Program: Artificial intelligence – promises and threats

Robyn Williams: Last week we heard briefly from the Reith Lectures, just ended, on RN, presented by Stuart Russell of Berkeley. He talked of artificial intelligence and how it could give us colossal wealth if handled properly. Here's a brief reminder:

Stuart Russell: If all goes well, it will herald a golden age for humanity. Our civilisation is the result of our intelligence and having access to much greater intelligence could enable a much better civilisation.

One rather prosaic goal is to use general-purpose AI to do what we already know how to do more effectively, at far less cost and at far greater scale. We could, thereby, raise the living standard of everyone on Earth in a sustainable way to a respectable level. That amounts to a roughly ten-fold increase in global GDP, the cash equivalent, or the net present value, as economists call it, of the increased income stream is about £10 quadrillion or $14 quadrillion. All of the huge investments happening in AI are just a rounding error in comparison.

Robyn Williams: Stuart Russell. And with me is Toby Walsh, Professor of AI at the University of New South Wales. What did you think of the lectures?

Toby Walsh: I would encourage everyone to hear them all, they are a wonderful introduction to one of the more important questions of the century.

Robyn Williams: He was rather startling with some of the things he said. In the first one, he was talking, for example, about if we took the right decisions we would have something like ten times the global GDP that we have now.

Toby Walsh: I think he could be right. If you think back, since the Industrial Revolution we now live like kings or queens compared to people back in Victorian times, and we are going to go through another period of rapid technological change that is going to produce a bounty. The interesting question (and this is one that Stuart dwells on) is how well that wealth is then going to be shared.

Robyn Williams: But is the path to that wealth clear?

Toby Walsh: I think we are starting to see already the benefits that artificial intelligence are bringing, how they are starting to invade pretty much every aspect of our lives, and it's hard to think of an aspect of our lives ultimately that it won't touch. There is huge uncertainty though how long it's going to take, and Stuart does talk about this. If you ask most of my colleagues, they say it's going to happen in the next 40 or 50 years. We've still got a long way to match the human brain. And that's really quite profound. If we told you that alien life was about to arrive and we were no longer going to be the smartest things on the planet, there would be crisis, there would be meetings at the UN, and we would be very concerned how would we deal with this strange occurrence.

Robyn Williams: Robots…now, you've got one sitting next to me here which is very sweet. What's its name, by the way?

Toby Walsh: He is called Alpha.

Robyn Williams: Alpha, excellent.

Toby Walsh: They should be given names that remind us that they aren't human, they are robot.

Robyn Williams: I look forward to Epsilon. Now, the one problem that we have with AI, with all these machines doing all these things, is many intrude when we don't necessarily want them. I do not want my fridge to decide my shopping and to have all those sorts of things which make people more or less just lumps of flesh sitting there while there are films on television about robots taking over. How do you escape just pure product placement?

Toby Walsh: I think we do have to be careful that we don't become redundant to the machines, that our biology is wired for immediate reward and gratification, and if we're not careful we may discover that we have let them do too many things and that we aren't in charge anymore. But these are all about us making the right choices, it's all in our destiny, the machines only do what we tell them to do, they don't do anything other than what we tell them to do, so it's entirely up to us.

Robyn Williams: As you summed it up in the beginning, Stuart saying we will have this immense wealth, but the spread won't be necessarily as it should be, but the implication was that life will become less costly, we'd be able to afford more and may not need product placement to be quite so intense.

Toby Walsh: Yes, on the other side to the fact that we are generating all this extra wealth, the cost of living, the necessities of life, housing and clothing and food, should all become much cheaper because we won't have to pay humans to do those dull jobs, to till the land, to make our clothes. And so the basic cost of living will come down. And indeed, we are actually starting to see that, there are a number of companies in the UK and New Zealand that are trialling a four-day week. You have to work less to earn the amount of money to be able live a comfortable life.

Robyn Williams: In one of the programs you also talked about something which you have dwelt on quite a lot in your work and that is the autonomous weapons that may be used in war that go out and have warfare for you and go perhaps far too far.

Toby Walsh: Yes, this is something that does keep me awake at night, this is various technologies that shouldn't be used for fighting war, we decided that about biological weapons, chemical weapons, even today nuclear weapons, and blinding lasers, cluster munitions. We've said as a global community we don't need this to fight war, there are plentiful terrible ways of fighting war, we have plentiful methods of deterrence, and this is one of them, handing over the decision to machines as to who lives and who dies. And there are ongoing discussions at the United Nations. I would hope that the Australian delegates would engage more meaningfully with these discussions, and we have, sadly seen the first such weapons being used in anger. There is a report at the United Nations saying that the first autonomous drones were used to hunt down and kill people without any humans in the loop just last year in Syria.

Robyn Williams: Another terrifying final example I'll draw from the Reith Lectures was the picture of there being cheap weapons on sale, as I said, in supermarkets, they mentioned Tesco's, for about $7 or $10.

Toby Walsh: Yes, these are not going to be expensive weapons, and that does offer a significant challenge, they will be weapons that you could perhaps 3D-print. The hard part would be the intelligence, the smartness, the code that is making them smart. But there are plentiful bad people out there, North Korean hackers or Russian hackers, who might be able to procure that code at little cost. And so there is a real risk that these will proliferate, they will become, as has been described, as the Kalashnikovs of the future, readily available to all the wrong people, to terrorists and rogue states, and they will be the perfect weapons for terrorists to use, they will undertake any operation, however evil. You can tell them to go out and kill all the children, go out and kill all the women and children in a city and they will do that, however evil that thing is, because they are a cold, heartless machine.

Robyn Williams: Well, the people who are making them do that are people behaving badly, but your new book is called Machines Behaving Badly. What had you in mind?

Toby Walsh: There is one whole chapter devoted to the risk that autonomous weapons pose. And this is the real problem with technology like artificial intelligence; it offers us much promise, and equally there are many misuses the same technology can be used for. So in terms of the promise, the same algorithms that are going to be developed for those autonomous drones that are going to identify targets on the ground and track them and kill them are going to go into your autonomous cars that are going to give mobility to the elderly, to people with disability, to the people who are too young to drive, and make driving far safer.

More than a million people die every year in road traffic accidents caused by human error, caused by human incompetence at driving, our overconfidence in our ability to drive, and that will go away. One thousand people every year in Australia die in road traffic accidents, each of those road traffic accidents cost $1 million to clear up the pieces, that's a billion-dollar friction on the planet, adding on top of which of course all the human hardship that most of us know someone whose life, someone's family's life that has been impacted by one of those terrible accidents. In this country, as in almost every developed country, if you survive birth, the most common cause for you to die before the age of 30 is road traffic accident, and that will go away.

We will look back in 20, 30 years’ time when we have autonomous cars and think it looks like the Wild West. We let people drive cars without any oversight from the machines who could drive much better, that don't drive when they are drunk or when they are tired or when they are texting, all the other mistakes that we humans make.

Robyn Williams: Finally, and I don't want to go through your book completely because we can do it when it's published; do you have the same outcome in your book as Stuart Russell? Do you have good news for us about the promised future?

Toby Walsh: I have good news and bad news. I am optimistic, I do think that we face some wicked problems, the climate emergency, getting out of this pandemic, growing inequality, the breakdown of our political organisations, it really is going to be a difficult couple of decades. But the only hope we have is to embrace, as we did in the past, technologies so that our children and our grandchildren can live better lives than us. That's the good news.

The bad news is I think it's going to be a very bumpy road, that we do face some really significant challenges, and we are at a unique point in history, this is the first time in history where our children can expect to live worse quality lives than we did. That's a terrible legacy that we are leaving our children, but I am inspired by the optimism at the possibilities that technologies offer us for the future, and that if we are going to tackle these problems, human ingenuity, humanity will step in and we will find a way through by building on technologies like artificial intelligence.

Robyn Williams: Professor of AI, Toby Walsh, at the University of New South Wales, and his book, Machines Behaving Badly, comes out in a few weeks. And you can find the Reith Lectures on the BBC Radio Four website.