Instant Genius - Why AI is not the enemy - Jim Al-Khalili

Episode Date: August 30, 2018

Jim Al-Khalili explains how artificial intelligence has changed the world, who benefits from it, and why we probably shouldn’t be afraid of it destroying humanity. Hosted on Acast. See acast.com/privacy for more information. Learn more about your ad choices. Visit podcastchoices.com/adchoices

You said this place was steps from the water. We just haven't found the steps yet. How much did we save? Enough. Enough to get lost. Or you could book a stay with Hilton. Welcome to your oceanfront room. Just steps from the water.

The Hilton sale is on now. Book on Hilton.com or the Hilton app and save up to 20% to get the stay you expected. When you want savings, not surprises. It matters where you stay. Hilton, for the stay. Study And play.

Come together on a Windows 11 PC. And for a limited time, college students get the best of both worlds. Get the Unreal College deal, everything you need to study and play with select Windows 11 PCs. Eligible students get a year of Microsoft 365 premium

and a year of Xbox GamePass Ultimate with a custom color Xbox wireless controller. Learn more at Windows.com slash student offer. While supplies last, ends June 30th, terms at AKA.m.S. College PC. Ambition comes in all shapes and sizes. At First Citizens Bank, we roll with your goals because we're built for what you're building.

Fit for your ambition for Citizens Bank. This podcast is sponsored by name, audio and focal. Streaming has made music more accessible than ever, but true listening is about more than ease. It's about quality. British audio experts name audio, alongside French acoustic specialist focal, Combine handcrafted tradition with cutting-edge innovation and high-end materials, delivering digital precision with analogue warmth. So you can experience exceptional sound at home.

Music just as the artist intended. Visit name audio.com to learn more. To some extent, we should be nervous. But not nervous about AI is taking over the world, not nervous because of the sky net. I think we should be more nervous about the fact that AI is still very dumb and the fact that potentially dumb but powerful people will control it. You're listening to the Science Focus podcast from the BBC Focus magazine team.

We're the UK's best-selling science and technology monthly, available in print and in several digital formats throughout the world. Find out more at ScienceFocus.com or look out for us in your app store. Hello and welcome to the Science Focus podcast. I'm Alice Lipscomb, Southwell, the production editor of BBC Focus magazine. Today, there is barely any device that doesn't have a computer brain powering it under the hood. But how smart really is this artificial intelligence? We seem happy to let AI improve our smartphone photos or turn the lights on when it gets dark.

But the thought of autonomously driving our cars leaves some of us uneasy. And of course, there is the dystopian vision of the future with robot terminators enslaving humanity. But it doesn't have to be this way. Online editor Alexander McNamara speaks to theoretical physicist Jim Al-Qa. Klili, ahead of his new BBC 4 documentary, The Joy of AI, in which explores the history of artificial intelligence, what it can do today and why embracing the positives of AI into our lives can lead to a better future for us all. So you've got a new programme that's out soon called The Joy of AI. Yes. I was just wondering if you'd be able to sort of just kick things off by

explaining to us what exactly is AI? Well, AI stands for artificial intelligence and it really does seem to be the buzz phrase around at the moment because the technology involving artificial intelligence very generally is rapidly, rapidly evolving. The thing is that when we talk about AI, we tend to mean something very general indeed. And I think some of the people working in the field might argue, well, that's not really artificial intelligence. I mean, an example, we all use Siri on smartphones or what's the other one in the homes? Alexa or Alexa, yeah. I don't have it so it's not on.

So Siri and Alexa, you know, which sort of help, you know, they answer questions. They sound as though they're intelligent. They sound as though they're sort of figuring stuff out for you. The same with things like Google Translate or even just the way, you know, Amazon or Facebook are able to sort of directly advertise stuff to you. How do they know that? But of course it's because of what you searched for before. So all of these things in technology are sort of classed broadly in artificial intelligence.

What I'm interested in, and certainly what this program is about, is where we're going in the future, because artificial intelligence is basically a machine behaving intelligently, doing intelligent things, not just number crunching. And that, we've not got there yet, but we're moving very fast towards it. So it's very exciting. But also, of course, what comes with it are these sort of words of warning that we need to make sure we know what's coming over the horizon. So what's the difference between a regular computer programme and something

that actually can be classed as an intelligent. artificial application? Well, there are various ways of defining it, because of course, a computer program, if it's complex enough, can do things much more quickly than a human brain can, but also can give the appearance for being intelligent. You know, a computer program that can beat a grandmaster at chess is just running an algorithm, you know, lines of code saying, if this, then do that, if that, then do that. But, you know, a game of chess, what it can do is explore all possible moves into the future

and decide what is the optimum piece to move, giving it the appearance of intelligent, whereas what it's really doing is number crunching. Very often what we mean by true intelligence nowadays is if the machine uses something called machine learning, whereby it's not simply given a set of instructions by a programmer and it just follows them blindly, but it learns as it goes. So it sort of builds its own code, as it were. It becomes increasingly sophisticated. And that's what modern AI machines are doing now. They're learning as they go. They use something called neural networks, which are meant to mimic in a sense the way our brain works, the way our brain is wired and all the neurons and how they're all interconnected.

A neural network is an abstract mathematical version of the human brain. And they are starting to show real intelligence by anyone's definition. So that's what we're doing. We're sort of creating a brain, a computer brain, as it were. How similar, or what's the size comparison between this machine learning? brain and a human brain? Well, I mean, in terms of physical size, I suppose, it's not really so relevant. You know, you could imagine some small thing, but wires connected down the corridor in a

computer lab to a vast array of huge machines. So in terms of physical size, I suppose, it's not so important. But in terms of the capabilities, of course, machines, artificial intelligences are way, way behind what a human brain can do. I mean, they give the appearance of being intelligent. You know, an AI can look at a picture and understand from the pixels what it represents. It can recognize pictures of a dog, for example. But it doesn't know what a dog is.

It doesn't understand the concept of dog. And so while it can do certain things faster and more efficiently and even sometimes in a more clever way than humans, that's still not intelligent in the way that we would talk about it. And I think the problem is we're very often too infected by Hollywood movies. And so, you know, we think of AIs or being intelligent as somehow being conscious, being sentient, being self-aware. That may come in the future, but we're a long, long way, many decades away from that. So is that sort of sentient, the difference as it is between artificial and intelligence we know and human intelligence? It's the difference between artificial intelligence

and intelligence in lots of living things, mainly mammals. You know, I don't think there's nothing special about human intelligence as such. You know, a dog can feel happy and angry and sad and jealous, which we regard as human traits. No artificial intelligence is anywhere near being able to have emotions in that sense, you know, because these are sort of higher level thinking that really only come about in sentient beings. And that's not just constrained to humans. So is calling it artificial intelligence really the right sort of wording? Is that we've got that quite right? Probably not in hindsight. I mean, this grew out. I mean, we're talking the program about how it first came about, you know, programming machines to think for themselves in a very

general way. Probably if we had to sort of redefine it, we'd call it something else. You know, machine learning or deep learning or artificial neural networks may be more accurate. They're just not quite as catchy, I guess, as artificial intelligence AI. I mean, the term is here to stay, so we just have to be careful what we mean by it. So coming back to that point of how it all began, how did we arrive to the point that we are today with artificial intelligence or machine learning or wherever we are? Well, it grew in parallel with computer science in general,

but the development of algorithms that you can code into an electronic device using binary, you know, set of zeros and ones, is something, you know, we go back long before people like Alan Turing, people are looking like Babbage, for example, you know, calculating machines that automate a process of numerical calculations. Artificial intelligence grew in sort of the middle of the 20th century, very slowly, but it allowed computers to try and solve problems. that maybe we were unable to solve, not just because they could do things more quickly,

but they could solve logic problems in a way that wasn't simply down to lines of code algorithm. So in that sense, it's different from general computer coding, computer programming. But the lines are blurred, and it's really only in the last decade or so, that we started to see AI machines doing something very different that is not the same as just making more and more powerful computers that do calculations faster and faster. I mean, a lovely example is the neural net codes at DeepMind, the British London-based company now owned by Google. but they are essentially probably the world leaders in AI and they're pioneered this idea of deep learning. So they have these AI machines, Alpha Go, Alpha Zero,

which do lots of different things as well as beat the best computer chess players in the world. Computers can already beat the best grandmasters at chess. These AI machines, just from a few hours of looking at a chess board and playing against themselves, they've learnt and have become sort of superhuman in their abilities. So that's different from just writing long lines of code and building more, more powerful computers. There's a lovely bit in the programme.

You mentioned AlphaGo there, where the sort of move that the computer did was so different to anything that a human has done. And is that the machine learning things in a different way to how humans work? Yes, I mean, that is what is so exciting and some might argue also very scary because what the AI did there was make a move. This is the Chinese game of Go, which people regard us as a better indicator of innovative ways of thinking than say chess. Chess is very, you know, if you do this, that happens, you do this, that happens. So you can explore all possibilities. With Go is much more about intuition.

and what this AI did was make a move that all the sort of grandmasters watching it play against one of their own thought was a silly move. Why did it do that? You know, that doesn't make sense. That's not what we would do. It just doesn't make sense. And it was only much later on in the game that they appreciated what a genius move it was because it became important as the game developed. So it had discovered a solution to a problem. problem, the people who built the machine didn't understand and weren't expecting.

So that's what's different. These AIs are starting to do stuff that we don't quite understand how they've come to. That sort of real innovation, real intuition that's going on inside these deep learning machines. That sounds both incredible, but that also, to me, seems where the scary bit does come in, like, how much they're going to learn? And what? Yeah. I mean, I think that this is what, you know, a lot of people are concerned about, that, you know, if they're doing things that we can't understand, if they become so true black

boxes, then how can we maintain control over them? What if they start getting smarter and smarter ever more quickly and reach what we refer to as the singularity? They develop general, you know, artificial general intelligence. They really become self-aware. and then decide that they don't need humans. This is the Skynet scenario from the Terminator. And people who work in AI and oh, God, let's see how long it takes

before someone brings up the Terminator, because that's always the worry. People who work in AI are much more optimistic, that we can maintain control and that we can build in sufficient safeguards. Even though AIs may end up being more intelligent than us, they would still be in our control. Now, that may sound, you know, rather somewhat difficult to achieve.

And I understand people's issues with it. Actually, this was the reason why when we were making, first the talks about making this program, and of course the joy of AI follows in a series of documentaries made by the same production company, Wingspan, for the BBC, starting with the joy of stats, which was presented by the late Hans Rosling, a statistician, and that documentary won a BAFTA.

It was fantastic. There have been others following on from that. So the joy of data was presented by another mathematician, Hannah Fry. So this is the next of the series, the joy of AI. And I did say it warned the production. I said, well, a lot of people won't see the words joy and AI fitting very comfortably together. So we need to be rather mindful. of that, that people are nervous about it. And I think, you know, to some extent, we should be nervous.

But not nervous about AI is taking over the world, not nervous because of the sky net. I think we should be more nervous about the fact that AI is still very dumb and the fact that potentially dumb but powerful people will control it for reasons that are not so good. You know, technology itself is not good or evil. Technology is just the application of scientific ideas. It's how we humans use technology. So AI can be used for good or for evil. A very simple example is you could use AI together with robotics for sending in autonomous bomb disposal devices, robots that will go in and understand their surroundings and diffuse a bomb so that humans don't have to risk their lives. Similarly, you could have AIs in autonomous and killer drones, you know,

sort of various countries are actually thinking about using. So these drones that will fly over enemy lines, not only just to scout out and send images back, but maybe take an autonomous decision as to whether they should take a human life. And those sorts of things are rather more worrisome. You know, so we need to know what's coming. I think that sometimes people worry about, oh, you know, driverless cars, oh, that's a terrible thing. Well, no, that'll come. Driverless cars will be here and we'll be absolutely fine with them.

And we'll solve all our problems that we have with, you know, driverless cars, computers driving cars. And it'll become the most natural thing for us. And we will wonder how we were ever worried about such a thing. But there are other aspects of AI that we do need to be more careful about. I sort of think of it in a way like the when the industrial revolution happened and there were the the Luddites were smashing the machines

because they saw them as a threat I wonder if we're now at the beginning of a sort of artificial intelligence revolution whereas 100 years down the line will go actually it wasn't so bad life is much better now but through the back of it Oh I'm convinced that will be the case of course at the time when

technologies start to replace human labor, there's always a pushback. And understandably so, people will lose their jobs, you know, but that has happened, as you say, since the Industrial Revolution. You know, ever since steam engines came in, because they could do the job more efficiently, more cheaply, and because they were stronger than human muscle, of course they replaced humans. you know, robotic arms on factory production lines replaced humans because they could do things more efficiently and more quickly and so on and more cheaply. So this is just another in the long line of technologies replacing humans in doing certain jobs, but on the other hand, making our lives

easier. You know, try persuading a 1950s housewife who's just been given a washing machine that in fact, you know, technology is evil, that she should go back to doing her or washing by hand. I think the difference with the AI revolution is that we are still not sure how much of an impact it will have. It'll have a huge impact. I always say that if you look back,

we've only had the Internet and the World Wide Web for a quarter of a century. AI will be equivalent to that in terms of it changing our lives. And imagine now life without the internet. The AI will change our lives equally, if not more, than the internet has. And it'll do it far more quickly than the next 25 years. What it'll do will change in the next 10, 15 years.

So it's coming so fast and it's going to change so many aspects of our lives. We just need to be ready for it. We shouldn't be afraid of it. We just need to be prepared. I guess that's one of those things. People are, when they hear about AI, they think always going to take our jobs. You know, I'm lucky.

I'm in the position where I get to speak to people for, you know, as part of my job. But there are some that will obviously be more susceptible to being taken over. Will AI create more jobs around that? So, yeah, will it create more jobs around it as opposed to just taking them away as people are afraid of?

I predict it certainly will. Of course, what it'll do will create new jobs that haven't been invented yet that we can't even imagine right now. And that's the difficulty of working. So I said, was the quote, prediction is very unreliable, particularly about the future. Some Danish proverb, I think. Yes, new technologies have always meant that certain jobs have. been transferred from humans to machines, but they've always brought with them new jobs that weren't around before. But we don't know what they are, and we don't know whether the number of new jobs created will, you know, it's not necessarily a zero-sum game. The number of new jobs created will exactly match the number of jobs that AIs have taken

over from humans. And so many people are understandably, and I think quite rightly, nervous about what AI will do because, you know, companies and, you know, an industry are always looking, they're looking for, you know, profits. That's their bottom line and efficiency. And if an AI can do something more efficient, then we're already seeing that, you know, you even when you, you know, if you very often, you'll phone up a restaurant to, you know, to book a table and you'll talk to someone who's actually isn't a someone, it's a machine, you know, because it can do things efficiently and it can, I can recognize your frequently asked questions and it will have the stock answers.

And so jobs will go and a lot of them. What we don't know is, you know, is it 10%, is it 60%? Will it affect certain countries more than others, depending on what their economy is like, depending on what sorts of industries they have? These are unknowns. But the fact is we can't stop the march of progress. AI is coming. And so what we should do is rather than hide away from it

or think it's evil and not want to engage with it, we need to be prepared for it. Because it'll come. And if society decides they're scared of AI and governments decide they don't want to talk about it or be prepared for it, I mean, others will. You know, the big corporations, the Googles and Amazon's and Facebooks and so on,

they will use AI and let alone, you know, whether rogue states make use of it to interfere with democracies or whatever, cyber terrorists, you know, it's going to come. We just need to make sure we know what it can and can't do. While, of course, you know, trying to minimize its impact on society on people's livelihoods. So it's sort of embracing AI at this point rather than just having the fear of what's to come.

Yeah, I mean, I think people are right to be nervous, but having a fear of what's to come sometimes for me means, oh, I don't want to know, I don't want anything to do with it. That's not going to help. Technology is coming and technology. It'll be wonderful if we're prepared for it. Just think about the advances AI could do in medicine and healthcare, you know, and, you know, recognizing tumors that no human could possibly see, helping us with, you know, helping surgeons, you know, with augmented reality.

AI just, you know, just the standard, you know, driverless cars would cut down on 80% of road accidents because 80% of road accidents are down to human error, you know, losing concentration, tiredness or whatever. that wouldn't happen with AIs. I always think there's a lovely, so in the program, I talked to a robotics expert, Paul Newman, they're developing driverless cars, and he told me, I'm not quite sure actually when it made it into the program, but he told me this wonderful notion that if he were to take his car out tomorrow and make a silly, you know, do a silly mistake that causes an accident,

chances are in the future he won't make that mistake, again. He will have learnt from his mistake. But others may make that mistake, of course. Whereas if AI controls cars, one AI making a mistake, because they're interconnected, all other cars on the planet will never make that mistake themselves again. Because it's learning all the time. So it just gets better and better. And we can, one day we will sit back in our, in our cars without even, the cars won't even have a steering wheel. Just jump in and tell you. it where we want to go. And we'll wonder what all the fuss was about.

And we'll look back over those years and just go, what are we doing, driving at all? Well, absolutely. Yes, back in the day when we actually had to manually steer a car. Wow, how crazy. No wonder there were so many road accidents, so many people killed on the roads. That sort of brings me through to my next question, actually,

which was, I was just thinking, you know, the show is itself very positive. And actually, you come out with a very positive view of AI at the end of it. I was just wondering, what do you think will be the next big thing that we see in AI that's like, you know, very revolutionary that will really make

everyone go, wow, this is great. I think things like, you know, certainly things like driverless cars are still probably a decade or more away. So what's happening now is the advances in, you know, things like chatbots,

the likes of, you know, the likes of, you know, Siri on your phone. It's incremental. So it's not suddenly there's a revolution that's AI's do something that hasn't occurred to anyone. I think over the coming five years or so, we're just going to see increased sophistication. You know, it wasn't that long ago when we were all utterly blown away by Google Translate, that you could type in a word into Google and it'll tell you the translation of that. And then you can cut and paste a whole paragraph and translate it.

Now you just walk around with your smartphone, you just talk to it. You just talk to Syria, you know, translate. You say something in your language and it speaks the other language to whoever you want to talk to. What's what, you know, the next step will be some sort of universal translator that, you know, as much loved by science fiction, that's not so far away.

That's not so crazy. So, yeah, AI will sort of integrate throughout so many different areas of life. So that's something called the Internet of Things, which I don't cover actually in the program, where, you know, all home appliances, we're already seeing home appliances all being interconnected so you can speak into your phone to turn your central heating on or to turn the oven on, your fridge might tell you're running low on milk. That's probably not quite there yet. But the notion that everything in the home or in the workplace is interconnected

and being controlled by some AI that ensures everything runs efficiently, whether it's environmentally efficient or whether it's simply to make your life more comfortable. That's going to be with us in the coming years. and he'll be with us and we won't sort of notice a point when yesterday we didn't have it and today we do because these things happen gradually and incrementally. It's just a very steep incremental rise

because things are changing so quickly. I'm just thinking with the idea that it's all that everything is speaking to each other which just sort of made me think who is ultimately deciding what the AI decides as good and this is the optimum position for this This is the growth thing.

Is it, you know, is it democratically done as an average and everything? Or are the biases of the developers and the people who create it? Are those being fed into it somehow? That, I think, for me, is the crucial thing that we need to make sure we get right. That is what we should be most worried about with the coming of AI and not your Skarnet scenario of the AIs themselves being in control. It's those in control of the AIs. The worry is that artificial intelligence may lead to further inequality, you know, between the haves and have-nots.

You know, if you're controlled by, you know, your life is run by this AI, which itself is run and controlled by some organizational corporation, that's what we need to sort of guard against. That's why there needs to be more public debate. you have to be in control of your own AI at home. You have to be able to optimize it. The decisions it makes should supplement and help your decision-making process. It should make your lives easier, but not at the cost of being controlled by someone else who wants something insidious from you. So yes, you know, is it the programmer who, who,

who controls the AI, who controls what the AI can or can't do. That's what we need to discuss very carefully. Is that sort of a role for governments, or is that, as you mentioned earlier, there could be governments that take it all, you know, Googles and Apples and Facebooks and Microsoft, they will be the ones who ultimately have the power and take it? Yeah, certainly governments have a role to play to make sure that these powerful organizations and corporations

are not in control of the way AI goes, but also there needs to be wider societal understanding of AI. So people, you know, whether this program helps, there's one tiny drop in the ocean, but there needs to be more public dialogue. And those who do understand and those, you know, the governments, certainly in the UK, for example, which is one of the countries that leads the world in AI, there are government committees, there are, groups who are who are writing reports on the concerns about AI that are published,

the worry is that they just get read by a small fraction of society, whereas there needs to be a broader push to educate people to be. So people have just much more aware of AI. And then, of course, in a democracy, it's people who then, vote for governments that they think are going to look after their concerns, as is always the case. And I don't think at the moment there's enough understanding of what AI is about in wider society. Most people either don't know what's coming or are too scared to engage or simply don't understand the capabilities that are just around the corner and that has to change.

stepping aside and not necessarily excluding AI you're obviously most well known as a physicist what do you predict to be the big breakthroughs in physics in 2019 or the next couple of years oh big breakthroughs in physics um i i think it's probably not going to come from so the sexy areas of astronomy or particle physics you know the large had drunk Collider is currently going to shut down and we're not going to be announcing new particles discovered. I don't know whether there's going to be big discoveries in astronomy. I think the big advances are coming where physics overlaps with other areas.

We're interdisciplinary science. So, you know, AI is a good example. You know, it's not just computer scientists working in AI. They're working with neuroscientists and physicists. and mathematicians and so on. Physicists are working with biologists in genetic engineering and genomics, in nanotechnology and new sort of smart materials. So those are areas that we're going to see some interesting advances.

There's a lot of scope for progress. So I don't think it's going to be in this more esoteric. theoretical boundaries of our knowledge. It's more in learning about the world and what we can do with new devices, new instrumentation. Things like artificial intelligence, I guess, is an example. So I don't see any big revolutions around the corner,

but exciting smaller advances that are going to be, probably of more immediate applications to everyday lives. I sort of, I hear that a lot. A lot of the people I speak to, they say it's not so much the individual disciplines. It's the crossing over of the disciplines and everyone working together. And that's what's really changing things at the moment. Yeah, and we're seeing that more and more. It used to be frowned upon.

It used to be sort of, oh, well, you know, if you work in interdisciplinary science, then you're just a jack-of-all-trades, master of none. but certainly for example on you know the scientists i interview on on life scientific on radio four invariably these are people who either work in interdisciplinary sciences or more often than not have started in one field and applied their expertise in a completely different field and so you know have seen things in a different from a different angle and that has led to innovation and and advances So while we do need people to be experts borrowing down deeply into their own specialist area, the really exciting stuff these days, yes, as I say, are likely to come from the crossovers where we know.

Unless you talk to people from other disciplines in science, you will never have come across. It's only that cross-fertilisation of ideas that leads to genuinely new advances. That was Professor Jim Al Khalili on artificial intelligence. The Joy of AI is on BBC 4 next Tuesday, the 4th of September at 9pm. Thanks for listening to the Science Focus podcast. The September issue of BBC Focus magazine is out now. And in it, we look at the future of food from how farming is preparing for population growth and climate change and also see what we'll be eating in 2028.

We also delve deep into the minds of psychopaths, uncover crimes using DNA ancestry kits, and ask why people believe in conspiracy theories that fly in the face of scientific evidence. As always, there is much, much more inside. Thank you for listening to the Science Focus podcast from the BBC Focus magazine team. We're the UK's best-selling science and technology monthly, available in print and in several digital formats throughout the world. Find out more at ScienceFocus.com will look out for us in your app store. This podcast is sponsored by name, audio and focal.

The texture and emotional depth of music can be lost through digital sources or poor signal. Name Audio believes you can have digital precision with analog warmth. Alongside French acoustic specialist vocal, Name creates high-end audio systems combining innovation with craftsmanship, so you can listen to music, just as the artist intended. Discover more at nameadio.com. Some follow the noise. Bloomberg follows the money, whether it's the funds fueling AI or crypto's trillion dollar swings.

There's a money side to every story. Get the money side of the story. Subscribe now at Bloomberg.com. There's a moment when you start to wonder, what's the right next step? Not about changing who they are, just finding the right kind of support. At Kingsley Manor, life stays expressive, connected, and full of character, shaped by people who have lived interesting lives and aren't finished yet. So it doesn't feel like a change. It feels like a continuation. Explore your options at kingsley manor.org, a non-profit month-to-month senior community within the Front Porch family.