Hod Lipson: Robots that are "self-aware"

117,350 views ใƒป 2007-10-13

TED


ืื ื ืœื—ืฅ ืคืขืžื™ื™ื ืขืœ ื”ื›ืชื•ื‘ื™ื•ืช ื‘ืื ื’ืœื™ืช ืœืžื˜ื” ื›ื“ื™ ืœื”ืคืขื™ืœ ืืช ื”ืกืจื˜ื•ืŸ.

ืžืชืจื’ื: Ido Dekkers ืžื‘ืงืจ: Sigal Tifferet
00:25
So, where are the robots?
0
25000
2000
ืื–, ืื™ืคื” ื”ืจื•ื‘ื•ื˜ื™ื?
00:27
We've been told for 40 years already that they're coming soon.
1
27000
3000
ืกื™ืคืจื• ืœื ื• ื‘ืžืฉืš 40 ืฉื ื” ื›ื‘ืจ ืฉื”ื ื›ื‘ืจ ืžื’ื™ืขื™ื.
00:30
Very soon they'll be doing everything for us.
2
30000
3000
ื‘ืงืจื•ื‘ ื”ื ื™ืขืฉื• ื”ื›ืœ ื‘ืฉื‘ื™ืœื ื•:
00:33
They'll be cooking, cleaning, buying things, shopping, building. But they aren't here.
3
33000
5000
ื”ื ื™ื‘ืฉืœื•, ื™ื ืงื•, ื™ืงื ื• ื“ื‘ืจื™ื, ื™ืขืฉื• ืฉื•ืคื™ื ื’, ื™ื‘ื ื•. ืื‘ืœ ื”ื ืœื ืคื”.
00:38
Meanwhile, we have illegal immigrants doing all the work,
4
38000
4000
ื‘ื™ื ืชื™ื™ื, ื™ืฉ ืœื ื• ืžื”ื’ืจื™ื ื‘ืœืชื™ ื—ื•ืงื™ื™ื ืฉืขื•ืฉื™ื ืืช ื”ืขื‘ื•ื“ื”,
00:42
but we don't have any robots.
5
42000
2000
ืื‘ืœ ืื™ืŸ ืœื ื• ืจื•ื‘ื•ื˜ื™ื.
00:44
So what can we do about that? What can we say?
6
44000
4000
ืื– ืžื” ืื ื—ื ื• ื™ื›ื•ืœื™ื ืœืขืฉื•ืช ื‘ืงืฉืจ ืœื–ื”? ืžื” ืื ื—ื ื• ื™ื›ื•ืœื™ื ืœื”ื’ื™ื“?
00:48
So I want to give a little bit of a different perspective
7
48000
4000
ืื– ืื ื™ ืจื•ืฆื” ืœืชืช ืœื›ื ืคืจืกืคืงื˜ื™ื‘ื” ืฉื•ื ื”
00:52
of how we can perhaps look at these things in a little bit of a different way.
8
52000
6000
ืขืœ ืื™ืš ืื ื—ื ื• ื™ื›ื•ืœื™ื ืœื”ืกืชื›ืœ ืขืœ ื”ื“ื‘ืจื™ื ื”ืืœื” ื‘ื“ืจืš ืื—ืจืช.
00:58
And this is an x-ray picture
9
58000
2000
ื•ื–ื• ืชืžื•ื ืช ืจื ื˜ื’ืŸ
01:00
of a real beetle, and a Swiss watch, back from '88. You look at that --
10
60000
5000
ืฉืœ ื—ื™ืคื•ืฉื™ืช ืืžื™ืชื™ืช, ื•ืฉืขื•ืŸ ืฉื•ื•ืฆืจื™, ืžืฉื ืช 88. ืืชื ืžืกืชื›ืœื™ื ืขืœ ื–ื” --
01:05
what was true then is certainly true today.
11
65000
2000
ืžื” ืฉื”ื™ื” ื ื›ื•ืŸ ืื– ื‘ื˜ื•ื— ืฉื ื›ื•ืŸ ื”ื™ื•ื.
01:07
We can still make the pieces. We can make the right pieces.
12
67000
3000
ืื ื—ื ื• ืขื“ื™ื™ืŸ ื™ื›ื•ืœื™ื ืœื™ื™ืฆืจ ืืช ื”ื—ืœืงื™ื, ืื ื—ื ื• ื™ื›ื•ืœื™ื ืœื™ื™ืฆืจ ืืช ื”ื—ืœืงื™ื ื”ื ื›ื•ื ื™ื,
01:10
We can make the circuitry of the right computational power,
13
70000
3000
ืื ื—ื ื• ื™ื›ื•ืœื™ื ืœื™ื™ืฆืจ ืืช ื”ืžืขื’ืœื™ื ืฉืœ ื›ื•ื— ื”ื—ื™ืฉื•ื‘ ื”ืžืชืื™ื,
01:13
but we can't actually put them together to make something
14
73000
3000
ืื‘ืœ ืื ื—ื ื• ืœื ืžืžืฉ ื™ื›ื•ืœื™ื ืœื—ื‘ืจ ืื•ืชื ื›ื“ื™ ืœื™ืฆื•ืจ
01:16
that will actually work and be as adaptive as these systems.
15
76000
5000
ืžืฉื”ื• ืฉืžืžืฉ ื™ืขื‘ื•ื“ ื•ื™ื”ื™ื” ืžืกืชื’ืœ ื›ืžื• ื”ืžืขืจื›ื•ืช ื”ืืœื”.
01:21
So let's try to look at it from a different perspective.
16
81000
2000
ืื– ื‘ื•ืื• ื ืกืชื›ืœ ืขืœ ื–ื” ืžืคืจืกืคืงื˜ื™ื‘ื” ืฉื•ื ื”.
01:23
Let's summon the best designer, the mother of all designers.
17
83000
4000
ื‘ื•ืื• ื ืงืจื ืœืžืขืฆื‘ื™ื ื”ื˜ื•ื‘ื™ื ื‘ื™ื•ืชืจ, ื”ืืžื ืฉืœ ื›ืœ ื”ืžืขืฆื‘ื™ื:
01:27
Let's see what evolution can do for us.
18
87000
3000
ื‘ื•ืื• ื ืจืื” ืžื” ื”ืื‘ื•ืœื•ืฆื™ื” ื™ื›ื•ืœื” ืœืขืฉื•ืช ื‘ืฉื‘ื™ืœื ื•.
01:30
So we threw in -- we created a primordial soup
19
90000
4000
ืื– ื–ืจืงื ื• ืคื ื™ืžื” -- ื™ืฆืจื ื• ืžืจืง ืงื“ืžื•ื ื™
01:34
with lots of pieces of robots -- with bars, with motors, with neurons.
20
94000
4000
ืขื ื”ืจื‘ื” ื—ืœืงื™ื ืฉืœ ืจื•ื‘ื•ื˜ื™ื: ืขื ืžื•ื˜ื•ืช, ืขื ืžื ื•ืขื™ื, ืขื ื ื•ื™ืจื•ื ื™ื.
01:38
Put them all together, and put all this under kind of natural selection,
21
98000
4000
ืฉืžื ื• ืืช ื›ื•ืœื ื‘ื™ื—ื“, ื•ืฉืžื ื• ืืช ื›ืœ ื–ื” ืชื—ืช ืกื•ื’ ืฉืœ ื‘ืจื™ืจื” ื˜ื‘ืขื™ืช,
01:42
under mutation, and rewarded things for how well they can move forward.
22
102000
4000
ืชื—ืช ืžื•ื˜ืฆื™ื”, ื•ืชื’ืžืœื ื• ื“ื‘ืจื™ื ืœืคื™ ื›ืžื” ื˜ื•ื‘ ื”ื ื™ื›ื•ืœื™ื ืœื”ืชืงื“ื.
01:46
A very simple task, and it's interesting to see what kind of things came out of that.
23
106000
6000
ืžืฉื™ืžื” ืคืฉื•ื˜ื” ืžืื•ื“, ื•ื–ื” ืžืขื ื™ื™ืŸ ืœืจืื•ืช ืื™ื–ื” ื“ื‘ืจื™ื ื™ืฆืื• ืžื–ื”.
01:52
So if you look, you can see a lot of different machines
24
112000
3000
ืื– ืื ืชื‘ื™ื˜ื•, ืชื•ื›ืœื• ืœืจืื•ืช ื”ืจื‘ื” ืžื›ื•ื ื•ืช ืงื˜ื ื•ืช
01:55
come out of this. They all move around.
25
115000
2000
ื™ื•ืฆืื•ืช ืžื–ื”. ื›ื•ืœืŸ ื–ื• ืžืกื‘ื™ื‘,
01:57
They all crawl in different ways, and you can see on the right,
26
117000
4000
ื›ื•ืœืŸ ื–ื•ื—ืœื•ืช ื‘ื“ืจื›ื™ื ืฉื•ื ื•ืช, ื•ืชืจืื• ืžื™ืžื™ืŸ,
02:01
that we actually made a couple of these things,
27
121000
2000
ืฉืžืžืฉ ื™ืฆืจื ื• ื›ืžื” ื›ืืœื”,
02:03
and they work in reality. These are not very fantastic robots,
28
123000
3000
ื•ื”ื ืขื•ื‘ื“ื™ื ื‘ืžืฆื™ืื•ืช. ืืœื” ืœื ืจื•ื‘ื•ื˜ื™ื ืคื ื˜ืกื˜ื™ื™ื,
02:06
but they evolved to do exactly what we reward them for:
29
126000
4000
ืื‘ืœ ื”ื ืžืชืคืชื—ื™ื ืœืขืฉื•ืช ื‘ื“ื™ื•ืง ืžื” ืฉืžืชื’ืžืœื™ื ืื•ืชื ืœืขืฉื•ืช:
02:10
for moving forward. So that was all done in simulation,
30
130000
3000
ืœืชื–ื•ื–ื” ืงื“ื™ืžื”. ืื– ื›ืœ ื–ื” ื ืขืฉื” ื‘ืกื™ืžื•ืœืฆื™ื”,
02:13
but we can also do that on a real machine.
31
133000
2000
ืื‘ืœ ืื ื—ื ื• ื™ื›ื•ืœื™ื ืœืขืฉื•ืช ืืช ื–ื” ื’ื ื‘ืžื›ื•ื ื•ืช ืืžื™ืชื™ื•ืช.
02:15
Here's a physical robot that we actually
32
135000
5000
ื”ื ื” ืจื•ื‘ื•ื˜ ืคื™ืกื™ ืฉื™ืฉ ืœื ื•
02:20
have a population of brains,
33
140000
3000
ืžืžืฉ ืื•ื›ืœื•ืกื™ื” ืฉืœ ืžื•ื—ื•ืช,
02:23
competing, or evolving on the machine.
34
143000
2000
ืžืชื—ืจื™ื, ืื• ืžืชืคืชื—ื™ื, ืขืœ ื”ืžื›ื•ื ื”.
02:25
It's like a rodeo show. They all get a ride on the machine,
35
145000
3000
ื–ื” ื›ืžื• ืชื—ืจื•ืช ืจื•ื“ื™ืื•: ื›ื•ืœื ื–ื•ื›ื™ื ืœืจื›ื‘ ืขืœ ื”ืžื›ื•ื ื”,
02:28
and they get rewarded for how fast or how far
36
148000
3000
ื•ื”ื ืžืชื•ื’ืžืœื™ื ืขืœ ื›ืžื” ืžื”ืจ ืื• ื›ืžื” ืจื—ื•ืง
02:31
they can make the machine move forward.
37
151000
2000
ื”ื ื™ื›ื•ืœื™ื ืœื’ืจื•ื ืœืžื›ื•ื ื” ืœื–ื•ื– ืงื“ื™ืžื”.
02:33
And you can see these robots are not ready
38
153000
2000
ื•ืืชื ื™ื›ื•ืœื™ื ืœืจืื•ืช ืฉื”ืจื•ื‘ื•ื˜ื™ื ื”ืืœื” ืœื ืžื•ื›ื ื™ื
02:35
to take over the world yet, but
39
155000
3000
ืœื”ืฉืชืœื˜ ืขืœ ื”ืขื•ืœื ืขื“ื™ื™ืŸ, ืื‘ืœ
02:38
they gradually learn how to move forward,
40
158000
2000
ื”ื ืœื•ืžื“ื™ื ื‘ื”ื“ืจื’ื” ืื™ืš ืœื–ื•ื– ืงื“ื™ืžื”,
02:40
and they do this autonomously.
41
160000
3000
ื•ื”ื ืขื•ืฉื™ื ืืช ื–ื” ืœื‘ื“.
02:43
So in these two examples, we had basically
42
163000
4000
ืื– ื‘ืฉืชื™ ื”ื“ื•ื’ืžืื•ืช ื”ืืœื”, ื”ื™ื• ืœื ื•
02:47
machines that learned how to walk in simulation,
43
167000
3000
ืžื›ื•ื ื•ืช ืฉืœืžื“ื• ืื™ืš ืœืœื›ืช ื‘ืกื™ืžื•ืœืฆื™ื”,
02:50
and also machines that learned how to walk in reality.
44
170000
2000
ื•ื’ื ืžื›ื•ื ื•ืช ืฉืœืžื“ื• ืื™ืš ืœืœื›ืช ื‘ืžืฆื™ืื•ืช.
02:52
But I want to show you a different approach,
45
172000
2000
ืื‘ืœ ืื ื™ ืจื•ืฆื” ืœื”ืจืื•ืช ืœื›ื ื’ื™ืฉื” ืฉื•ื ื”,
02:54
and this is this robot over here, which has four legs.
46
174000
6000
ื•ื–ื” ืจื•ื‘ื•ื˜, ื›ืืŸ, ืฉื™ืฉ ืœื• ืืจื‘ืข ืจื’ืœื™ื™ื,
03:00
It has eight motors, four on the knees and four on the hip.
47
180000
2000
ื™ืฉ ืœื• ืฉืžื•ื ื” ืžื ื•ืขื™ื, ืืจื‘ืขื” ื‘ื‘ืจื›ื™ื™ื ื•ืืจื‘ืขื” ื‘ื™ืจื›ื™ื™ื.
03:02
It has also two tilt sensors that tell the machine
48
182000
3000
ื™ืฉ ืœื• ื’ื ืฉื ื™ ื—ื™ื™ืฉื ื™ ื”ื˜ื™ื™ื” ืฉืื•ืžืจื™ื ืœืžื›ื•ื ื”
03:05
which way it's tilting.
49
185000
3000
ืœืื™ื–ื” ื›ื™ื•ื•ืŸ ื”ื™ื ื ื•ื˜ื”.
03:08
But this machine doesn't know what it looks like.
50
188000
2000
ืื‘ืœ ื”ืžื›ื•ื ื” ื”ื–ื• ืœื ื™ื•ื“ืขืช ืื™ืš ื”ื™ื ื ืจืื™ืช.
03:10
You look at it and you see it has four legs,
51
190000
2000
ืืชื ืžืกืชื›ืœื™ื ืขืœื™ื” ื•ืจื•ืื™ื ืืจื‘ืข ืจื’ืœื™ื™ื,
03:12
the machine doesn't know if it's a snake, if it's a tree,
52
192000
2000
ื”ืžื›ื•ื ื” ืœื ื™ื•ื“ืขืช ืื ื”ื™ื ื ื—ืฉ, ืื ื”ื™ื ืขืฅ,
03:14
it doesn't have any idea what it looks like,
53
194000
3000
ืื™ืŸ ืœื” ืฉื•ื ืžื•ืฉื’ ืื™ืš ื”ื™ื ื ืจืื™ืช,
03:17
but it's going to try to find that out.
54
197000
2000
ืื‘ืœ ื”ื™ื ืขื•ืžื“ืช ืœื ืกื•ืช ืœื”ื‘ื™ืŸ.
03:19
Initially, it does some random motion,
55
199000
2000
ื‘ื”ืชื—ืœื”, ื”ื™ื ืขื•ืฉื” ืชื ื•ืขื•ืช ืืงืจืื™ื•ืช,
03:21
and then it tries to figure out what it might look like.
56
201000
3000
ื•ืื– ื”ื™ื ืžื ืกื” ืœื”ื‘ื™ืŸ ืื™ืš ื”ื™ื ื ืจืื™ืช --
03:24
And you're seeing a lot of things passing through its minds,
57
204000
2000
ื•ืืชื ืจื•ืื™ื ื”ืจื‘ื” ื“ื‘ืจื™ื ืขื•ื‘ืจื™ื ืœื” ื‘ืจืืฉ,
03:26
a lot of self-models that try to explain the relationship
58
206000
4000
ื”ืจื‘ื” ืžื•ื“ืœื™ื ืขืฆืžื™ื™ื ืฉืžื ืกื™ื ืœื”ืกื‘ื™ืจ ืืช ื”ื™ื—ืก
03:30
between actuation and sensing. It then tries to do
59
210000
3000
ื‘ื™ืŸ ืชื–ื•ื–ื” ืœื—ื™ืฉื” -- ื•ืื– ืžื ืกื” ืœืขืฉื•ืช
03:33
a second action that creates the most disagreement
60
213000
4000
ืชื ื•ืขื” ืฉื ื™ื” ืฉื™ื•ืฆืจืช ืืช ื—ื•ืกืจ ื”ื”ืกื›ืžื” ื”ืžื™ืจื‘ื™
03:37
among predictions of these alternative models,
61
217000
2000
ื‘ื™ืŸ ื”ืชื—ื–ื™ื•ืช ืฉืœ ื”ืžื•ื“ืœื™ื ื”ืืœื˜ืจื ื˜ื™ื‘ื™ื™ื,
03:39
like a scientist in a lab. Then it does that
62
219000
2000
ื›ืžื• ืžื“ืขืŸ ื‘ืžืขื‘ื“ื”. ืื– ื”ื•ื ืขื•ืฉื” ืืช ื–ื”
03:41
and tries to explain that, and prune out its self-models.
63
221000
4000
ื•ืžื ืกื” ืœื”ืกื‘ื™ืจ ืืช ื–ื”, ื•ืžื•ื—ืง ืืช ื”ืžื•ื“ืœื™ื ื”ืขืฆืžื™ื™ื.
03:45
This is the last cycle, and you can see it's pretty much
64
225000
3000
ื–ื” ื”ืžื—ื–ื•ืจ ื”ืื—ืจื•ืŸ, ื•ืืชื ื™ื›ื•ืœื™ื ืœืจืื•ืช ืฉื”ื•ื ื“ื™
03:48
figured out what its self looks like. And once it has a self-model,
65
228000
4000
ื”ืฆืœื™ื— ืœื”ื‘ื™ืŸ ืืช ื”ืžืจืื” ื”ืขืฆืžื™ ืฉืœื•, ื•ื‘ืจื’ืข ืฉื™ืฉ ืœื• ืžื•ื“ืœ ืขืฆืžื™,
03:52
it can use that to derive a pattern of locomotion.
66
232000
4000
ื”ื•ื ื™ื›ื•ืœ ืœื”ืฉืชืžืฉ ื‘ื• ื›ื“ื™ ืœื™ืฆืจ ื“ืคื•ืกื™ ืชื ื•ืขื”.
03:56
So what you're seeing here are a couple of machines --
67
236000
2000
ืื– ืžื” ืฉืืชื ืจื•ืื™ื ืคื” ื”ืŸ ื›ืžื” ืžื›ื•ื ื•ืช --
03:58
a pattern of locomotion.
68
238000
2000
ื“ืคื•ืก ืฉืœ ืชื ื•ืขื”.
04:00
We were hoping that it wass going to have a kind of evil, spidery walk,
69
240000
4000
ืงื™ื•ื•ื™ื ื• ืฉืชื”ื™ื” ืœื• ืžื™ืŸ ืชื ื•ืขื” ืžืจื•ืฉืขืช, ืขื›ื‘ื™ืฉื™ืช,
04:04
but instead it created this pretty lame way of moving forward.
70
244000
4000
ืื‘ืœ ื‘ืžืงื•ื, ื”ื•ื ื™ืฆืจ ื“ืจืš ื“ื™ ืขืœื•ื‘ื” ืœืชื–ื•ื–ื” ืงื“ื™ืžื”.
04:08
But when you look at that, you have to remember
71
248000
3000
ืื‘ืœ ื›ืฉืžืกืชื›ืœื™ื ืขืœ ื–ื”, ืฆืจื™ืš ืœื–ื›ื•ืจ
04:11
that this machine did not do any physical trials on how to move forward,
72
251000
6000
ืœื ืขืฉืชื” ืฉื•ื ื ื™ืกื•ื™ื™ื ืคื™ืกื™ื™ื ืขืœ ืื™ืš ืœื–ื•ื– ืงื“ื™ืžื”,
04:17
nor did it have a model of itself.
73
257000
2000
ื•ืœื ื”ื™ื” ืœื” ืžื•ื“ืœ ืฉืœ ืขืฆืžื”.
04:19
It kind of figured out what it looks like, and how to move forward,
74
259000
3000
ื”ื™ื ื‘ืขืจืš ื”ื‘ื™ื ื” ื‘ืขืฆืžื” ืื™ืš ื”ื™ื ื ืจืื™ืช, ื•ืื™ืš ืœื–ื•ื– ืงื“ื™ืžื”,
04:22
and then actually tried that out.
75
262000
4000
ื•ืื– ืžืžืฉ ื ื™ืกืชื” ืืช ื–ื”.
04:26
(Applause)
76
266000
5000
(ืžื—ื™ืื•ืช ื›ืคื™ื™ื)
04:31
So, we'll move forward to a different idea.
77
271000
4000
ืื–, ื ืžืฉื™ืš ืœืจืขื™ื•ืŸ ืื—ืจ.
04:35
So that was what happened when we had a couple of --
78
275000
5000
ืื– ื–ื” ืžื” ืฉืงืจื” ื›ืฉื”ื™ื• ืœื” ื›ืžื” --
04:40
that's what happened when you had a couple of -- OK, OK, OK --
79
280000
4000
ื–ื” ืžื” ืฉืงืจื” ื›ืฉื”ื™ื• ืœื” ื›ืžื” -- ืื•ืงื™, ืื•ืงื™, ืื•ืงื™ --
04:44
(Laughter)
80
284000
2000
(ืฆื—ื•ืง)
04:46
-- they don't like each other. So
81
286000
2000
-- ื”ื ืœื ืื•ื”ื‘ื™ื ืื—ื“ ืืช ื”ืฉื ื™. ืื–
04:48
there's a different robot.
82
288000
3000
ื™ืฉ ืจื•ื‘ื•ื˜ ืื—ืจ.
04:51
That's what happened when the robots actually
83
291000
2000
ื–ื” ืžื” ืฉืงืจื” ื›ืฉื”ืจื•ื‘ื•ื˜ื™ื ืžืžืฉ
04:53
are rewarded for doing something.
84
293000
2000
ืžืชื•ื’ืžืœื™ื ืขืœ ืขืฉื™ื™ืช ืžืฉื”ื•.
04:55
What happens if you don't reward them for anything, you just throw them in?
85
295000
3000
ืžื” ืฉืงื•ืจื” ืื ืœื ืชื’ืžืœืชื ืื•ืชื ืขืœ ืฉื•ื ื“ื‘ืจ, ืคืฉื•ื˜ ื–ื•ืจืงื™ื ืื•ืชื ืคื ื™ืžื”?
04:58
So we have these cubes, like the diagram showed here.
86
298000
3000
ืื– ื™ืฉ ืœื”ื ืืช ื”ืงื•ื‘ื™ื•ืช ื”ืืœื”, ื›ืžื• ืฉื”ืชืจืฉื™ื ืžืจืื” ืคื”.
05:01
The cube can swivel, or flip on its side,
87
301000
2000
ื”ืงื•ื‘ื™ื•ืช ื™ื›ื•ืœื•ืช ืœื”ืชืคืชืœ, ืื• ืœื”ืชื”ืคืš ืขืœ ื”ืฆื“,
05:04
and we just throw 1,000 of these cubes into a soup --
88
304000
4000
ื•ืื ื—ื ื• ืคืฉื•ื˜ ื–ื•ืจืงื™ื 1000 ื›ืืœื” ืœืชื•ืš ืžืจืง --
05:08
this is in simulation --and don't reward them for anything,
89
308000
2000
ื–ื• ื”ื“ืžื™ื” -- ื•ืœื ืžืชื’ืžืœื™ื ืื•ืชื ืขืœ ื›ืœื•ื,
05:10
we just let them flip. We pump energy into this
90
310000
3000
ืคืฉื•ื˜ ื ืชื ื• ืœื”ื ืœื”ืชื”ืคืš. ื”ื–ืจืžื ื• ืื ืจื’ื™ื” ืœืชื•ืš ื–ื”
05:13
and see what happens in a couple of mutations.
91
313000
3000
ื•ืจืื™ื ื• ืžื” ืงื•ืจื” ืื—ืจื™ ื›ืžื” ืžื•ื˜ืฆื™ื•ืช.
05:16
So, initially nothing happens, they're just flipping around there.
92
316000
3000
ืื– ื‘ื”ืชื—ืœื”, ื›ืœื•ื ืœื ืงื•ืจื”, ื”ื ืคืฉื•ื˜ ืžืชื”ืคื›ื™ื ืฉื.
05:19
But after a very short while, you can see these blue things
93
319000
4000
ืื‘ืœ ืื—ืจื™ ื–ืžืŸ ืงืฆืจ, ืืชื ื™ื›ื•ืœื™ื ืœืจืื•ืช ืืช ื”ื“ื‘ืจื™ื ื”ื›ื—ื•ืœื™ื ื”ืืœื”
05:23
on the right there begin to take over.
94
323000
2000
ื‘ืฆื“ ื™ืžื™ืŸ ืฉื ืžืชื—ื™ืœื™ื ืœื”ืฉืชืœื˜.
05:25
They begin to self-replicate. So in absence of any reward,
95
325000
4000
ื”ื ืžืชื—ื™ืœื™ื ืœื”ืฉืชื›ืคืœ. ืื– ื‘ื”ืขื“ืจ ืชื’ืžื•ืœ,
05:29
the intrinsic reward is self-replication.
96
329000
3000
ื”ืชื’ืžื•ืœ ื”ืขืฆืžื™ ื”ื•ื ืฉื™ื›ืคื•ืœ ืขืฆืžื™.
05:32
And we've actually built a couple of these,
97
332000
1000
ื•ื‘ืขืฆื ื‘ื ื™ื ื• ื›ืžื” ื›ืืœื”,
05:33
and this is part of a larger robot made out of these cubes.
98
333000
4000
ื•ื–ื” ื—ืœืง ืžืจื•ื‘ื•ื˜ ื’ื“ื•ืœ ื™ื•ืชืจ ืฉื‘ื ื•ื™ ืžื”ืงื•ื‘ื™ื•ืช ื”ืืœื”,
05:37
It's an accelerated view, where you can see the robot actually
99
337000
3000
ื–ื” ืžื‘ื˜ ืžื•ืืฅ, ื›ืฉืืชื ื™ื›ื•ืœื™ื ืœืจืื•ืช ืืช ื”ืจื•ื‘ื•ื˜ ืžืžืฉ
05:40
carrying out some of its replication process.
100
340000
2000
ืžื‘ืฆืข ื—ืœืง ืžืชื”ืœื™ืš ื”ืฉื›ืคื•ืœ.
05:42
So you're feeding it with more material -- cubes in this case --
101
342000
4000
ืื– ืžื–ื™ื ื™ื ืœื• ืขื•ื“ ื—ื•ืžืจื™ื -- ืงื•ื‘ื™ื•ืช ื‘ืžืงืจื” ื”ื–ื” --
05:46
and more energy, and it can make another robot.
102
346000
3000
ื•ืขื•ื“ ืื ืจื’ื™ื”, ื•ื”ื•ื ื™ื›ื•ืœ ืœื™ืฆื•ืจ ืจื•ื‘ื•ื˜ ื ื•ืกืฃ.
05:49
So of course, this is a very crude machine,
103
349000
3000
ืื– ื›ืžื•ื‘ืŸ, ื–ื• ืฉื™ื˜ื” ืžืื•ื“ ื’ืกื”,
05:52
but we're working on a micro-scale version of these,
104
352000
2000
ืื‘ืœ ืื ื—ื ื• ืขื•ื‘ื“ื™ื ืขืœ ืฉื™ื˜ื” ืžื•ืงื˜ื ืช ืฉืœ ื–ื”,
05:54
and hopefully the cubes will be like a powder that you pour in.
105
354000
3000
ื•ื‘ืชืงื•ื” ื”ืงื•ื‘ื™ื•ืช ื™ื”ื™ื• ื›ืžื• ืื‘ืงื” ืฉืืชื ืฉื•ืคื›ื™ื ืคื ื™ืžื”.
05:57
OK, so what can we learn? These robots are of course
106
357000
5000
ืื•ืงื™ื™, ืื– ืžื” ืื ื—ื ื• ื™ื›ื•ืœื™ื ืœืœืžื•ื“? ื”ืจื•ื‘ื•ื˜ื™ื ื”ืืœื” ื›ืžื•ื‘ืŸ
06:02
not very useful in themselves, but they might teach us something
107
362000
3000
ืœื ืžืžืฉ ื™ืขื™ืœื™ื ื‘ืขืฆืžื, ืื‘ืœ ื”ื ืื•ืœื™ ื™ืœืžื“ื• ืื•ืชื ื• ืžืฉื”ื•
06:05
about how we can build better robots,
108
365000
3000
ืขืœ ืื™ืš ืœื‘ื ื•ืช ืจื•ื‘ื•ื˜ื™ื ื˜ื•ื‘ื™ื ื™ื•ืชืจ,
06:08
and perhaps how humans, animals, create self-models and learn.
109
368000
5000
ื•ืื•ืœื™ ืื™ืš ืื ืฉื™ื, ื—ื™ื•ืช, ื™ื•ืฆืจื™ื ืžื•ื“ืœื™ื ืขืฆืžื™ื™ื ื•ืœื•ืžื“ื™ื.
06:13
And one of the things that I think is important
110
373000
2000
ื•ืื—ื“ ื”ื“ื‘ืจื™ื ืฉืื ื™ ื—ื•ืฉื‘ ืฉื”ื ื—ืฉื•ื‘ื™ื
06:15
is that we have to get away from this idea
111
375000
2000
ื–ื” ืฉืื ื—ื ื• ืฆืจื™ื›ื™ื ืœื•ื•ืชืจ ืขืœ ื”ืจืขื™ื•ืŸ
06:17
of designing the machines manually,
112
377000
2000
ืฉืœ ืชื›ื ื•ืŸ ื”ืžื›ื•ื ื•ืช ื‘ืื•ืคืŸ ื™ื“ื ื™,
06:19
but actually let them evolve and learn, like children,
113
379000
3000
ืืœื ื‘ืขืฆื ืœืชืช ืœื”ื ืœื”ืชืคืชื— ื•ืœืœืžื•ื“, ื›ืžื• ื™ืœื“ื™ื,
06:22
and perhaps that's the way we'll get there. Thank you.
114
382000
2000
ื•ืื•ืœื™ ื›ื›ื” ื ื’ื™ืข ืœืฉื. ืชื•ื“ื” ืจื‘ื”.
06:24
(Applause)
115
384000
2000
(ืžื—ื™ืื•ืช ื›ืคื™ื™ื)
ืขืœ ืืชืจ ื–ื”

ืืชืจ ื–ื” ื™ืฆื™ื’ ื‘ืคื ื™ื›ื ืกืจื˜ื•ื ื™ YouTube ื”ืžื•ืขื™ืœื™ื ืœืœื™ืžื•ื“ ืื ื’ืœื™ืช. ืชื•ื›ืœื• ืœืจืื•ืช ืฉื™ืขื•ืจื™ ืื ื’ืœื™ืช ื”ืžื•ืขื‘ืจื™ื ืขืœ ื™ื“ื™ ืžื•ืจื™ื ืžื”ืฉื•ืจื” ื”ืจืืฉื•ื ื” ืžืจื—ื‘ื™ ื”ืขื•ืœื. ืœื—ืฅ ืคืขืžื™ื™ื ืขืœ ื”ื›ืชื•ื‘ื™ื•ืช ื‘ืื ื’ืœื™ืช ื”ืžื•ืฆื’ื•ืช ื‘ื›ืœ ื“ืฃ ื•ื™ื“ืื• ื›ื“ื™ ืœื”ืคืขื™ืœ ืืช ื”ืกืจื˜ื•ืŸ ืžืฉื. ื”ื›ืชื•ื‘ื™ื•ืช ื’ื•ืœืœื•ืช ื‘ืกื ื›ืจื•ืŸ ืขื ื”ืคืขืœืช ื”ื•ื•ื™ื“ืื•. ืื ื™ืฉ ืœืš ื”ืขืจื•ืช ืื• ื‘ืงืฉื•ืช, ืื ื ืฆื•ืจ ืื™ืชื ื• ืงืฉืจ ื‘ืืžืฆืขื•ืช ื˜ื•ืคืก ื™ืฆื™ืจืช ืงืฉืจ ื–ื”.

https://forms.gle/WvT1wiN1qDtmnspy7