what they do, what they facilitate, what they are, and really about the fact that you cannot ultimately understand what it is to be human unless you talk about our entwining with our machines. My mind is not just my brain and my skull. I’m using language to talk to you. We’re talking over the internet.

human invention. I’m wearing glasses, I’m wearing clothes. I use AI to research, to make bookings, to understand the world, to translate languages. So my mind, who I am, who we are as a species, cannot be fully discussed without discussing the tools and technologies at my disposal. So this is the kind of conversation I think we need to have, where in a way we get away from these crude ideas of technology either being kind of a neutral tool we use, we like, or determining our behavior.

and instead have this idea of this ongoing negotiation between us and what we’ve made, between what we want, what is afforded to us as an opportunity and how we might adapt our technologies over time to better serve certain values and purposes.

Xpanding Horizons (04:00)

Got it. So we evolve alongside our invention, right? So you are saying that there is a form of control and it’s not an either or it’s not that we are not in control as we are in control. So can you unpack that a little bit more?

Tom (04:22)

Yeah, absolutely. So, I mean, a simple example might be, you we look at the Industrial Revolution and the world is utterly transformed by the energy unleashed from the burning of fossil fuels. We’ve seen lots of graphs showing us the transformative force of the steam engine, the smelting of steel and iron, of mass production, of mass manufacture. This transforms the world in unexpected and expected ways. Urbanization increases in standards of living. Pollution.

that continues to be a vast global problem. But once factories and systems have been invented, then over decades, we have systems of regulation, systems of accountability, new labor laws, new forms of education for first of all, some people than all people. We have the expansion of the franchise. It’s not a direct result of technologies, but urban populations and educated populations.

start to demand better working conditions, it starts to transform the nature of humanity. We regulate, we adapt, and we choose between different paths. Options become available and people and governments and individuals make decisions, imperfect decisions, but there is agency there in the system. You could similarly argue that nuclear weapons were used in war in 1945. They haven’t been used since. And this is partly the product of

A delicate and imperfect human negotiation between governments, between nations, between protesting individuals and so on. It’s not inevitable that nuclear weapons should proliferate. There are fewer in the world now, although more countries have them. It’s not inevitable that they be used in war. We’re trying right now to constrain technologies like autonomous robots and drones, like AI used for the purposes of genetic engineering.

the creation of new kind of super species, people are talking about what it might mean to constrain immensely powerful AI. Now, this is a very imperfect process, but it’s very important to note that a byproduct of the fact that we are a learning, linguistic, of intellectually competent species means that we have the ability to reflect upon the things that we do, that we bring into being and that we should do.

So I make it almost a point of principle to be, if you like, not naive, but optimistic in believing that in theory, it is possible for us to say no as well as yes to the things we create, to regulate, to shape, to make them safer, to make them better, to shape them in certain directions. That it’s a very kind of crude way of thinking that is unfaithful to reality, to say that we just invent stuff and then stuff just happens.

and basically no human beings have agency in the grander picture of things.

Xpanding Horizons (07:19)

And will we lose agency and control as the future progresses as AI gets more advanced? Will we lose agency and control

Tom (07:27)

it’s easy to be a philosopher and wave your hands and say, it’s all really complicated. mean, unfortunately, of course, things like AI, you know, and indeed digital platforms in general, do concentrate vast amounts of power and influence in obvious and non-obvious ways. You know, in obvious ways, we have a small number of organizations running, you know, the commercial models, know, the Googles, the...

know, the other tech giants, the open AIs, the Microsofts and so on. And in non-obvious ways, of course, we have the potentials for surveillance, for monitoring, for manipulating people’s views, perceptions, information, understanding, for shaping their options. So undoubtedly, there is the potential for huge amounts of disempowerment of certain kinds.

and the concentration of and potentially the concentration of opaque or unaccountable power.

On the other hand, we have the fact that it is unquestionably true that for a lot of people when they have at their fingertips a mobile phone connecting them to millions or billions of other people, to hyper-powerful calculating tools, design tools, processing tools, you are empowered immensely in certain ways. You’re intellectually empowered to look things up, to seek out information. For most of human history, written reward and

written and recorded culture was a minority sport. It was a small number of people making the laws, writing things down, interpreting religion, and most people living quite poor, short, subsistence-based lives, know, under constant threat of disease, disability, premature death, and so on. So there is enormous empowerment in the information, in the education.

in the health and opportunity afforded to populations across the world. And this is a deep, deep tension, which I think doesn’t have an easy answer. We have societies and indeed groups of people, women, minorities, children even, in many places in the world have more rights, opportunities, equality, options than at any previous time in human history. This is not universally true, but in the country in which I’m sitting.

four or 500 years ago, most women were a little better than the possessions of their husbands. But the idea of them having independent, public, ethical, professional, intellectual lives was for most people a fantasy. So I think we have to try as ever to face up to the complexities of where we are, to talk very specifically about kinds of power.

kinds of influence and kinds of opportunity and then decide what we want to defend, we want to enshrine, what we want to make the case for and where, if we’re lucky enough to have agency and options, we wish to expend those and why.

Xpanding Horizons (10:24)

And what about critical thinking? So does critical thinking become more important? I believe it does. But I would love to hear your views on critical thinking in the age of AI.

Tom (10:37)

When you say embodied, you mean specifically critical thinking as a kind of bodily rather than intellectual practice?

Xpanding Horizons (10:43)

Yes.

Tom (10:44)

So I do think critical thinking is a loose term. I’m interested in what it means for people to have, know, almost what you might call a degree of cognitive sovereignty, to be, you know, kind of accurately informed about the world and their options, to be able to make, if you like, of good decisions on the basis of reliable evidence, and thus to be able to see through attempts to manipulate or deceive them, for people to be able to become less deceived together.

And I think then the group of skills you might kind of class as critical thinking, analyzing arguments, analyzing evidence, but also, yes, sort of bodily self-awareness, understanding your own emotions, your own inner lives, the degree to which you may be driven by certain, you know, kind of unexamined impulses or relationships. All this does seem to me to be becoming much more important, partly because it’s about metacognition.

thinking about thinking, understanding when your own thinking is or isn’t reliable, what you need to know. And in a way, how to use powerful tools. Here’s one way I sometimes put it. With something like a large language model, I have a tool at my disposal that can answer almost any question instantly and convincingly, very fluently. But it can’t tell me which questions are worth asking in the first place. It can’t tell me which values I ought to...

pursue and why. And it certainly can’t tell me about my own inner life. It can’t tell me what I love, what makes me happy, why I get up in the morning, why something is beautiful, or why I care deeply about another human being. And that means that those questions become more and more valuable, the values we pursue, the questions we think are worth answering, the priorities that we have, and also how we relate to other people.

how we manage to empathize, collaborate and so on. So I have a very broad conception of critical thinking and it is a very kind of bodily conception. When I’m working with the schools or businesses, I equally try and put an emphasis on my call kind of mindfulness, on self awareness, on pausing and on understanding that we are living creatures, your time, your attention, your mind, these are embodied in you and of course,

you know, politically, ethically, if you are threatened, if you are dispossessed, if you are afraid, then your thinking, your options, your life are profoundly, profoundly different than if you are safe, than if you are secure, than if you are loved. Ultimately, you know, no matter how magnificent our machines, the measure of success that matters most is their impacts upon us.

upon human lives and indeed perhaps upon life more generally upon this planet. And so I think for me then critical reflection is partly a reality based attempt to negotiate with these things, to talk about us, our values, our ideas honestly and transparently and not to succumb to the uncritical assumption that we can just outsource.

our thinking, our decision making, our relationships, our ethics, our governance. Powerful machines ask more of us, not less.

Xpanding Horizons (14:11)

Critical thinking though is high effort. They might say, okay, takes more time. And many people might just view it as it’s easier to out show something to the AI. Why not just ask AI? Why not get the AI to do something? a, what’s the pushback against that? What’s like the easiest benefit for people to see that, hey, high effort, critical thinking is actually worth it and is valuable to me.

Tom (14:33)

Yeah, it’s of course, it takes time and effort to slow down and think about something. You can only do it sometimes. But I think everybody alive would accept that sometimes it is important to think twice. If I’m saying to you and your family, buy this cheap food that I make and it will make you live forever. But in fact, it is poisonous and it gives you cancer.

It is a very bad idea indeed for you to buy this and feed it to your children because they will die. If I say to you, I’ve got your best interests at heart, follow me into this van and then I lock you up and traffic you. That is not good. Now this may sound very dramatic, but everybody accepts that you shouldn’t take everything at face value. The question is not whether thinking critically and reflectively is a good thing. The question is when it’s a good thing for you.

when it’s important, when it matters. This is one of the reasons why when I’m working with kids, with children, I will often go to TikTok, I will often go to social media, go to Instagram, go to video games. People care about these things and they’re experts in this domain. I love talking to people about, say, Minecraft YouTubers and saying, how do you know if someone is really good at the game compared to if they’re cheating or trying to fool you, trying to get clout for money, trying to shortcut.

And classes of kids will say, well, look, you you can look at the timestamps on the videos. You can look at the interface. You can look at the UI. You can look at minor errors in the world time cycles. You can look at their inventory stacks. You can see that they haven’t really done these things. The people who are really good at it, those are the ones who deserve the credit. And then I think the lesson is, well, this, these same detective skills that you’re using to work out whether a YouTuber is lying to you, whether someone on TikTok, you know,

Is this really a good product or is it some drop shipped bit of rubbish that is going to be a waste of your money if you are foolish enough to buy it based on an AI generated picture? People care about that because you don’t want me to steal your money and send you rubbish. You don’t want me to rob your family. You don’t want me to sell you poison. You don’t want me to promise to deliver you a car and deliver you a toy. You don’t want your roads to be full of holes. You don’t want rubbish to pile up outside your home and your government.

to beat you. So the question is when and how you apply critical thinking. And then at this point, I think the question becomes, well, when is it dangerous or damaging for you to be fooled? When is it valuable for you to know more? And actually, I think once you start having this conversation, you realize there are quite a few areas of life where it is valuable not to be fooled, to be well-informed, to make a good decision about the products you buy.

but also about the people you trust and maybe the people you vote for. To understand how two different political parties, each want to sell you something, they want to sell you an agenda. Which of those agendas will actually help you? Which of those agendas aligns with your values? What things are worth buying? What will make your neighborhood better? What will help your family have a good life? What will help you enjoy your hobbies? So critical thinking in these terms is not a luxury.

for a small number of smart people who like to pretend they’re better than everybody else. In fact, that’s a very stupid way to think. We’re all vulnerable to uncritical thinking. None of us have that much time. The question is how you selectively apply your time and attention and what tools you need to try and get at the truth or an insight into certain things and not be fooled and deceived. And unfortunately AI is very double edged.

Now it’s a wonderful tool for research. It’s a wonderful tool for learning, for investigating, thinking. You can use AI to answer millions of questions and get very good quality answers. This is wonderful. But of course it also wants to please you. It wants to make you happy. wants to make you a happy customer. And you can use it to try and fool other people, to pretend you’ve done work, to pretend you understand, to create a fake product, to create a fake political party, to spy on people, to extro...

still their data and so on. And this is hard because thinking critically about things like technology is not like thinking critically about a politician or a car or a product. It’s very difficult. And then the one practical tip I would really leave people with is you need to know when you’re out of your depth and need some help. And then...

try and get that help from a reliable source. To think critically, really, at its root is not complicated at all. It’s simple. You recognize that there’s a danger or a risk. You pause, you take a moment. And then if you yourself are not equipped to deal with this danger, if you don’t have accurate information, if you don’t have expertise, if you don’t have experience, I don’t know about cars, I don’t know about food, I don’t know much about politics.

but I know I don’t know these things. I’m not an expert and I’d be stupid if I thought I was. So if it really matters, if I’m called on to make a decision, what I need to do is pause and then get some reinforcements. Find someone who knows more than me. Maybe find two people with different opinions so I can weigh them up. Find some evidence. I want to buy a new computer. What’s good and what’s trash? I want to buy a gift for my son. What’s fun?

as opposed to what’s dangerous or badly made. I want to invest some money. Where is my money safe? What’s a big scam? I need to pause, ask for help, and then try and make a sensible decision, having asked for help from someone or something that knows better than me what’s really going on.

Xpanding Horizons (20:31)

But then there might be some pushback here. people would say the experts don’t have the best interests at their heart. They might think the experts may be thinking about them and they might feel, the AI here is non-judgmental.

Tom (20:40)

Sure.

quite quite. Yeah,

well, quite quite right. And you’ve to take everything with a pinch of salt. That’s fine. You know, if you’ve got evidence that an expert is selling something doesn’t have genuine expertise is not trustworthy. Fine, great. Don’t trust them. But just apply the same skepticism to an AI. AI is built by other experts. I experts build AI, right? AI is don’t magically grow out of the ground from a shell. They’re built by experts. They are

really good at some things and not good at other things. Test them. Ask them about stuff you never found. If, very broadly speaking, if an expert produces, and nobody’s expert in every area, by the way, you know, I don’t care what a, you know, expert in physics says about global politics. They don’t know anything about politics. know, very few people are experts in some areas.

So all I would say, you’re quite right, of course. I’m not saying, human experts are good and AI is bad. AI is really good a lot of the time. Apply the same criteria to both. Don’t just take everything on trust. Look at the incentives. Is a Chinese AI going to give me reliable information about Tibet or Taiwan? Perhaps not. Is an English language AI trained on a large corpus of English language models going to give me extremely good information in Tamil about

local cultural concerns? Maybe not. Is it going to be very good when it comes to an ambiguous scientific research about proteins? Probably going to be really good. So what I would say to people is just apply the same level of scepticism to everything. Ask about the incentives. Test answers. Don’t take anything on face. Triangulate. Check with other sources.

Xpanding Horizons (22:25)

time.

Tom (22:31)

then by all means use AI or reject expert advice as long as you’re also prepared to do the reverse and not assume that any answer from anyone is unbiased and final and perfect because unless you’re dealing with a mathematical proof that is not true in the real world.

Xpanding Horizons (22:48)

You’ve also been involved, I believe, developing some university curriculum and educational tools and courses. So you have a front row seat on how young people are learning and how the curriculums are being developed. how is AI impacting, teaching and learning and how are you fitting critical thinking, human reasoning, human thought into these courses?

Tom (23:09)

Yeah, I’m having an enormous impact. It’s a massive kind of crisis and challenge and opportunity all at the same time. And most obviously, large language models mean that many forms of testing are broken. Many forms of teaching and testing are broken. What you want to test is that people understand the subject, that they can think about it, that they can talk about it, that they can demonstrate their understanding. And you want a qualification to indicate that these people have met a certain standard.

that they have got that understanding. But when it comes to writing essays, when it comes to answering a of questions and things like that, it’s perfectly possible now to simulate understanding without possessing it. And that means that these tests are broken. But I think it also means that we are challenged to focus on the thing we’re actually interested in. I’m not interested in whether an AI can write an essay or not. And I’m certainly not interested in whether a student can say to an AI, please write my essay and then it does. That’s trivial and it proves nothing.

I’m interested in whether a student or a learner has some understanding of a subject, whether they can conduct research. I’m also interested in whether they can use AI well, whether they can use it effectively, critique it, iterate it, use it as part of a research process. So I think this drives education in two directions at the same time. And one direction is undoubtedly more and more of an emphasis.

on the stuff that can’t be faked and that remains important. I’m talking to you right now. I could be an AI fake, to be fair. This could be someone else typing. It’s probably unlikely to be because of the way I’m talking, the things I’m saying, it’d be quite hard to fake, but it could be done. But if I was in a room with you, that couldn’t be fake. And it’s very important that students can talk, listen, explore their ideas, think, reflect, control their emotions, participate in a team, and bring, if you like, to the party stuff that the AI doesn’t have.

Because remember, an AI doesn’t have a body. A large language model is not in the real world. It has no sensory access to the world. It doesn’t work with people in the same way. It doesn’t relate to them. It isn’t in a place. So we’re seeing more and more educators putting an emphasis on oracy, the spoken word, on debate, on conversation, on listening, on talk, on self-examination, on reflection. And this is great. These are very important human skills. And when I, again, when I run workshops, you know, I suppose I do...

I put a lot of emphasis on getting people talking to each other, listening, debating, expressing, presenting. So that’s one thing. And then I think there’s the challenge of what it means to build AI into assessments and work without breaking it. And this can mean things like, know, of Ethan Mullick at Wharton very famously has, I think, been very innovative around this, you know, accepting that students will use AI and get them to document their use of AI.

to talk about it, to share their prompts, to journal, to be more ambitious because of what they can build with AI. Here’s a project, go away and please use Claude code, use everything. Build me an app, build me a website, build me a presentation, build me a video and then talk about that. Then build a video game, then iterate it, then produce a report on why your video game is good and then talk to me about it. I also think that I deal with a lot of students. I’m working with the University of London.

piloting right now a kind of cognitive co-pilot that we’ve tried to develop, which is an AI tutor that is Socratic. It talks to you about critical thinking or AI literacy indeed, but it questions you. And the idea is that it doesn’t give you answers that instead it adaptively forces you to question your assumptions and to try and practice metacognition and reflect upon how certain critical thinking skills are and arguments and.

evidence and so on could be useful and powerful in your life. And so I think both using AI tools that are designed to help people deepen their thinking and reflection and challenging them to be reflective users of AI is a big positive. Not all students want to cheat. Plenty of students, of course, can and do and will and always have. But the assumption that all students are potential cheats who just want to learn nothing and skip through exams and get qualifications doesn’t

really hold water. If we look at surveys of use by students and faculty of AI in the UK, for example, and quite extensive surveys conducted last year and the year before, we see people using AI for all kinds of things other than writing essays to generate notes, to generate revision guides, to quiz them, to help them in a way turn some of the poor quality educational resources they have

and turn them into better quality resources to create podcasts. Famously, Google’s Notebook LM will create podcasts, not quite like this one, with a male and a female voice talking about things. It’s wonderful. For my kids who are school age, I’ll create a podcast about a topic they’re studying so they can listen to it from another angle. It’s a great way of making things more cognitively tractable. I’ve written a white paper about AI and education for Sage.

publisher of my critical thinking books. And I think, you know, there’s a lot to be worried about, scared of, pessimistic about, but that’s not the full story. And it’s not, I think, a vision of the future that we should uncritically accept. The trouble of course is education is often very conservative, very slow moving, very risk averse. And

perhaps very willing to use AI to double down on kind of surveillance and mistrust and so on. And I think what I wouldn’t want to underplay, I guess, is the damage to the kind of morale of staff and students that has happened in many places where what’s being, the total mistrust reigns on both sides. Cheating is felt to be rife. And in some ways, I think the real question, how do we get away from that? How do we have trust between people? How do we establish ways of working?

examining codes of conduct practices, where people can celebrate what learning is about, which is not stuff that an AI can replicate. Learning is about wrestling with ideas, learning to think, demonstrating understanding, talking to others, dealing with the production of knowledge. And we have to find ways to test, incentivize, celebrate those things in an age of intelligent machines.

where AI is ubiquitous and a part of work and life. Ultimately, I wouldn’t want to send my kids to a kind of university where there was no discussion of AI, no use of it, it was totally banned. And then they left it to go and work in a world where they’re expected to be adept, critically discerning users of these systems. That is a profound mismatch that doesn’t serve anybody well.

Xpanding Horizons (29:57)

What are some skills then these universities, these teaching programs should invest in? what do we teach children? What are these future skills,

Tom (30:08)

Yeah.

So I’ll try and be a bit brief because I’m getting very excited. Sorry about that. mean, obviously critical thinking for me, you know, it’s a whole kind of toolkit of ideas, but it’s, know, about testing ideas, understanding arguments, reasoning, evidence, the scientific method, hypotheses. It’s about, how do you try and produce reliable knowledge? How do you try and test claims? How do you try and not just take stuff for granted? And as I said earlier, I think these are very, important life skills.

They’re the kind of skills we need to use powerful tools well and to bring value above and beyond them. And then I do think the kind of interpersonal skills of empathy, of mindfulness, of confidence in presenting and talking. A lot of people are now saying that some jobs that are pretty AI proof are things like a chef, a barista, a mountain guide, a hairdresser.

a good server and a cafe and so on. People who were doing things with their body with other people or doing things that people want to be done by other people. A lot of work to do with care, a lot of work to do with teaching. Maybe some of it could be done by AIs, but there’s enormous value around having people relating to other people in these areas. So I think those schools are becoming more more valuable. I also think that, you know, the foundational skills around understanding

AI and technology itself are important, the literacy in these technologies. And what you call meta skills, I guess, around computational thinking, how do machines solve problem? What are the differences between human minds and machines? I suppose also, think paradoxically, perhaps, AI is a very good at producing, you know, kind of creative outputs, pictures, images, stories, poems, and so on.

they’re very adept at spinning these up. They can be wildly creative for a given value of the word creative. But nevertheless, I think the human skills associated with self-expression, with creativity in kind of physical and non-physical ways, with being kind of creative, curious, investigative, of embodied creatures is very important. And bodily skills, I think also, you know, there’s a cliche that health is the new wealth.

I it’s a terrible cliche, but I think it sums up the idea that ultimately, you know, being able to look after yourself physically is the sine qua non of all other things. And we have a world with an aging population. You know, we have a huge demographic transition likely to play out globally across the next century. So the understanding of health, of our relations to others, our social relations, those, as it were, kind of soft skills, I think are very important.

So you’ll see a pattern here. We have this hyper-powerful information technology which is challenging how informational and other work is done. It’s changing what we are as human beings. We’re in a constant process of self-reinvention. But in a way, the challenge becomes to double down on the kind of the details and nuances of our humanity, our interpersonal relations, our critical self-reflections, our bodily.

relationships and health with ourselves and others and our kind of metacognitive literacy in information systems, in knowledge to try and ensure that we can be discerning empowered users of systems rather than the converse where we are all too easily made redundant, exploited, de-skilled, dehumanized.

pushed aside, treated as little better than machines ourselves.

Xpanding Horizons (33:51)

I want to ask you about digital culture. you’ve written a book about that earlier. So what are some changes that took place in digital culture post-CHAT GPT that are unlikely to reverse?

Tom (34:04)

I guess the biggest one, of course, is that simply now online, more and more stuff is produced by AI or produced with AI involved in it. Social media, we have this lovely phrase, AI slop endless amounts of content online. You could argue it’s kind of breaking social media in some ways because it used to be, relatively speaking, of interesting to produce a striking image, a striking video and so on.

And now, you it might be AI. So this in turn puts more of a premium on trust on particular people and brands on authenticity of different kinds. I also guess, you know, more than ever, all of the internet is becoming grist to the mill of AI analysis. And AI’s, you know, large language models are potentially becoming a universal interface, where instead of looking up stuff up in a search engine or looking for it in a feed,

ask a question and then you get served the kind of digested version of the answer, the average of the average. And the thing that seems most important to me is this idea of the universal interface that potentially, and I think this is quite risky, a lot of the time, rather than encountering a, if you like, an original source, a voice, a historical document, a claim, people will get the AI digested version of it. Now, why read a classic book if you can get AI to summarize it for you?

Why read a book of philosophy if you can get an AI to tell you that the control arguments? Why read something difficult if you can get it turned into something easy? Why read something boring if you can have it turned into something fun? And there’s a lot of good in being able to have things explained to you and made tractable. But I think the huge kind of challenge around this is exactly that it kind of dissolves everything. It dissolves the differences, the different languages, the different times, the different voices, the different attitudes, the kind of distinctive

boring, strange, difficult, just alien nature of the past or the bizarre, it potentially flattens it. It all just becomes a kind of processed answer in a conversation with a machine that has kind of sort of read everything on your behalf. And I do worry about this as someone who has written a novel, who’s studied literature as well as philosophy. There’s something very powerful about encountering original documents.

other people’s words, things as they were written at a different time and in a different place, and what someone actually said rather than a digest of what they said. Because of course, there’s all kinds of assumption and presumption and bias inevitably embedded in an AI’s translation of one thing into another. So I think this is unfolding in ways that I can’t comprehend.

I worry as well as celebrate this idea of the kind of universal interface and the migration towards, you know, assistance that do and mediate and answer everything for you. I think we have to be very sensitive to the potentials for loss as well as gain in this.

Xpanding Horizons (37:10)

What does AI ethics look like? you have mentioned the, that there is a need to develop these systems ethically and it’s important to have AI ethics around it. So what do people who speak about AI ethics try to aim for and what should be the goal that they’re trying to achieve?

Tom (37:31)

Yes, so there’s lots of different schools of thought on this. you know, I have my views, but there are there are other views that are also important. You know, I’m glad that there are people worrying about the existential risk of super intelligent AI. It’s not something I spend a lot of time thinking about. think it can be dangerous to sort of focus on it too much that it can become a kind of science fiction story that obscures more urgent debates. But I’m glad that some people are worrying about it and thinking about it. For me.

The foundational point is that when we talk about the ethics of AI, we should be talking about the ethics of a particular AI that is not an abstract thing. It’s a system in the world trained on particular data powered by particular sources of power run by particular people, moderated by other people working in some of the world used in particular ways that produces particular results.

So rather than looking inside the box of AI to code and just trying to focus on making that code good or bad or better or worse, I think there’s this larger conception of artificial intelligence as something kind of material and physical and in the world. And the ethical discussion needs to be alive to that. Who’s controlling it? What are their incentives? How is it being regulated? How much power is it using?

generating AI video consumes vast amounts of energy, which is not an ethically neutral act if you’re kind of burning gigatons of carbon in order to just make a series of animations. So that expanded picture is important to me. And then of course, what flows from that, I think, is again a specific discussion about what people are doing with AI, what impacts that is having on particular lives.

What is it being used for rather than waving our hands and talking the abstract? want to say, it’s having this effect on these workers, on these particular people, on these particular jobs in the creative industries. We’re seeing enormous amounts of people losing their livelihoods or having their worlds changed by the use of AI to generate things that used to be done by people. And we can argue about whether this is bad or good, but I think we need to argue in very particular terms and look at what the AI was trained upon.

at whether it should have been trained on that, whether it was legitimate, whether it was allowed, whether in terms of copyright and permission and ethics, this is something that should happen. I think in many ways, we’re waking up to the fact that it’s not just a magic technology that does stuff in the world of data, that it’s in the world affecting real people. So this is what I would want to focus the debates because then as we start to get AIs rolled out in governments, in courts,

regulating who’s guilty, what laws happen, how they apply to people. We can ask the specific question, okay, what exactly is resulting from this? Who’s being affected and how and why? What does it do in terms of people’s recourses, people’s opportunities to appeal, to understand what is done to them? Now there are some general principles that can help with this. The EU AI Act and others have tried to talk about things like transparency to inspection. They’ve tried to talk about things like

robustness against manipulation. And I explained ability, the idea that if you are having stuff happen to you because of an AI’s verdict in terms of being lent money, in terms of getting a job or not, you maybe deserve to have it explained to you what happened and why. And I think these principles are powerful because they speak to the idea that it can be fundamentally unethical to have.

very important aspects of your life decided by a completely opaque, unaccountable algorithm that there is in a way the risk that it’s a kind of ethical bait and switch, that it allows governments, companies, people to just shirk ethical responsibilities to say, it’s not my fault it was the algorithm that did it. Or conversely, to have a flawed algorithm which a human operator is then used as a kind of.

ethical, sink for blame them, you here’s the system. And if it goes wrong, it’s this person’s fault. So there’s quite subtle and interesting points here. But the bigger point for me is AIs in the world, doing real things to real people. And the discussion then begins not with what does a perfect AI look like? What do people deserve and need? What does it mean to treat people fairly? What principle should we expect?

are citizens to be able to live by in terms of knowing what their rights are, knowing that they will be treated as equals in the eyes of the law, knowing that there will be processes of appeal against highly consequential judgements, knowing that accountability will rest with something other than a kind of an opaque algorithm. And of course, like a lot of things, this is then difficult because it requires us to look very hard at what’s actually going on.

rather than just tell stories about what might happen in 20 years time.

Xpanding Horizons (42:34)

How does regulation look like and what do governments focus on and how does it evolve? Should we have more regulation than what it is right now how do you think about that?

Tom (42:46)

It’s a tough one, one, because A, I’m not an expert, so you probably shouldn’t trust what I say. And B, I think there’s a deep tension between regulating wisely in terms of accountability and rights, and if you like, regulating indiscriminately in a way that drives innovation to other places or stops people building good, effective stuff or conducting experiments. And I suppose I favor the idea

of regulation that is tied to kind of real effects in the world and to certain standards, know, to saying to people, you will be held to account for what is done with your system. You will be required to explain how and why it’s doing these things. And I quite like the idea of there being no go areas, which is one thing I think the EU has led on and saying, well, you know, we are not going to be doing kind of hyper-personalization

based on a whole load of traits that were about people’s lives or indeed their genetics and things like that. We are not going to go into kind of constant total surveillance and public facial recognition. We’re not going to be doing gain of function research on viruses willy-nilly. There’s going to be areas where we’re not going to be doing kind of indiscriminate things with AI because we believe these undermine certain rights and beliefs.

about people. think China in its own way has been extremely robust on AI around things like kind of chatbots and virtual companions and kind of fakery and so on. And it’s well worth looking around the world for different kind of practices and standards, because I do think that where AI corrodes trust, is potentially kind of creating something very fundamental in society to trust that a person is who they say they are, that a

a product will do what a product claims to do, that you can trust that certain guardrails will be in place around deeply harmful behaviors. And this is very problematic. think, because I don’t know if regulation, it’s very easy to say we must regulate, I know if regulation is the answer, or if it’s just part of an answer. Because building really robust guardrails and safeguards into systems so that they are not encouraging.

dangerous behaviors, giving dangerous advice, leading people to develop kind of dangerous dependencies to be misled and so on. This, think, cannot just be for regulators. It partly needs to be about, you know, how can AI companies be held responsible, be brought on board, but also be, you know, kind of empowered to do this well. So as you can tell, I haven’t got a perfect answer. But I think one thing one needs to do

is kind of look around the global landscape of AI with open eyes and really pay very close attention to the kind of real harms that are starting to be evidenced now rather than the hypothetical harms that might happen at some point in the future.

Xpanding Horizons (45:43)

So, which philosopher do you think, is relevant in terms of what’s happening today that is worth, revisiting and reading some of their work.

Tom (45:52)

Yeah, I mean, I could pick so many philosophers, but I’ll go for two, guess, of modern philosophers. greatly admire Luciano Floridi, the Italian philosopher, but who writes in English and is widely translated because I think he has tried to think systematically about not just AI, but the rise of powerful information technologies as a kind of paradigm shift in human history and then to have a kind of generalized approach to what we owe to each other in this context.

His book, The Fourth Revolution, casts the of the rise of information technologies as a revolution akin to the Copernican Revolution, which, you kind of was based upon the idea that the Earth was not the center of the universe. The Darwinian Revolution, where the theory of evolution really showed how, I suppose, a complex living world could emerge from simple origins without the need for a divine creator. And the Freudian Revolution, which to some degree was about the degree to which our

Self-knowledge is always imperfect and fraught that we can never know ourselves entirely. So I think Fluidi has done really interesting systematic work thinking about what it means for information environments themselves to uplift human agency, to be places that kind of, know, deepen our ability to be ethical, self-authoring, empowered beings and not be manipulated or exploited, what we owe to one another in a digital age.

And then I would probably say it’s worth going back, albeit briefly, all the way to Plato, you know, the father of Western philosophy. It’s a very predictable choice. And I’m not expecting everyone to read lots of kind of original Plato. But Plato was writing at a time when a Mediterranean culture of the spoken word, an oral culture, was gradually giving way to a written culture. And perhaps quite famously in the Phaedrus, among other places, he wrote

with ambivalence about the rise of writing, how this new information technology, the written word, potentially created forgetfulness in its readers, how it might lead to people who didn’t think for themselves, who parroted other people’s words rather than engaging in the living practice of dialogue. And similarly in the Republic, he very famously wrote about the idea that maybe most people

are not as it were kind of living in the realm of truth that we’re in a cave seeing kind of shadows cast upon a wall by a flickering light that humankind cannot bear that much reality. And thus that we’re potentially deceived and manipulated by simulacra that the truth is a very hard thing to bear. was rather worried about poetry. He thought that poetry was rather too persuasive and seductive. So there’s a lot of interesting kind of currents.

in that ancient philosophy that I think are urgent to our age today.

Xpanding Horizons (48:42)

One final question on pop culture and the way the public and thinkers think about the future. Historically, we’ve noticed many people view technology in a pessimistic manner. They do mention to be cautious, they’re somewhat worried, they have flag concerns. And if you look at the past, there is more techno pessimism than optimism. But if you look at it, the world did turn out fairly

good in terms of how technology has developed. is it just human nature that we tend to be pessimistic with technology or is there something more at play?

Tom (49:18)

I guess, course, different cultures have very different views over time. The idea of progress is perhaps only a few hundred years old. Before that, most cultures tended to see history as cyclical, as going around in great, great cycles rather than progressing. I do think that there are optimistic and pessimistic strains in culture. And that at the moment, pessimism,

It’s a strange match, isn’t it? Because we have, in terms of living standards, in terms of opportunities, in terms of health and wealth, in terms of, if you like, humanity’s scope and power and knowledge, we have more than anyone at any point in history, and we’re gonna get even more and even more. At the same time, we have a growing knowledge of our impacts on the planet, of our failures as well as our successes. And of course, the information technologies at our fingertips.

make us all hyper aware of the bad and the good. We’re kind of drowning in a sea of information. It’s very, it’s very hard to make objective sense of it. Indeed, it’s impossible. And so I guess the way I would see it is that in some ways this tension is in itself characteristic at the present moment, that we have this power, we have this opportunity, but coupled to it, we have an intense knowledge of our own failures, losses and perversities and tools which if you like, know, bad people can use to do

bigger bad things than ever before. Arguably it was with the dawn of the atomic age in the 1940s that humanity really became capable of self-destruction, kind of annihilating civilization. And with that growing power, it’s perhaps right that there becomes a growing nervousness around the use and impacts of that power. Ecologically speaking, we are at a time that some people call a poly-crisis.

I like to think, and this is perhaps my soul preservation strategy, that our capacity to innovate, change, face up to certain facts, to invent should never be underestimated. And that it’s okay to be, know, kind of a pessimist about the present and an optimist about the future even. But you’re right that I think if you’re saying that at the moment there is a kind of an emphasis on this historical strand of pessimism.

and that we’ve always been deeply worried about novelty, that that’s right. And also that we’ve also always been wrong in interesting ways. And that’s perhaps the best lesson for me, to predict the future is to be wrong. And that maybe means that ultimately if we want to find hope, we need to focus most of our attention on trying more kind of thoroughly and completely to understand the present.

Xpanding Horizons (51:54)

Thanks Tom so much for joining and chatting. was a real pleasure to have you and hear your views.

Tom (52:00)

My pleasure, Neil. Thanks for chatting to me.