Otherwise, temperatures generally in the high teens to low 20s for scotland and Northern Ireland, and about 20 to 23 widely across england and wales. It then starts to get a bit warmer for scotland and Northern Ireland as we start to get a southerly drift, so temperatures more widely will reach the 20s as we go through the weekend. Therell still be a lot of dry and sunny weather for england and wales. Just an outside chance of a Shower Working in to the coast of southern england. Our temperatures, well, coming up to about 22 in glasgow, so feeling quite a bit warmer here. And a fine weekend, sunday looks good as well. Could see 25 across parts of eastern england, but then some showers around next week. Welcome to hardtalk. Im stephen sackur. Wherever you live, however you live, Artificial Intelligence, Machine Learning, is going to change your life in the next few years. Will it be for the better . Well, there are Tech Visionaries with very different views some benign, others, dystopian. But on one thing, the futurists can agree what will matter most to the outcome isnt The Machines, its the humans. My guest, neil lawrence, is a professor of Machine Learning who says we face a choice. Either we use alas a tool or we will become a tool for al. So, which is more likely . Neil lawrence, neil lawrence, welcome to hardtalk. Welcome to hardtalk. Thank you very much thank you very much for having me. For having me. Its a great pleasure its a great pleasure to have you. To have you. Now, you are a computer what weve seen with a lot now, you are a computer scientist, but it seems to me, scientist, but it seems to me, as youve journeyed deep into the potentiality as youve journeyed deep into the potentiality of Artificial Intelligence, of Artificial Intelligence, youve also thought a great youve also thought a great deal about human Intelligence Deal about Human Intelligence and what is so very special and what is so very special and unique about us humans. And unique about us humans. Can you try to put can you try to put that into words . That into words . Yeah, i think, for me, yeah, i think, for me, what weve seen with a lot of the artificial Intelligence Debate has been a sort then i think theres a question about, well, should we use the Word Intelligence for our ecology, which is also an Information Processing Ecosystem . And we dont tend to do that unless were sort of doing animistic practices. So i think thats a really important question, goes to the heart of the debate, because i think people are confused by its use when were talking about machines. You have caused quite a stir with this book, the atomic human. Its subtitled understanding ourselves in the age of ai, and what i find so interesting about it, and were going to get deep into ai in a minute, butjust to stick with Human Beings for a second, is that you talk about, quote, the immense cognitive power of the Human Brain, and yet you also point out that the Human Brain is utterly inadequate to compete with the computer when it comes to conveying, processing information. Try and explain to me how you can both describe us as having this immense cognitive ability, and yet, in some ways, be quite useless. Yeah, its to do with the physics. So, our brain is using neurons, which use electricity to compute. And its an extraordinary, extraordinary entity. I mean, i dont want to call it a machine. Its beyond our best understanding at the moment. And it uses electricity. Now, that comes from our evolution of animals, which first used electricity to move, and a multicellular organism, which has each cell, which probably on its own is thinking, im a little cell well, its not thinking. They become coordinated through these electrical signals. Now, when i communicate with another human, i have to use sound waves. Now, the Speed Of Sound is, roughly speaking, A Million times slower than the speed of light. And that means that the rate of communication i can do with another human is roughly A Million times slower. And you break it down to bits of information. And you quantify, in any given minute of human speech, how many bits of information we can convey to another. And you compare that with how a computer talking to another computer can convey, immeasurably, more bits of information in that same minute. So, what roughly is the ratio . Well, depending on whether youre talking about ten gigabit or one gigabit internet, but basically 300 Million times faster. Right which is why, when it comes to our discussion today, which is going to be about Big Data and algorithms and how one can use immense, unimaginably large data sets to perform all sorts of tasks through algorithms, the human Being Cant compete. Well, its interesting. I wouldnt say we cant compete, but we can be easily manipulated by something that we wouldnt think of as intelligent. Theres certainly tasks that we do a lot better than The Machine, but The Machine is doing many things in a very different way. So, even when we think, oh, its doing something relatively simple, likejust looking at our data and choosing whether or not we like certain posts, classic Machine Learning algorithms, it can manipulate us because it sees us in a very different way to the way we see Each Other. Again, philosophically speaking, much of the way in which Human Beings convey meaning to Each Other isnt based on words at all. Its based on many different forms of expression. You could think of art and the way we convey meaning through paint or music. One could think of the facial expressions that, through socialisation, weve learned to invest with great meaning. Are those powers that are really important, do you think, to the way we shape the next century, Living Alongside these Machine Learning devices . Well, i want to be careful about Living Alongside, because i dont quite know what we mean by that. They should be tools that are operating under our guidance, but even that is a difficult question to answer, what does that look like . Because they are Making Decisions that are often beyond our conception. And those decisions could be things that go in ways that are not the decisions we would make if we brought our context to it. And i think the point you made about theres so much to us that is different from The Machine, in terms of, like, were here physically. I mean, how much more, how much different is it for you, even if youre not doing this on zoom to be together in the studio . Youre right. Fundamentally different. And so, how much different is it when the thing youre discussing with doesnt even have a bodily form . Im very aware that you are different from some other leaders in this sort of tech debate, because you do not take what i might call a sort of Doomsday Approach to the future. There are many eminent thinkers, like geoff hinton, who i imagine you probably know, who used to work at google. He quit. Hes a professor of computer science. He believes that al represents an existential threat to humanity. Elon musk has said that he believes that agi, as he puts it, that is artificial General Intelligence, will supersede the intelligence of all of Human Beings within, he says, a year. You dont seem to share these views and these fears. Well, that would require a Rankable Definition of intelligence, and we dont have that. Intelligence is a multifaceted thing. We cant simply rank it in that way. In fact, the idea that we can rank intelligence is fundamentally eugenic. It comes from the eugenicists, and the term General Intelligence comes from eugenics. So i think we have to be very careful with that, because its the same form of thinking that the eugenicists participated in when they said, oh, we could breed more intelligent people by measuring the intelligence of people. Intelligence is nothing like that. So, the idea of an artificial General Intelligence would be like, whats the artificial general vehicle . Whats the best vehicle for all tasks . It does everything better than all other vehicles. This is a ridiculous notion. The vehicle you would like to use is dependent on the circumstance, and in intelligence, unless we define the context, the right form of Intelligence Cant be defined. So the word is more like beauty, in that we all have a sense of what it is, but actually we cant say much more about it other than our sense. We cant define it from first principles. What matters most to you in the Here And Now of the Development And Deployment of Artificial Intelligence . I think were in a very difficult situation, where people are being told that this technology is too complex for them to understand, that they shouldnt be worrying about it, that only certain people, only certain companies, should be controlling this technology, and ifind that horrific. You mean you are seeing the rise of sort of Monopolistic Tech Oligarchs who are, in a sense, telling The Public, dont worry about it, well offer you services that you wont understand, but believe us, theyre in your best interest . And theyre gaining mindshare with governments. Of course, theyre only doing it in their. Thats their objective to, as i think Satya Nadella said, capture the ai market. But we know that the history of that, this is one of the most important changes that has occurred, certainly in 500 years since the printing press, but we can go back to 5,500 years, the development of writing. We are talking about a fundamental change in which the way we share information between Each Other, and were talking about a small group of people who are saying, you regular people, you dont understand this, let us look after it for you, and there is no Track Record of those entities doing this well. If we look at the history, these companies, they do not have a good understanding of how society works. They do not have a good understanding of the things that are most important to people, such as health, Social Care, education, security. Im struggling to square what youve just said with the fact that, for three years, very recently, you went off to work for amazon, one of the biggest Ai And Big Data corporations in the world. Why do you struggle to square that . Well, youve just described to me your deep mistrust of the agenda of. You didnt name amazon, but when you talked about the technocracy, im imagining amazon would be up there as one of the key members of the technocracy. Well, i think increasingly, because theyre increasingly moving away from their core business, which is selling directly to customers and putting customers first, where those customers are directly aligned with the purchase of a product, and theyre more getting in the business of selling to other businesses or advertising, and as these businesses. If i may interrupt, i mean, they are one of the greatest owners of data on individual consumers in the whole world today. I mean, again, imjust struggling to see how you can be so worried and fearful of these biggest global Tech Companies, and yet go off and work for one. Because you have to understand these systems if youre going to understand how to intervene, and the first level of understanding, and one of the areas i worked in predominantly was amazons Supply Chain and the possibilities for benefit through better organisation of Supply Chain, particularly when you start realising, well, Supply Chain of goods is similar to Supply Chain of availability of staff in hospitals. So, the same sets of algorithms that allow you to efficiently move goods Around The World could also allow you to more efficiently deploy nurses and doctors. So, if you want to understand the problems that come up, you have to work with these companies. Ive also worked with oil companies, you know, and understood whats going on in terms of how we extract oil. But im also worried about the climate. I think its not a healthy world where we totally disassociate ourselves from these companies. Theres very many well meaning people in these businesses. Well, do you think politicians, who, in oursystem, ultimately make the rules that the big Tech Companies have to abide by, do you think politicians have been naive in the way theyve approached this challenge of how best to control, regulate, make accountable Big Tech . I think its, part of being a politician is an extremely difficultjob, and i think a lot of it. Were so lucky in the uk that we have really good advice structures to support the politicians. But, of course, this isntjust a uk issue. It isnt, and imjust, again, looking at something you said, a sort of Siren Statement of yours society is indulgent of companies, that is the big Tech Companies that are using ai, in the way that parents are indulgent of children. We allow companies to do things that we would never let other public institutions get away with. You know, as you look Around The World, do you see that still being a profound problem . Thats a deep problem, isnt it . I mean, its extraordinary, when companies do the most basic of things, like, oh, weve opened a Research Centre in nairobi, they expect us to all cheer them as if theyve suddenly become the most saintly thing, like a child, and its because we have a somehow a moral exception around companies. We dont expect them to behave like we would expect a civil institution to behave. But we also want them to generate economic wealth. I mean, you know, if you imagine a politician in the united states or the uk, they dont want to impose a set of rules on a company at the Cutting Edge of ai, which that company might well argue would dampen down their desire to innovate, to be entrepreneurial, to push the boundaries of the technology, you know, because then they would say, we might go work somewhere else. Absolutely. But theres a sort of confusion about this space. This is one of these subtleties. So, the uks policy is actually pro innovation and one i greatly support. But pro Innovation Isnt like pro very large companies that dominate the ecosystem and prevent small players entering. The Way Market economics works is if im a doctor or a teacher and i have a good idea about how to use al to benefit patients or my colleagues, that i should be able to enter the market with that idea, and thats not the situation. Weve got a situation where digital markets are dominated by a few very large players, and government has already recognised this. So, one of the sort of extraordinary things that happened just before the election came to be, 24th of may, the digital Markets Competition and Consumer Act got passed, almost unreported in the wider press. This is one of the most important acts because its trying to address that problem, which is a problem of how Market Economics Works in the digitised era, with these information passing. As you point out, we do have laws and regulations. The eu has begun implementation of its Ai Act as well. We see that, but you go beyond that. Youve come up with this idea of Data Trusts, that peoples data, which, in a sense, is one of the most important things they own in this digital world, it should be handled in a way which they are fully aware of. There is transparency. They know what happens to their data. Theres accountability for their data. Is this world of Data Trusts ever going to happen . I hope so. Its a difficult thing to introduce because it involves educating people about how their data is being used. And it undermines the commercial principles on which so many big Tech Companies are Making Gazillions its a difficult thing to introduce because it involves educating people about