Better Ways of Talking About AI, AI Tutors and AI Sentience

Nine-minute radio clip and transcript on several topics in AI including brief discussion of some new concepts from my new book

Above is a nine-minute interview on several aspects of AI including How we can talk about AI better (from my new book Surfing AI: 30 Fresh Terms and Smarter Ways of Talking About Artificial Intelligence). AI tutors and AI in schools. Is AI intelligent and sentient?

If you are interested in only one topic just go to the relevant heading below (the transcript has been lightly edited for clarity).

Dr Paul Duignan’s interview on artificial intelligence with Mai FM

Mai FM: We’ve got a special guest. Yes, we have a clinical and strategy psychologist. It’s our mate, Dr. Paul. Good morning.

Mai FM: Hi, Dr. Paul.

Kia ora [Hello], everybody.

Mai FM: You’re writing a couple of books and doing some presentations about AI in schools. And when you say AI in schools, I bet you there’s still a lot of people in this world that freak out when you say that.

Dr Paul: Yes, well, it’s natural enough. It’s fair enough to be a bit suspicious about AI. But this recent announcement from Judith Collins the Minister of Science, Innovation and Technology about AI tutors is kind of exciting, I think it’s actually a good move.

Is AI intelligent or sentient?

Mai FM: Dr. Paul, I’ve got to ask, is AI intelligent or sentient? Like, will it take over or…?

Dr Paul Duignan: That’s the big question. I’ll come back in about 20 years to tell you what happens. But I think AI has enormous opportunities, but it has a lot of threats. People sort of go on one side or the other, whereas I kind of go down the middle. We need to use its opportunities, but we need to be really serious about managing its threats. Just on the question about is it sentient or intelligent, we’ve just got to remember these words were created to talk about humans. So I don’t think it’s the best way of really talking about it. It’s more like ‘What can it do and what can’t it do?’ And can this create risks? So I think it’s a better way of looking at it. So it’s not going to be like a person because it’s not a person. It’s going to be different. So it’s a bit like saying, do humans have silicon? There’s silicon in those chips. But it’s mad to say, do humans have silicon? We don’t. We’re quite different things.

Mai FM; There’s some people I met in Las Vegas who had silicon.

Dr Paul Duignan: There’s a bit of that around.

Are AI tutors a good idea?

Mai FM: Yeah, but I get what you mean. What’s so exciting about having AI tutors and AI in schools?

Dr Paul Duignan: Well, the great thing is that the research shows that having a tutor actually does help kids at school. And the problem, of course, is that no one can afford that. Well, just about hardly anyone can afford to have a tutor. So it’s a big educational inequality problem there. So what Judith Collins is saying is, ‘well, hey, maybe we could get some AI in there so everybody could have a tutor.’ Now, if we do it safely, pilot it, it’s actually a really cool idea because AI can kind of monitor exactly where the kid’s at and then give them exactly what they need.

The other day I said to AI, how would you teach your five-year-old kid about nuclear physics? And it started saying, oh, talk about the waves at the beach or all this cool stuff. So [AI] tutors can mean that basically every kid in the country potentially could have their own tutor.

If we do it, I think we need to look at what people are doing internationally, not reinvent the wheel. But I think that we need to do it, but we need to make sure it’s piloted and safe.

Mai FM: When we start by introducing the idea of a tutor, do teachers get their backs ups? Because you’re like, well, if you’re giving them tutors, then all of a sudden there’s the next stage, you’re replacing teachers in general and kids are just learning off AI and off the internet and off bots.

Dr Paul Duignan: Well, the first thing is that we tend to think of a tutor just like one kid just talking to an actual device of some sort, sitting on their own, right?

Mai FM: Yeah.

Dr Paul Duignan: But in actual fact, AI is really cool at coordinating. So there’s no reason why AI couldn’t be saying, ‘Hey, there’s one kid on this [side of the room], some kid’s working away at algebra.’ And then say, ‘Hey, go and talk to that kid on the other side of the room. because they can help you with that.’ And we know that teaching others helps the person who’s teaching learn too. Or it can say to the teacher, ‘This person really needs your help.’ Teachers have got a real problem knowing where each kid is. So the first point, I think, is it can help coordinate, not just the one-to-one thing.

But that then relates to the fact, you know what we learned during the pandemic? No one wants their kids sitting at home, just sitting on a computer all day, because we know where that’ll end up. So in actual fact, I don’t think teachers need to be worried about AI at all at the moment. We’re still going to send our kids to school, but this will free teachers up to not be so preoccupied with academics, but maybe more about the social side of things, to help them through resources, help them through administration.

Of course, it needs to be piloted and done safely, of course, to check that it’s going to work and not cause any problems.

Doesn’t AI just let students cheat?

Mai FM: Yeah. Dr. Paul, I’ve got to ask, as a student, like years ago, probably like 20 plus years ago. Long time ago. Long, long time ago. I was the type of student that would find loopholes to get exams done, you know, essays done the easiest way. So what about students out there that like to look for loopholes? Can they use AI to their advantage, I guess? Yeah, chat GPT to write their exams. This is what people worry about at the moment.

Dr Paul Duignan: Now, the thing is, sometimes the solution to an AI problem is actually more AI, because in a sense, at the moment, someone can just ask ChatGP to write an assignment, right?

Mai FM: Yeah.

Dr Paul Duignan: But once you get the AI tutor, it’s going to be keeping an eye on you, right? It’s going to be keeping a better eye on you than the teacher can because you can be sitting down the back there, looking at the computer, the teacher doesn’t know what you’re doing. But if the computer’s got AI in it, it’ll be watching every move. So it can make sure you’re not cheating, but then it can just gently get you to do what you need to do and make it interesting, you know so I think I think in this case more AI might help rather than less AI.

Mai FM: Do you know what this reminds me of it reminds me of that little paper clip that used to be on Microsoft, Clippy.

Dr Paul Duignan: Yeah he was a bit of a disaster because he wasn’t quite up to it but its just like that. Just someone who just guides you if you have questions.

Mai FM: So what I’m taking from this is it sounds like the bloody NRL a couple of years ago where you’ve got one ref, now you want another ref.

Dr Paul Duignan: Yeah, but it’s like the super ref, you know, who comes in and sorts everything out.

A new way of talking about AI - Paul’s new book on AI - Surfing AI

Mai FM: So you’ve written a book called Surfing AI. Did you write this or did AI write this?

Dr Paul Duignan: No, I wrote it. In actual fact, on the cover I’ve got human written.

Mai FM: Yeah, I love that. Love it.

Dr Paul Duignan: Apparently there’s at least 5,000 books out about AI. I think about 90% would be written by AI. And that’s just a little point, that for things created by humans in this new world actually there may be an advantage in saying, ‘Hey, this is created by a human. It’s not just something pumped out by someone using AI.’ So there’s kind of an advantage in that.

Mai FM: So is this a book that helps encourage people to embrace the new technology as we move forward?

Dr Paul Duignan: Well, as I said, I’m sort of going down the middle. I think great opportunities, great risks. So what I’m trying to do in this book is very specific. It creates a new language for talking about AI.

Say we’re going on a road trip. We wouldn’t talk so much about gearboxes, steering columns, and wheel alignment, we talk about GPS apps, refueling and accommodation, right? Now at the moment, we don’t have this language to talk about AI. So I’m really trying to invent some new language at the right level to talk about AI. And that’s terms like infotrash. We’re about to be swamped by a whole lot of infotrash. That’s sort of creating what I’m calling the trustability crisis. If anything can be made up by AI, how do you know what’s true and what’s not true?

Will we all run away into sort of firewalled communities where we just kind of pull up the drawbridge and not accept communications from anyone? Other concepts I introduce are like forecastability AI. This is where AI is able to figure out exactly what you’re going to do. Totally predict what you’re going to do. So is this going to freak everybody out? Are we going to worry about it?

Another term I introduce is infonakedness. Are we going to worry if AI can find out anything about us, it’ll know what we’re going to do? A futher concept is AI compelled, will people in court say, ‘Look, I couldn’t help it, the AI made me do it.’

So I’m trying to introduce these new words that we can use to have a more strategic discussion about AI rather than just sticking at this level of people talking about generative AI and computer systems and machine learning. That stuff’s great for technologists, but not so great for the rest of us.

Mai FM: Yeah. I hear what you’re saying, Dr. Paul, but I’ve got to ask you a question. Is this purely passion around AI, or are you an ambassador influencer, or are you getting in the bag behind the scenes?

Dr Paul Duignan: No, I’m not getting any kickbacks from the industry. As a psychologist, what excites me, like, up until now, there’s been humans, humans are intelligent, We’ve got emotions and stuff. As a psychologist, we study them, right?

All of a sudden, it’s like there’s this brand-new intelligent entity in the world. So you see how it can kind of blow a psychologist’s mind because you can actually give AI tests. You know, there’s these tests that we give to humans. You take exactly the same test and test AI to see how it’s doing. You know, kind of like IQ tests and all sorts of others.

So I’m really, just as a psychologist, really, really interested in it. I’ve also been, over the years, involved in work on technology and risk and all that kind of stuff. So that’s where I’m coming from.

Mai FM: Yeah. Love it. If you text ‘AI’ to 463, we can send you back more details in regards to Surfing AI, the book that Dr. Paul is publishing. And you can learn a little bit more. But it’s always good to have you on the show, Dr. Paul. Thank you so much.

Dr Paul Duignan: Thanks so much, guys

.

Mai FM: All right. See you later. Take care.