Can robots be creative? - Gil Weinberg

495,388 views ・ 2015-03-19

TED-Ed


Please double-click on the English subtitles below to play the video.

00:10
How does this music make you feel?
0
10727
2305
00:13
Do you find it beautiful?
1
13032
1669
00:14
Is it creative?
2
14701
1636
00:16
Now, would you change your answers
3
16337
1705
00:18
if you learned the composer was this robot?
4
18042
3480
00:21
Believe it or not,
5
21522
1979
00:23
people have been grappling with the question of artificial creativity,
6
23501
3317
00:26
alongside the question of artifcial intelligence,
7
26818
3007
00:29
for over 170 years.
8
29825
3007
00:32
In 1843, Lady Ada Lovelace,
9
32832
3008
00:35
an English mathematician considered the world's first computer programmer,
10
35840
4309
00:40
wrote that a machine could not have human-like intelligence
11
40149
3428
00:43
as long as it only did what humans intentionally programmed it to do.
12
43577
5430
00:49
According to Lovelace,
13
49007
1702
00:50
a machine must be able to create original ideas
14
50709
2985
00:53
if it is to be considered intelligent.
15
53694
3698
00:57
The Lovelace Test, formalized in 2001, proposes a way of scrutinizing this idea.
16
57392
6422
01:03
A machine can pass this test if it can produce an outcome
17
63814
3442
01:07
that its designers cannot explain based on their original code.
18
67256
5004
01:12
The Lovelace Test is, by design, more of a thought experiment
19
72260
3349
01:15
than an objective scientific test.
20
75609
2411
01:18
But it's a place to start.
21
78020
2019
01:20
At first glance,
22
80039
1194
01:21
the idea of a machine creating high quality, original music in this way
23
81233
3920
01:25
might seem impossible.
24
85153
2155
01:27
We could come up with an extremely complex algorithm
25
87308
2894
01:30
using random number generators, chaotic functions, and fuzzy logic
26
90202
4141
01:34
to generate a sequence of musical notes
27
94343
2459
01:36
in a way that would be impossible to track.
28
96802
2674
01:39
But although this would yield countless original melodies never heard before,
29
99476
3795
01:43
only a tiny fraction of them would be worth listening to.
30
103271
3943
01:47
With the computer having no way to distinguish
31
107214
2337
01:49
between those which we would consider beautiful
32
109551
2423
01:51
and those which we won't.
33
111974
2637
01:54
But what if we took a step back
34
114611
1637
01:56
and tried to model a natural process that allows creativity to form?
35
116248
4973
02:01
We happen to know of at least one such process
36
121221
2252
02:03
that has lead to original, valuable, and even beautiful outcomes:
37
123473
4522
02:07
the process of evolution.
38
127995
2788
02:10
And evolutionary algorithms,
39
130783
2198
02:12
or genetic algorithms that mimic biological evolution,
40
132981
3259
02:16
are one promising approach
41
136240
2303
02:18
to making machines generate original and valuable artistic outcomes.
42
138543
6020
02:24
So how can evolution make a machine musically creative?
43
144563
2991
02:27
Well, instead of organisms,
44
147554
2111
02:29
we can start with an initial population of musical phrases,
45
149665
11713
02:41
and a basic algorithm
46
161378
1463
02:42
that mimics reproduction and random mutations
47
162841
2924
02:45
by switching some parts,
48
165765
2412
02:48
combining others,
49
168177
1762
02:49
and replacing random notes.
50
169939
2414
02:52
Now that we have a new generation of phrases,
51
172353
2698
02:55
we can apply selection using an operation called a fitness function.
52
175051
4954
03:00
Just as biological fitness is determined by external environmental pressures,
53
180005
4302
03:04
our fitness function can be determined by an external melody
54
184307
4157
03:08
chosen by human musicians, or music fans,
55
188464
2846
03:11
to represent the ultimate beautiful melody.
56
191310
8805
03:20
The algorithm can then compare between our musical phrases
57
200115
3186
03:23
and that beautiful melody,
58
203301
1797
03:25
and select only the phrases that are most similar to it.
59
205098
3853
03:28
Once the least similar sequences are weeded out,
60
208951
3018
03:31
the algorithm can reapply mutation and recombination to what's left,
61
211969
4529
03:36
select the most similar, or fitted ones, again from the new generation,
62
216498
4304
03:40
and repeat for many generations.
63
220802
7610
03:48
The process that got us there has so much randomness and complexity built in
64
228412
4316
03:52
that the result might pass the Lovelace Test.
65
232728
3570
03:56
More importantly, thanks to the presence of human aesthetic in the process,
66
236298
3948
04:00
we'll theoretically generate melodies we would consider beautiful.
67
240246
9081
04:09
But does this satisfy our intuition for what is truly creative?
68
249327
4240
04:13
Is it enough to make something original and beautiful,
69
253567
2544
04:16
or does creativity require intention and awareness of what is being created?
70
256111
6286
04:22
Perhaps the creativity in this case is really coming from the programmers,
71
262397
3796
04:26
even if they don't understand the process.
72
266193
2825
04:29
What is human creativity, anyways?
73
269018
2186
04:31
Is it something more than a system of interconnected neurons
74
271204
2835
04:34
developed by biological algorithmic processes
75
274039
3143
04:37
and the random experiences that shape our lives?
76
277182
3433
04:40
Order and chaos, machine and human.
77
280615
3537
04:44
These are the dynamos at the heart of machine creativity initiatives
78
284152
4116
04:48
that are currently making music, sculptures, paintings, poetry and more.
79
288268
6002
04:54
The jury may still be out
80
294270
1477
04:55
as to whether it's fair to call these acts of creation creative.
81
295747
4304
05:00
But if a piece of art can make you weep,
82
300051
2381
05:02
or blow your mind,
83
302432
1541
05:03
or send shivers down your spine,
84
303973
2161
05:06
does it really matter who or what created it?
85
306134
3452
About this website

This site will introduce you to YouTube videos that are useful for learning English. You will see English lessons taught by top-notch teachers from around the world. Double-click on the English subtitles displayed on each video page to play the video from there. The subtitles scroll in sync with the video playback. If you have any comments or requests, please contact us using this contact form.

https://forms.gle/WvT1wiN1qDtmnspy7