Rodney Brooks, the roboticist featured in Errol Morris’ great documentary Fast, Cheap and Out of Control, is interviewed by Joanne Pransky of Robotics Business Review about the future of AI. A few exchanges follow:
Joanne Pransky:
Let’s assume that your life is only 50 per cent complete. What groundbreaking challenges do you think you’ll be working on 25 and 50 years from now?
Rodney Brooks:
Twenty-five years from now: getting into and out of bed. Fifty years from now: going to the bathroom. I think robotics for eldercare and homecare are going to be important because of demographic inversion, and that’s going to be the big market for robots going forward. In one of my talks, I put up a picture of a Mercedes-Benz 2014 S-Class, and I asked the audience, “What is this?” And they say, “Oh, it’s a car. Oh, it’s a Mercedes”. And somebody said it’s an S-Class. I said, “It’s an eldercare robot”. Because what it’s going to do is let me drive much longer and safely, before my kids pry my keys from my “cold, dead hands”, so to speak. This is an example of a technology which is going to allow the elderly to have dignity and independence longer, and we baby boomers are going to be demanding those as we get older, as there aren’t going to be enough young people to serve our elderly needs.
Joanne Pransky:
If you could wave a magic wand, what technological item would you give to the world?
Rodney Brooks:
There’s two: a technological hand like a human, and object recognition like a child. We have image-based object recognition, but we don’t have the category recognition that a child can do.
Joanne Pransky:
How far away do you think we are from that vision recognition?
Rodney Brooks:
When I did my PhD on that topic in 1977, I thought we were a long way away and it’s still a long way away. We can now do vision a lot better using different techniques, but not in the same “general” way that people can do it. That may take a long time. We’ve had airplanes for over a hundred years. It’s only in the last few years that people have gotten model airplanes to land on branches. We are just understanding STOL (short takeoff and landing) now, which birds use all the time, for flying machines. That took a hundred years.
Joanne Pransky:
And what do you think the future human–robot interface (HRI) will be like? Will it be directly in the brain, as other science fiction people state? Will it be with our eyes?
Rodney Brooks:
I saw my first touch screen probably around 1988/1989 at CMU and I thought, “That’ll never work.” When I go to some of the academic human robot interaction conferences, I like to characterize some of the papers as, “Well, we tested this variation on that variation, and 60 per cent of people preferred Method A, and the other two preferred Method B.” I think that’s “want-to-be” scientist stuff. It’s asking questions at the wrong level. I think we haven’t invented it. I think a university should be inventing wild HR interactions and seeing what sticks, instead of, “Oh, well, should it be displayed this way or should I have this?” They haven’t invented this interface yet, whatever it’s going to be. That’s what people should be doing, trying different things, most of which will fail. But everyone wants the paper that just gets accepted, just enough science. I don’t know what it’s going to be, but things will change.•