Margaret Boden: artificial intelligence is around us everywhere.
Feature - Upfront
Margaret Boden
by Matt Nippert
Sussex University professor Margaret Boden, OBE, recently found herself in the middle
of an academic “pissing competition” with linguist and deified intellectual Noam
Chomsky. Boden is a leader in the field of cognitive science, which spans philosophy,
psychology and computer science, and published a 1500-page history of the subject.
Visiting Auckland University as a Hood Fellow, Boden discussed artificial intelligence, “computer companions”, and her spat with Chomsky.
Last December, Noam Chomsky published a blistering 10-page critique of your book, Mind as Machine, in the journal Artificial Intelligence. What did you do to deserve that? Actually, I got the most incredibly funny email yesterday. It said: “As a urologist, I can recognise a pissing competition when I see one.” The point is, it wasn’t a pissing competition when I started it. To cut a long story short, when I started looking into it, I discovered this appalling intellectual situation in linguistics: there are Chomskians and anti-Chomskians, and the vitriol – disagreement is one thing, you’re used to that in academia – but the vitriol is quite extraordinary. This is linguists talking! And I said, “Why?”
Pissing competition or not, the controversy can’t have been bad for sales. You’re right! Many more people will have heard about the book now than if he hadn’t weighed in.
When it comes to “Artificial Intelligence” [AI], most people think of The Terminator and machines running amok. How divorced from reality is this view? [Laughs] That’s pretty divorced from reality for at least two reasons. First, it’s too damn difficult to devise such systems. And, second, we’d have to choose to do it. If we were close to doing it, there would be all sorts of ethical and political questions raised about whether we should go on. Worries about robots running amok really are nonsense.
How, then, does AI work in practice? AI is around us everywhere. If you spend something on your credit card – if you go into a shop and buy a diamond necklace or a Harley-Davidson – you might well be asked about it. In the process of putting the card through, AI runs checks on your past patterns of spending, and if it thinks this is something you don’t normally spend that amount of money on, it will put in various checks and ask the person in the shop to put you on the phone.
AI is also making a splash in financial circles. Some hedge funds controlled entirely by computer are performing quite well. I don’t have any expert knowledge of that area, but I believe it’s the case that they’ve had to be slowed down. If they connect with one another, and they make their decisions and recognise the patterns so incredibly quickly, all sorts of transactions can take place within a tiny amount of time that we actually don’t want taking place. I think there are some problems, really, in that the technology’s too good.
Feedback loops could crash the market before anyone realises what’s going on. That’s right.
You’re interested in the area of “computer companions”. Are these oversized Tamagotchi pets? No. What I mean by a computer companion is a computer system – a robot or maybe just a computer with a screen – in your home, which is intended not just as a tool but as something that gives you something like a human conversation.
But what bothers me about this area is the notion that these can be companions, and can in some sense be substitutes for relationships with other human beings. We ascribe agency to things that aren’t really agents. A dog or a cat is an agent, but of a very simple sort compared to us. And I think there are real problems here – ethical problems, moral problems, as well as the technological problems.
If we can make computers that play chess better than people, surely the technology can’t be too far off? That’s an interesting case, because you say they play chess better than people. Well, admittedly, Deep Blue beat Gary Kasparov.
Surely that’s a good measure. Well, in one sense it is, but in another it isn’t. Deep Blue won those games because it was based on specially designed hardware that was making 200 million calculations a second. Of course we can’t do anything like that. If you like, it was cheating because it wasn’t playing chess in the way in which Kasparov and every other human being play chess.
