Gadgetoff 2007 - Marvin MinskyShow Video Details ↓ [noise] … …Marvin Minsky: What I'd really love to do is tell you for about 2 hours about the real gadget I want to make which is a way to backup the human brain so that if something bad happens to you, nobody cares because you can reload all the information that was in your brain into a new robot. And, of course, if you have a new robot you can live forever because the parts and so on. But that's another subject. Maybe Aubrey de Grey will talk about it. I just wanted to talk to you about practical gadgets. And in my pocket I usually have this. Anyway, if you have a bunch of gadgets in your pocket, [laughter] that's great. Then you want to hang them from your belt because otherwise they'll wear a hole in your pocket. This is a typical collection. Of course you want to have your laser and there's a fairly recent laser which contains a flashlight so that you use the same batteries. This is a typical set. You want a gravity pen so that you can write upside down under water. You should have a stick of hot-melt glue, an erasable pen, a 10 power magnifier and 2 gigabytes of RAM and a tape measure and so forth. And everyone needs all of this. I fix about 4 things a day. Of course I have a digital camera and have you seen this? This is a 9-volt battery and a little thing. Here's the digital watch you saw except this one is smaller and it can play an hour of video and record 8 hours of audio and costs 50 dollars and so forth. And we saw a cockroach with legs and you've seen all sorts of vehicles recently and I think there's a big future in vehicles with legs because vehicles with wheels require an incredible amount of infrastructure. Several percent of the national budget goes into maintaining roads and stuff. And so here is a device that I invented a number of years ago and really never told anyone about. How do you make a leg? [silence] So this gadget has 2 parts and the part on the left does exactly what the part on the right does. It has 4 joints, it's a little flimsy right now but if I move this joint then only the corresponding other joint moves and you're supposed to imagine that there's 10 miles here. Now what goes between this set of 4 joints and this set of 4 joints is 6 wires. And it has the wonderful feature that if you break any 2 wires... You see it needs 4 degrees of freedom, 1, 2, 3, 4. If you break any 2 of the 6 wires it will still work the same. And in fact 1 of them is broken and I never got around to repairing it. How does that work? Well most robots have motors to control joints. This robot has a wire that goes around each joint, either this way or that way, either left or right. But it's not attached to the joint and so, to make the arm do what you want, you imagine not what the joints do, but a Fourier series. So the first component is straight, it just goes like that. The second component is U-shaped and the third component is S-shaped. And the fourth component is W-shaped. So you want a 4 by 4 matrix. But what I did, being an ex-mathematician, is I made a 6 by 4 matrix which has the property that any subset of 4 columns is still non-singular. The result is a machine where you can break any part and any 2 out of 6 parts and it doesn't matter. And I've never seen any mechanical device that actually shows any thought about reliability. [laughter] Well, I've used up my 4 minutes but my theory is that... There are lots of theories about how the brain works and you can see some guy saying I have a neural net theory of how to make a machine that will learn anything and this one has a statistical theory about how to learn anything and this one says I'm going to make a simulated evolution and this one says I'm going to make a rule-based system. And there are about 10 movements in AI that, since about 1980, have gotten some good results but stopped making progress. And the reason is everybody is trying to find the best way to do something. Well what you want is something like this where you have 6 pretty good ways of doing something and if some of them don't work maybe the others will. And if you read my new book called The Emotion Machine, there's about 300 pages of ways to exploit that idea. And to me we're just big gadgets and we're made out of lots of little gadgets and the important thing is to find out how to put them all together, not holistically but reductionistically so that if anything breaks, something else will takeover. [applause] [applause] [silence] |