Interview from the BBC World Service radio show Digital Planet, broadcast 22 May 2006

This is just a snippet of the program (all I was on) that I transcribed myself. There was also a nice interview of Kathleen Richardson who organised the Cambridge Robot Project.  The interviewer is Gareth Mitchell.
Mitchell:  One of those taking part [in the Cambridge Robot Project's production and celebration of Rossum's Universal Robots] is Joanna Bryson of the computer science department from the University of Bath in the West of England. And of course with story lines about robots taking over the world in the back of my head, I managed to grab a few words with Joanna earlier.

Bryson:  People worry about robots taking over the world, or that they would have ambitions, because, as we make our AI, in general, we make our AI smarter than we are. Of course, you probably have on your phone a calculator (in your pocket) that can do arithmetic a million times faster than you can. And yet, it hasn't taken over even your pocket, let alone the world, right?

Mitchell:  I remember how depressed I was when a very basic computer back in the 1980s easily beat me at chess. That was the wakeup moment for me.

Bryson: Well, that's true. And I'm not claiming that we do have a strong conception of ourselves. I think it was probably really depressing when people found out that the sun didn't go around the earth, too. And (of course, you can hear my accent, I'm an American) people still have issues with the notion of evolution.

Mitchell:  I think there's something both frustrating but also delightfully reassuring about the real world robots that we have these days. Because, generally, when you see they're actually not that good [JB laughs]. These vacuum cleaning robots, well they're brilliant if you have a totally flat floor [JB laughs again] and no stairs in your flat. and I look at that thing and I think, "These things aren't taking over the world at all!" It can barely vacuum up a room let alone take over the earth.

Bryson: Right, it is ludicrous to worry about the iRobot hoovers taking over the world... [[cut]] I think that one of the things that worries me about AI and ethics is that we do get on these completely bizarre trajectories that make us not think about what's really happening. So one of the things that is really happening is that we do have unmanned vehicles in the (you know) battlezone. And if you start saying "Oh, it's a real agent.  We've created life; it has ethics of its own," you can displace responsibility from the people who are the actual decision makers onto the object. And that's terrible. To me, that's a much bigger problem --- that people are trying to misrepresent these things as different than any other object we create. Just because they have some kind of cleverness.

And yes, they can beat us at chess, and maybe they can do speech recognition. And maybe we'll get to things that can really reason and that we can have decent conversations with. [But] AI is not really exceptional. It's the same kind of thing that happens with soldiers. It's easy to push blame down the ranks, and that's the kind of thing that I worry about.

I worry that we don't worry enough about, for example, our sewage --- our water systems, where is the water coming from for London into the future? And yet we obsess about these things, as you say, these hoovers that can't even get up a flight of stairs.


Gareth went on in the wrap up to make the point I was trying to make about a "strong conception" of self better than I did.  I think we worry inappropriately about AI because we aren't sure what it is that makes us special.  (See more by going back to my AI & Society page.)

His friend Bill then said that I don't worry enough about AI; that if we AI engineers make a mistake just once then it will be the AI not thinking about us rather than us not thinking about AI.  Er... no, that's the point, that would take a lot more than one "mistake."  Bill should maybe worry about nanotechnology or biotech --- those are things that can reproduce themselves a lot more quickly and easily than a robot, and could kill or consume resources a lot more coldly and efficiently than any cognitive system --- anything that has to reason, like a robot.


page author: Joanna Bryson
Last modified: Thu Nov 16 15:08:02 GMT 2006