ABOUT THE SPEAKERS
Sebastian Thrun - Educator, entrepreneur
Sebastian Thrun is a passionate technologist who is constantly looking for new opportunities to make the world better for all of us.

Why you should listen

Sebastian Thrun is an educator, entrepreneur and troublemaker. After a long life as a professor at Stanford University, Thrun resigned from tenure to join Google. At Google, he founded Google X, home to self-driving cars and many other moonshot technologies. Thrun also founded Udacity, an online university with worldwide reach, and Kitty Hawk, a "flying car" company. He has authored 11 books, 400 papers, holds 3 doctorates and has won numerous awards.

More profile about the speaker
Sebastian Thrun | Speaker | TED.com
Chris Anderson - TED Curator
After a long career in journalism and publishing, Chris Anderson became the curator of the TED Conference in 2002 and has developed it as a platform for identifying and disseminating ideas worth spreading.

Why you should listen

Chris Anderson is the Curator of TED, a nonprofit devoted to sharing valuable ideas, primarily through the medium of 'TED Talks' -- short talks that are offered free online to a global audience.

Chris was born in a remote village in Pakistan in 1957. He spent his early years in India, Pakistan and Afghanistan, where his parents worked as medical missionaries, and he attended an American school in the Himalayas for his early education. After boarding school in Bath, England, he went on to Oxford University, graduating in 1978 with a degree in philosophy, politics and economics.

Chris then trained as a journalist, working in newspapers and radio, including two years producing a world news service in the Seychelles Islands.

Back in the UK in 1984, Chris was captivated by the personal computer revolution and became an editor at one of the UK's early computer magazines. A year later he founded Future Publishing with a $25,000 bank loan. The new company initially focused on specialist computer publications but eventually expanded into other areas such as cycling, music, video games, technology and design, doubling in size every year for seven years. In 1994, Chris moved to the United States where he built Imagine Media, publisher of Business 2.0 magazine and creator of the popular video game users website IGN. Chris eventually merged Imagine and Future, taking the combined entity public in London in 1999, under the Future name. At its peak, it published 150 magazines and websites and employed 2,000 people.

This success allowed Chris to create a private nonprofit organization, the Sapling Foundation, with the hope of finding new ways to tackle tough global issues through media, technology, entrepreneurship and, most of all, ideas. In 2001, the foundation acquired the TED Conference, then an annual meeting of luminaries in the fields of Technology, Entertainment and Design held in Monterey, California, and Chris left Future to work full time on TED.

He expanded the conference's remit to cover all topics, including science, business and key global issues, while adding a Fellows program, which now has some 300 alumni, and the TED Prize, which grants its recipients "one wish to change the world." The TED stage has become a place for thinkers and doers from all fields to share their ideas and their work, capturing imaginations, sparking conversation and encouraging discovery along the way.

In 2006, TED experimented with posting some of its talks on the Internet. Their viral success encouraged Chris to begin positioning the organization as a global media initiative devoted to 'ideas worth spreading,' part of a new era of information dissemination using the power of online video. In June 2015, the organization posted its 2,000th talk online. The talks are free to view, and they have been translated into more than 100 languages with the help of volunteers from around the world. Viewership has grown to approximately one billion views per year.

Continuing a strategy of 'radical openness,' in 2009 Chris introduced the TEDx initiative, allowing free licenses to local organizers who wished to organize their own TED-like events. More than 8,000 such events have been held, generating an archive of 60,000 TEDx talks. And three years later, the TED-Ed program was launched, offering free educational videos and tools to students and teachers.

More profile about the speaker
Chris Anderson | Speaker | TED.com
TED2017

Sebastian Thrun and Chris Anderson: What AI is -- and isn't

Filmed:
1,575,780 views

Educator and entrepreneur Sebastian Thrun wants us to use AI to free humanity of repetitive work and unleash our creativity. In an inspiring, informative conversation with TED Curator Chris Anderson, Thrun discusses the progress of deep learning, why we shouldn't fear runaway AI and how society will be better off if dull, tedious work is done with the help of machines. "Only one percent of interesting things have been invented yet," Thrun says. "I believe all of us are insanely creative ... [AI] will empower us to turn creativity into action."
- Educator, entrepreneur
Sebastian Thrun is a passionate technologist who is constantly looking for new opportunities to make the world better for all of us. Full bio - TED Curator
After a long career in journalism and publishing, Chris Anderson became the curator of the TED Conference in 2002 and has developed it as a platform for identifying and disseminating ideas worth spreading. Full bio

Double-click the English transcript below to play the video.

00:12
Chris Anderson: Help us understand
what machine learning is,
0
904
2886
00:15
because that seems to be the key driver
1
3814
2054
00:17
of so much of the excitement
and also of the concern
2
5892
2737
00:20
around artificial intelligence.
3
8653
1494
00:22
How does machine learning work?
4
10171
1643
00:23
Sebastian Thrun: So, artificial
intelligence and machine learning
5
11838
3896
00:27
is about 60 years old
6
15758
2002
00:29
and has not had a great day
in its past until recently.
7
17784
4269
00:34
And the reason is that today,
8
22077
2924
00:37
we have reached a scale
of computing and datasets
9
25025
3973
00:41
that was necessary to make machines smart.
10
29022
2637
00:43
So here's how it works.
11
31683
1751
00:45
If you program a computer today,
say, your phone,
12
33458
3497
00:48
then you hire software engineers
13
36979
2335
00:51
that write a very,
very long kitchen recipe,
14
39338
3854
00:55
like, "If the water is too hot,
turn down the temperature.
15
43216
3132
00:58
If it's too cold, turn up
the temperature."
16
46372
2279
01:00
The recipes are not just 10 lines long.
17
48675
2849
01:03
They are millions of lines long.
18
51548
2603
01:06
A modern cell phone
has 12 million lines of code.
19
54175
4084
01:10
A browser has five million lines of code.
20
58283
2646
01:12
And each bug in this recipe
can cause your computer to crash.
21
60953
4969
01:17
That's why a software engineer
makes so much money.
22
65946
3075
01:21
The new thing now is that computers
can find their own rules.
23
69953
3660
01:25
So instead of an expert
deciphering, step by step,
24
73637
3606
01:29
a rule for every contingency,
25
77267
2148
01:31
what you do now is you give
the computer examples
26
79439
3074
01:34
and have it infer its own rules.
27
82537
1581
01:36
A really good example is AlphaGo,
which recently was won by Google.
28
84142
4306
01:40
Normally, in game playing,
you would really write down all the rules,
29
88472
3687
01:44
but in AlphaGo's case,
30
92183
1785
01:45
the system looked over a million games
31
93992
2066
01:48
and was able to infer its own rules
32
96082
2192
01:50
and then beat the world's
residing Go champion.
33
98298
2738
01:53
That is exciting, because it relieves
the software engineer
34
101853
3509
01:57
of the need of being super smart,
35
105386
1819
01:59
and pushes the burden towards the data.
36
107229
2325
02:01
As I said, the inflection point
where this has become really possible --
37
109578
4534
02:06
very embarrassing, my thesis
was about machine learning.
38
114136
2746
02:08
It was completely
insignificant, don't read it,
39
116906
2205
02:11
because it was 20 years ago
40
119135
1350
02:12
and back then, the computers
were as big as a cockroach brain.
41
120509
2907
02:15
Now they are powerful enough
to really emulate
42
123440
2331
02:17
kind of specialized human thinking.
43
125795
2076
02:19
And then the computers
take advantage of the fact
44
127895
2313
02:22
that they can look at
much more data than people can.
45
130232
2500
02:24
So I'd say AlphaGo looked at
more than a million games.
46
132756
3080
02:27
No human expert can ever
study a million games.
47
135860
2839
02:30
Google has looked at over
a hundred billion web pages.
48
138723
3182
02:33
No person can ever study
a hundred billion web pages.
49
141929
2650
02:36
So as a result,
the computer can find rules
50
144603
2714
02:39
that even people can't find.
51
147341
1755
02:41
CA: So instead of looking ahead
to, "If he does that, I will do that,"
52
149120
4312
02:45
it's more saying, "Here is what
looks like a winning pattern,
53
153456
3072
02:48
here is what looks like
a winning pattern."
54
156552
2079
02:50
ST: Yeah. I mean, think about
how you raise children.
55
158655
2517
02:53
You don't spend the first 18 years
giving kids a rule for every contingency
56
161196
3644
02:56
and set them free
and they have this big program.
57
164864
2347
02:59
They stumble, fall, get up,
they get slapped or spanked,
58
167235
2719
03:01
and they have a positive experience,
a good grade in school,
59
169978
2884
03:04
and they figure it out on their own.
60
172886
1834
03:06
That's happening with computers now,
61
174744
1737
03:08
which makes computer programming
so much easier all of a sudden.
62
176505
3029
03:11
Now we don't have to think anymore.
We just give them lots of data.
63
179558
3175
03:14
CA: And so, this has been key
to the spectacular improvement
64
182757
3422
03:18
in power of self-driving cars.
65
186203
3064
03:21
I think you gave me an example.
66
189291
1739
03:23
Can you explain what's happening here?
67
191054
2685
03:25
ST: This is a drive of a self-driving car
68
193763
3564
03:29
that we happened to have at Udacity
69
197351
1957
03:31
and recently made
into a spin-off called Voyage.
70
199332
2398
03:33
We have used this thing
called deep learning
71
201754
2574
03:36
to train a car to drive itself,
72
204352
1623
03:37
and this is driving
from Mountain View, California,
73
205999
2387
03:40
to San Francisco
74
208410
1168
03:41
on El Camino Real on a rainy day,
75
209602
2259
03:43
with bicyclists and pedestrians
and 133 traffic lights.
76
211885
3524
03:47
And the novel thing here is,
77
215433
2636
03:50
many, many moons ago, I started
the Google self-driving car team.
78
218093
3120
03:53
And back in the day, I hired
the world's best software engineers
79
221237
3181
03:56
to find the world's best rules.
80
224442
1607
03:58
This is just trained.
81
226073
1754
03:59
We drive this road 20 times,
82
227851
3336
04:03
we put all this data
into the computer brain,
83
231211
2447
04:05
and after a few hours of processing,
84
233682
2082
04:07
it comes up with behavior
that often surpasses human agility.
85
235788
3926
04:11
So it's become really easy to program it.
86
239738
2017
04:13
This is 100 percent autonomous,
about 33 miles, an hour and a half.
87
241779
3803
04:17
CA: So, explain it -- on the big part
of this program on the left,
88
245606
3630
04:21
you're seeing basically what
the computer sees as trucks and cars
89
249260
3257
04:24
and those dots overtaking it and so forth.
90
252541
2886
04:27
ST: On the right side, you see the camera
image, which is the main input here,
91
255451
3762
04:31
and it's used to find lanes,
other cars, traffic lights.
92
259237
2676
04:33
The vehicle has a radar
to do distance estimation.
93
261937
2489
04:36
This is very commonly used
in these kind of systems.
94
264450
2621
04:39
On the left side you see a laser diagram,
95
267095
1992
04:41
where you see obstacles like trees
and so on depicted by the laser.
96
269111
3200
04:44
But almost all the interesting work
is centering on the camera image now.
97
272335
3436
04:47
We're really shifting over from precision
sensors like radars and lasers
98
275795
3476
04:51
into very cheap, commoditized sensors.
99
279295
1842
04:53
A camera costs less than eight dollars.
100
281161
1987
04:55
CA: And that green dot
on the left thing, what is that?
101
283172
2793
04:57
Is that anything meaningful?
102
285989
1371
04:59
ST: This is a look-ahead point
for your adaptive cruise control,
103
287384
3668
05:03
so it helps us understand
how to regulate velocity
104
291076
2477
05:05
based on how far
the cars in front of you are.
105
293577
2634
05:08
CA: And so, you've also
got an example, I think,
106
296235
2716
05:10
of how the actual
learning part takes place.
107
298975
2381
05:13
Maybe we can see that. Talk about this.
108
301380
2458
05:15
ST: This is an example where we posed
a challenge to Udacity students
109
303862
3643
05:19
to take what we call
a self-driving car Nanodegree.
110
307529
3131
05:22
We gave them this dataset
111
310684
1495
05:24
and said "Hey, can you guys figure out
how to steer this car?"
112
312203
3054
05:27
And if you look at the images,
113
315281
1624
05:28
it's, even for humans, quite impossible
to get the steering right.
114
316929
4073
05:33
And we ran a competition and said,
"It's a deep learning competition,
115
321026
3591
05:36
AI competition,"
116
324641
1173
05:37
and we gave the students 48 hours.
117
325838
1887
05:39
So if you are a software house
like Google or Facebook,
118
327749
4172
05:43
something like this costs you
at least six months of work.
119
331945
2717
05:46
So we figured 48 hours is great.
120
334686
2202
05:48
And within 48 hours, we got about
100 submissions from students,
121
336912
3467
05:52
and the top four got it perfectly right.
122
340403
3370
05:55
It drives better than I could
drive on this imagery,
123
343797
2640
05:58
using deep learning.
124
346461
1189
05:59
And again, it's the same methodology.
125
347674
1799
06:01
It's this magical thing.
126
349497
1164
06:02
When you give enough data
to a computer now,
127
350685
2085
06:04
and give enough time
to comprehend the data,
128
352794
2140
06:06
it finds its own rules.
129
354958
1445
06:09
CA: And so that has led to the development
of powerful applications
130
357339
4845
06:14
in all sorts of areas.
131
362208
1525
06:15
You were talking to me
the other day about cancer.
132
363757
2668
06:18
Can I show this video?
133
366449
1189
06:19
ST: Yeah, absolutely, please.
CA: This is cool.
134
367662
2354
06:22
ST: This is kind of an insight
into what's happening
135
370040
3534
06:25
in a completely different domain.
136
373598
2429
06:28
This is augmenting, or competing --
137
376051
3752
06:31
it's in the eye of the beholder --
138
379827
1749
06:33
with people who are being paid
400,000 dollars a year,
139
381600
3454
06:37
dermatologists,
140
385078
1237
06:38
highly trained specialists.
141
386339
1983
06:40
It takes more than a decade of training
to be a good dermatologist.
142
388346
3561
06:43
What you see here is
the machine learning version of it.
143
391931
3196
06:47
It's called a neural network.
144
395151
1841
06:49
"Neural networks" is the technical term
for these machine learning algorithms.
145
397016
3742
06:52
They've been around since the 1980s.
146
400782
1789
06:54
This one was invented in 1988
by a Facebook Fellow called Yann LeCun,
147
402595
4640
06:59
and it propagates data stages
148
407259
3558
07:02
through what you could think of
as the human brain.
149
410841
2578
07:05
It's not quite the same thing,
but it emulates the same thing.
150
413443
2966
07:08
It goes stage after stage.
151
416433
1302
07:09
In the very first stage, it takes
the visual input and extracts edges
152
417759
3637
07:13
and rods and dots.
153
421420
2612
07:16
And the next one becomes
more complicated edges
154
424056
3037
07:19
and shapes like little half-moons.
155
427117
3191
07:22
And eventually, it's able to build
really complicated concepts.
156
430332
4443
07:26
Andrew Ng has been able to show
157
434799
2048
07:28
that it's able to find
cat faces and dog faces
158
436871
3480
07:32
in vast amounts of images.
159
440375
1661
07:34
What my student team
at Stanford has shown is that
160
442060
2724
07:36
if you train it on 129,000 images
of skin conditions,
161
444808
6073
07:42
including melanoma and carcinomas,
162
450905
2565
07:45
you can do as good a job
163
453494
3301
07:48
as the best human dermatologists.
164
456819
2197
07:51
And to convince ourselves
that this is the case,
165
459040
2549
07:53
we captured an independent dataset
that we presented to our network
166
461613
3990
07:57
and to 25 board-certified
Stanford-level dermatologists,
167
465627
4342
08:01
and compared those.
168
469993
1672
08:03
And in most cases,
169
471689
1504
08:05
they were either on par or above
the performance classification accuracy
170
473217
3875
08:09
of human dermatologists.
171
477116
1467
08:10
CA: You were telling me an anecdote.
172
478607
1746
08:12
I think about this image right here.
173
480377
1957
08:14
What happened here?
174
482358
1484
08:15
ST: This was last Thursday.
That's a moving piece.
175
483866
4008
08:19
What we've shown before and we published
in "Nature" earlier this year
176
487898
3600
08:23
was this idea that we show
dermatologists images
177
491522
2484
08:26
and our computer program images,
178
494030
1539
08:27
and count how often they're right.
179
495593
1627
08:29
But all these images are past images.
180
497244
1778
08:31
They've all been biopsied to make sure
we had the correct classification.
181
499046
3460
08:34
This one wasn't.
182
502530
1172
08:35
This one was actually done at Stanford
by one of our collaborators.
183
503726
3179
08:38
The story goes that our collaborator,
184
506929
2314
08:41
who is a world-famous dermatologist,
one of the three best, apparently,
185
509267
3391
08:44
looked at this mole and said,
"This is not skin cancer."
186
512682
2935
08:47
And then he had
a second moment, where he said,
187
515641
2476
08:50
"Well, let me just check with the app."
188
518141
1866
08:52
So he took out his iPhone
and ran our piece of software,
189
520031
2699
08:54
our "pocket dermatologist," so to speak,
190
522754
2121
08:56
and the iPhone said: cancer.
191
524899
2994
08:59
It said melanoma.
192
527917
1306
09:01
And then he was confused.
193
529849
1233
09:03
And he decided, "OK, maybe I trust
the iPhone a little bit more than myself,"
194
531106
4551
09:07
and he sent it out to the lab
to get it biopsied.
195
535681
2735
09:10
And it came up as an aggressive melanoma.
196
538440
2469
09:13
So I think this might be the first time
that we actually found,
197
541545
3067
09:16
in the practice of using deep learning,
198
544636
2487
09:19
an actual person whose melanoma
would have gone unclassified,
199
547147
3372
09:22
had it not been for deep learning.
200
550543
2115
09:24
CA: I mean, that's incredible.
201
552682
1560
09:26
(Applause)
202
554266
1769
09:28
It feels like there'd be an instant demand
for an app like this right now,
203
556059
3600
09:31
that you might freak out a lot of people.
204
559683
1966
09:33
Are you thinking of doing this,
making an app that allows self-checking?
205
561673
3527
09:37
ST: So my in-box is flooded
about cancer apps,
206
565224
4973
09:42
with heartbreaking stories of people.
207
570221
2303
09:44
I mean, some people have had
10, 15, 20 melanomas removed,
208
572548
3204
09:47
and are scared that one
might be overlooked, like this one,
209
575776
3952
09:51
and also, about, I don't know,
210
579752
1741
09:53
flying cars and speaker inquiries
these days, I guess.
211
581517
2732
09:56
My take is, we need more testing.
212
584273
2738
09:59
I want to be very careful.
213
587449
1778
10:01
It's very easy to give a flashy result
and impress a TED audience.
214
589251
3666
10:04
It's much harder to put
something out that's ethical.
215
592941
2627
10:07
And if people were to use the app
216
595592
2394
10:10
and choose not to consult
the assistance of a doctor
217
598010
2797
10:12
because we get it wrong,
218
600831
1583
10:14
I would feel really bad about it.
219
602438
1653
10:16
So we're currently doing clinical tests,
220
604115
1925
10:18
and if these clinical tests commence
and our data holds up,
221
606064
2798
10:20
we might be able at some point
to take this kind of technology
222
608886
2990
10:23
and take it out of the Stanford clinic
223
611900
1892
10:25
and bring it to the entire world,
224
613816
1658
10:27
places where Stanford
doctors never, ever set foot.
225
615498
2456
10:30
CA: And do I hear this right,
226
618617
2580
10:33
that it seemed like what you were saying,
227
621221
1966
10:35
because you are working
with this army of Udacity students,
228
623211
4254
10:39
that in a way, you're applying
a different form of machine learning
229
627489
3221
10:42
than might take place in a company,
230
630734
1735
10:44
which is you're combining machine learning
with a form of crowd wisdom.
231
632493
3484
10:48
Are you saying that sometimes you think
that could actually outperform
232
636001
3384
10:51
what a company can do,
even a vast company?
233
639409
2050
10:53
ST: I believe there's now
instances that blow my mind,
234
641483
2940
10:56
and I'm still trying to understand.
235
644447
1758
10:58
What Chris is referring to
is these competitions that we run.
236
646229
3937
11:02
We turn them around in 48 hours,
237
650190
2268
11:04
and we've been able to build
a self-driving car
238
652482
2252
11:06
that can drive from Mountain View
to San Francisco on surface streets.
239
654758
3387
11:10
It's not quite on par with Google
after seven years of Google work,
240
658169
3584
11:13
but it's getting there.
241
661777
2528
11:16
And it took us only two engineers
and three months to do this.
242
664329
3084
11:19
And the reason is, we have
an army of students
243
667437
2856
11:22
who participate in competitions.
244
670317
1850
11:24
We're not the only ones
who use crowdsourcing.
245
672191
2220
11:26
Uber and Didi use crowdsource for driving.
246
674435
2223
11:28
Airbnb uses crowdsourcing for hotels.
247
676682
2759
11:31
There's now many examples
where people do bug-finding crowdsourcing
248
679465
4007
11:35
or protein folding, of all things,
in crowdsourcing.
249
683496
2804
11:38
But we've been able to build
this car in three months,
250
686324
2915
11:41
so I am actually rethinking
251
689263
3655
11:44
how we organize corporations.
252
692942
2238
11:47
We have a staff of 9,000 people
who are never hired,
253
695204
4696
11:51
that I never fire.
254
699924
1308
11:53
They show up to work
and I don't even know.
255
701256
2362
11:55
Then they submit to me
maybe 9,000 answers.
256
703642
3058
11:58
I'm not obliged to use any of those.
257
706724
2176
12:00
I end up -- I pay only the winners,
258
708924
1991
12:02
so I'm actually very cheapskate here,
which is maybe not the best thing to do.
259
710939
3718
12:06
But they consider it part
of their education, too, which is nice.
260
714681
3185
12:09
But these students have been able
to produce amazing deep learning results.
261
717890
4201
12:14
So yeah, the synthesis of great people
and great machine learning is amazing.
262
722115
3861
12:18
CA: I mean, Gary Kasparov said on
the first day [of TED2017]
263
726000
2814
12:20
that the winners of chess, surprisingly,
turned out to be two amateur chess players
264
728848
5412
12:26
with three mediocre-ish,
mediocre-to-good, computer programs,
265
734284
5371
12:31
that could outperform one grand master
with one great chess player,
266
739679
3163
12:34
like it was all part of the process.
267
742866
1743
12:36
And it almost seems like
you're talking about a much richer version
268
744633
3335
12:39
of that same idea.
269
747992
1200
12:41
ST: Yeah, I mean, as you followed
the fantastic panels yesterday morning,
270
749216
3857
12:45
two sessions about AI,
271
753097
1994
12:47
robotic overlords and the human response,
272
755115
2167
12:49
many, many great things were said.
273
757306
1982
12:51
But one of the concerns is
that we sometimes confuse
274
759312
2687
12:54
what's actually been done with AI
with this kind of overlord threat,
275
762023
4062
12:58
where your AI develops
consciousness, right?
276
766109
3424
13:01
The last thing I want
is for my AI to have consciousness.
277
769557
2971
13:04
I don't want to come into my kitchen
278
772552
1716
13:06
and have the refrigerator fall in love
with the dishwasher
279
774292
4193
13:10
and tell me, because I wasn't nice enough,
280
778509
2124
13:12
my food is now warm.
281
780657
1837
13:14
I wouldn't buy these products,
and I don't want them.
282
782518
2891
13:17
But the truth is, for me,
283
785825
1802
13:19
AI has always been
an augmentation of people.
284
787651
2720
13:22
It's been an augmentation of us,
285
790893
1676
13:24
to make us stronger.
286
792593
1457
13:26
And I think Kasparov was exactly correct.
287
794074
2831
13:28
It's been the combination
of human smarts and machine smarts
288
796929
3849
13:32
that make us stronger.
289
800802
1464
13:34
The theme of machines making us stronger
is as old as machines are.
290
802290
4587
13:39
The agricultural revolution took
place because it made steam engines
291
807567
3758
13:43
and farming equipment
that couldn't farm by itself,
292
811349
2666
13:46
that never replaced us;
it made us stronger.
293
814039
2122
13:48
And I believe this new wave of AI
will make us much, much stronger
294
816185
3738
13:51
as a human race.
295
819947
1183
13:53
CA: We'll come on to that a bit more,
296
821765
1813
13:55
but just to continue with the scary part
of this for some people,
297
823602
3671
13:59
like, what feels like it gets
scary for people is when you have
298
827297
3558
14:02
a computer that can, one,
rewrite its own code,
299
830879
4618
14:07
so, it can create
multiple copies of itself,
300
835521
3584
14:11
try a bunch of different code versions,
301
839129
1897
14:13
possibly even at random,
302
841050
1775
14:14
and then check them out and see
if a goal is achieved and improved.
303
842849
3632
14:18
So, say the goal is to do better
on an intelligence test.
304
846505
3641
14:22
You know, a computer
that's moderately good at that,
305
850170
3894
14:26
you could try a million versions of that.
306
854088
2509
14:28
You might find one that was better,
307
856621
2090
14:30
and then, you know, repeat.
308
858735
2004
14:32
And so the concern is that you get
some sort of runaway effect
309
860763
3040
14:35
where everything is fine
on Thursday evening,
310
863827
3008
14:38
and you come back into the lab
on Friday morning,
311
866859
2336
14:41
and because of the speed
of computers and so forth,
312
869219
2449
14:43
things have gone crazy, and suddenly --
313
871692
1903
14:45
ST: I would say this is a possibility,
314
873619
2020
14:47
but it's a very remote possibility.
315
875663
1916
14:49
So let me just translate
what I heard you say.
316
877603
3337
14:52
In the AlphaGo case,
we had exactly this thing:
317
880964
2704
14:55
the computer would play
the game against itself
318
883692
2315
14:58
and then learn new rules.
319
886031
1250
14:59
And what machine learning is
is a rewriting of the rules.
320
887305
3235
15:02
It's the rewriting of code.
321
890564
1769
15:04
But I think there was
absolutely no concern
322
892357
2845
15:07
that AlphaGo would take over the world.
323
895226
2426
15:09
It can't even play chess.
324
897676
1464
15:11
CA: No, no, no, but now,
these are all very single-domain things.
325
899164
5147
15:16
But it's possible to imagine.
326
904335
2879
15:19
I mean, we just saw a computer
that seemed nearly capable
327
907238
3089
15:22
of passing a university entrance test,
328
910351
2655
15:25
that can kind of -- it can't read
and understand in the sense that we can,
329
913030
3688
15:28
but it can certainly absorb all the text
330
916742
1987
15:30
and maybe see increased
patterns of meaning.
331
918753
2899
15:33
Isn't there a chance that,
as this broadens out,
332
921676
3694
15:37
there could be a different
kind of runaway effect?
333
925394
2466
15:39
ST: That's where
I draw the line, honestly.
334
927884
2078
15:41
And the chance exists --
I don't want to downplay it --
335
929986
2643
15:44
but I think it's remote, and it's not
the thing that's on my mind these days,
336
932653
3672
15:48
because I think the big revolution
is something else.
337
936349
2512
15:50
Everything successful in AI
to the present date
338
938885
2922
15:53
has been extremely specialized,
339
941831
2214
15:56
and it's been thriving on a single idea,
340
944069
2489
15:58
which is massive amounts of data.
341
946582
2739
16:01
The reason AlphaGo works so well
is because of massive numbers of Go plays,
342
949345
4147
16:05
and AlphaGo can't drive a car
or fly a plane.
343
953516
3255
16:08
The Google self-driving car
or the Udacity self-driving car
344
956795
3031
16:11
thrives on massive amounts of data,
and it can't do anything else.
345
959850
3240
16:15
It can't even control a motorcycle.
346
963114
1727
16:16
It's a very specific,
domain-specific function,
347
964865
2762
16:19
and the same is true for our cancer app.
348
967651
1907
16:21
There has been almost no progress
on this thing called "general AI,"
349
969582
3236
16:24
where you go to an AI and say,
"Hey, invent for me special relativity
350
972842
4000
16:28
or string theory."
351
976866
1666
16:30
It's totally in the infancy.
352
978556
1931
16:32
The reason I want to emphasize this,
353
980511
2127
16:34
I see the concerns,
and I want to acknowledge them.
354
982662
3838
16:38
But if I were to think about one thing,
355
986524
2886
16:41
I would ask myself the question,
"What if we can take anything repetitive
356
989434
5563
16:47
and make ourselves
100 times as efficient?"
357
995021
3473
16:51
It so turns out, 300 years ago,
we all worked in agriculture
358
999170
4249
16:55
and did farming and did repetitive things.
359
1003443
2051
16:57
Today, 75 percent of us work in offices
360
1005518
2556
17:00
and do repetitive things.
361
1008098
2124
17:02
We've become spreadsheet monkeys.
362
1010246
2183
17:04
And not just low-end labor.
363
1012453
2054
17:06
We've become dermatologists
doing repetitive things,
364
1014531
2754
17:09
lawyers doing repetitive things.
365
1017309
1749
17:11
I think we are at the brink
of being able to take an AI,
366
1019082
3823
17:14
look over our shoulders,
367
1022929
1718
17:16
and they make us maybe 10 or 50 times
as effective in these repetitive things.
368
1024671
4058
17:20
That's what is on my mind.
369
1028753
1275
17:22
CA: That sounds super exciting.
370
1030052
2450
17:24
The process of getting there seems
a little terrifying to some people,
371
1032526
3530
17:28
because once a computer
can do this repetitive thing
372
1036080
3180
17:31
much better than the dermatologist
373
1039284
3434
17:34
or than the driver, especially,
is the thing that's talked about
374
1042742
3230
17:37
so much now,
375
1045996
1290
17:39
suddenly millions of jobs go,
376
1047310
1958
17:41
and, you know, the country's in revolution
377
1049292
2695
17:44
before we ever get to the more
glorious aspects of what's possible.
378
1052011
4329
17:48
ST: Yeah, and that's an issue,
and it's a big issue,
379
1056364
2517
17:50
and it was pointed out yesterday morning
by several guest speakers.
380
1058905
4196
17:55
Now, prior to me showing up onstage,
381
1063125
2754
17:57
I confessed I'm a positive,
optimistic person,
382
1065903
3739
18:01
so let me give you an optimistic pitch,
383
1069666
2389
18:04
which is, think of yourself
back 300 years ago.
384
1072079
4795
18:08
Europe just survived 140 years
of continuous war,
385
1076898
3996
18:12
none of you could read or write,
386
1080918
1711
18:14
there were no jobs that you hold today,
387
1082653
2945
18:17
like investment banker
or software engineer or TV anchor.
388
1085622
4096
18:21
We would all be in the fields and farming.
389
1089742
2414
18:24
Now here comes little Sebastian
with a little steam engine in his pocket,
390
1092180
3573
18:27
saying, "Hey guys, look at this.
391
1095777
1548
18:29
It's going to make you 100 times
as strong, so you can do something else."
392
1097349
3595
18:32
And then back in the day,
there was no real stage,
393
1100968
2470
18:35
but Chris and I hang out
with the cows in the stable,
394
1103462
2526
18:38
and he says, "I'm really
concerned about it,
395
1106012
2100
18:40
because I milk my cow every day,
and what if the machine does this for me?"
396
1108136
3652
18:43
The reason why I mention this is,
397
1111812
1702
18:46
we're always good in acknowledging
past progress and the benefit of it,
398
1114360
3603
18:49
like our iPhones or our planes
or electricity or medical supply.
399
1117987
3354
18:53
We all love to live to 80,
which was impossible 300 years ago.
400
1121365
4245
18:57
But we kind of don't apply
the same rules to the future.
401
1125634
4156
19:02
So if I look at my own job as a CEO,
402
1130621
3207
19:05
I would say 90 percent
of my work is repetitive,
403
1133852
3140
19:09
I don't enjoy it,
404
1137016
1351
19:10
I spend about four hours per day
on stupid, repetitive email.
405
1138391
3978
19:14
And I'm burning to have something
that helps me get rid of this.
406
1142393
3641
19:18
Why?
407
1146058
1158
19:19
Because I believe all of us
are insanely creative;
408
1147240
3003
19:22
I think the TED community
more than anybody else.
409
1150731
3194
19:25
But even blue-collar workers;
I think you can go to your hotel maid
410
1153949
3559
19:29
and have a drink with him or her,
411
1157532
2402
19:31
and an hour later,
you find a creative idea.
412
1159958
2717
19:34
What this will empower
is to turn this creativity into action.
413
1162699
4140
19:39
Like, what if you could
build Google in a day?
414
1167265
3442
19:43
What if you could sit over beer
and invent the next Snapchat,
415
1171221
3316
19:46
whatever it is,
416
1174561
1165
19:47
and tomorrow morning it's up and running?
417
1175750
2187
19:49
And that is not science fiction.
418
1177961
1773
19:51
What's going to happen is,
419
1179758
1254
19:53
we are already in history.
420
1181036
1867
19:54
We've unleashed this amazing creativity
421
1182927
3228
19:58
by de-slaving us from farming
422
1186179
1611
19:59
and later, of course, from factory work
423
1187814
3363
20:03
and have invented so many things.
424
1191201
3162
20:06
It's going to be even better,
in my opinion.
425
1194387
2178
20:08
And there's going to be
great side effects.
426
1196589
2072
20:10
One of the side effects will be
427
1198685
1489
20:12
that things like food and medical supply
and education and shelter
428
1200198
4795
20:17
and transportation
429
1205017
1177
20:18
will all become much more
affordable to all of us,
430
1206218
2441
20:20
not just the rich people.
431
1208683
1322
20:22
CA: Hmm.
432
1210029
1182
20:23
So when Martin Ford argued, you know,
that this time it's different
433
1211235
4341
20:27
because the intelligence
that we've used in the past
434
1215600
3453
20:31
to find new ways to be
435
1219077
2483
20:33
will be matched at the same pace
436
1221584
2279
20:35
by computers taking over those things,
437
1223887
2291
20:38
what I hear you saying
is that, not completely,
438
1226202
3078
20:41
because of human creativity.
439
1229304
2951
20:44
Do you think that that's fundamentally
different from the kind of creativity
440
1232279
3785
20:48
that computers can do?
441
1236088
2696
20:50
ST: So, that's my firm
belief as an AI person --
442
1238808
4434
20:55
that I haven't seen
any real progress on creativity
443
1243266
3803
20:59
and out-of-the-box thinking.
444
1247949
1407
21:01
What I see right now -- and this is
really important for people to realize,
445
1249380
3623
21:05
because the word "artificial
intelligence" is so threatening,
446
1253027
2903
21:07
and then we have Steve Spielberg
tossing a movie in,
447
1255954
2523
21:10
where all of a sudden
the computer is our overlord,
448
1258501
2413
21:12
but it's really a technology.
449
1260938
1452
21:14
It's a technology that helps us
do repetitive things.
450
1262414
2982
21:17
And the progress has been
entirely on the repetitive end.
451
1265420
2913
21:20
It's been in legal document discovery.
452
1268357
2228
21:22
It's been contract drafting.
453
1270609
1680
21:24
It's been screening X-rays of your chest.
454
1272313
4223
21:28
And these things are so specialized,
455
1276560
1773
21:30
I don't see the big threat of humanity.
456
1278357
2391
21:32
In fact, we as people --
457
1280772
1794
21:34
I mean, let's face it:
we've become superhuman.
458
1282590
2385
21:36
We've made us superhuman.
459
1284999
1764
21:38
We can swim across
the Atlantic in 11 hours.
460
1286787
2632
21:41
We can take a device out of our pocket
461
1289443
2074
21:43
and shout all the way to Australia,
462
1291541
2147
21:45
and in real time, have that person
shouting back to us.
463
1293712
2600
21:48
That's physically not possible.
We're breaking the rules of physics.
464
1296336
3624
21:51
When this is said and done,
we're going to remember everything
465
1299984
2943
21:54
we've ever said and seen,
466
1302951
1213
21:56
you'll remember every person,
467
1304188
1496
21:57
which is good for me
in my early stages of Alzheimer's.
468
1305708
2626
22:00
Sorry, what was I saying? I forgot.
469
1308358
1677
22:02
CA: (Laughs)
470
1310059
1578
22:03
ST: We will probably have
an IQ of 1,000 or more.
471
1311661
3077
22:06
There will be no more
spelling classes for our kids,
472
1314762
3425
22:10
because there's no spelling issue anymore.
473
1318211
2086
22:12
There's no math issue anymore.
474
1320321
1832
22:14
And I think what really will happen
is that we can be super creative.
475
1322177
3510
22:17
And we are. We are creative.
476
1325711
1857
22:19
That's our secret weapon.
477
1327592
1552
22:21
CA: So the jobs that are getting lost,
478
1329168
2153
22:23
in a way, even though
it's going to be painful,
479
1331345
2494
22:25
humans are capable
of more than those jobs.
480
1333863
2047
22:27
This is the dream.
481
1335934
1218
22:29
The dream is that humans can rise
to just a new level of empowerment
482
1337176
4247
22:33
and discovery.
483
1341447
1657
22:35
That's the dream.
484
1343128
1452
22:36
ST: And think about this:
485
1344604
1643
22:38
if you look at the history of humanity,
486
1346271
2021
22:40
that might be whatever --
60-100,000 years old, give or take --
487
1348316
3328
22:43
almost everything that you cherish
in terms of invention,
488
1351668
3726
22:47
of technology, of things we've built,
489
1355418
2151
22:49
has been invented in the last 150 years.
490
1357593
3099
22:53
If you toss in the book and the wheel,
it's a little bit older.
491
1361756
3048
22:56
Or the axe.
492
1364828
1169
22:58
But your phone, your sneakers,
493
1366021
2790
23:00
these chairs, modern
manufacturing, penicillin --
494
1368835
3551
23:04
the things we cherish.
495
1372410
1714
23:06
Now, that to me means
496
1374148
3658
23:09
the next 150 years will find more things.
497
1377830
3041
23:12
In fact, the pace of invention
has gone up, not gone down, in my opinion.
498
1380895
4154
23:17
I believe only one percent of interesting
things have been invented yet. Right?
499
1385073
4905
23:22
We haven't cured cancer.
500
1390002
1988
23:24
We don't have flying cars -- yet.
Hopefully, I'll change this.
501
1392014
3718
23:27
That used to be an example
people laughed about. (Laughs)
502
1395756
3257
23:31
It's funny, isn't it?
Working secretly on flying cars.
503
1399037
2992
23:34
We don't live twice as long yet. OK?
504
1402053
2683
23:36
We don't have this magic
implant in our brain
505
1404760
2785
23:39
that gives us the information we want.
506
1407569
1832
23:41
And you might be appalled by it,
507
1409425
1526
23:42
but I promise you,
once you have it, you'll love it.
508
1410975
2444
23:45
I hope you will.
509
1413443
1166
23:46
It's a bit scary, I know.
510
1414633
1909
23:48
There are so many things
we haven't invented yet
511
1416566
2254
23:50
that I think we'll invent.
512
1418844
1268
23:52
We have no gravity shields.
513
1420136
1306
23:53
We can't beam ourselves
from one location to another.
514
1421466
2553
23:56
That sounds ridiculous,
515
1424043
1151
23:57
but about 200 years ago,
516
1425218
1288
23:58
experts were of the opinion
that flight wouldn't exist,
517
1426530
2667
24:01
even 120 years ago,
518
1429221
1324
24:02
and if you moved faster
than you could run,
519
1430569
2582
24:05
you would instantly die.
520
1433175
1520
24:06
So who says we are correct today
that you can't beam a person
521
1434719
3569
24:10
from here to Mars?
522
1438312
2249
24:12
CA: Sebastian, thank you so much
523
1440585
1569
24:14
for your incredibly inspiring vision
and your brilliance.
524
1442178
2682
24:16
Thank you, Sebastian Thrun.
525
1444884
1323
24:18
ST: That was fantastic. (Applause)
526
1446231
1895

▲Back to top

ABOUT THE SPEAKERS
Sebastian Thrun - Educator, entrepreneur
Sebastian Thrun is a passionate technologist who is constantly looking for new opportunities to make the world better for all of us.

Why you should listen

Sebastian Thrun is an educator, entrepreneur and troublemaker. After a long life as a professor at Stanford University, Thrun resigned from tenure to join Google. At Google, he founded Google X, home to self-driving cars and many other moonshot technologies. Thrun also founded Udacity, an online university with worldwide reach, and Kitty Hawk, a "flying car" company. He has authored 11 books, 400 papers, holds 3 doctorates and has won numerous awards.

More profile about the speaker
Sebastian Thrun | Speaker | TED.com
Chris Anderson - TED Curator
After a long career in journalism and publishing, Chris Anderson became the curator of the TED Conference in 2002 and has developed it as a platform for identifying and disseminating ideas worth spreading.

Why you should listen

Chris Anderson is the Curator of TED, a nonprofit devoted to sharing valuable ideas, primarily through the medium of 'TED Talks' -- short talks that are offered free online to a global audience.

Chris was born in a remote village in Pakistan in 1957. He spent his early years in India, Pakistan and Afghanistan, where his parents worked as medical missionaries, and he attended an American school in the Himalayas for his early education. After boarding school in Bath, England, he went on to Oxford University, graduating in 1978 with a degree in philosophy, politics and economics.

Chris then trained as a journalist, working in newspapers and radio, including two years producing a world news service in the Seychelles Islands.

Back in the UK in 1984, Chris was captivated by the personal computer revolution and became an editor at one of the UK's early computer magazines. A year later he founded Future Publishing with a $25,000 bank loan. The new company initially focused on specialist computer publications but eventually expanded into other areas such as cycling, music, video games, technology and design, doubling in size every year for seven years. In 1994, Chris moved to the United States where he built Imagine Media, publisher of Business 2.0 magazine and creator of the popular video game users website IGN. Chris eventually merged Imagine and Future, taking the combined entity public in London in 1999, under the Future name. At its peak, it published 150 magazines and websites and employed 2,000 people.

This success allowed Chris to create a private nonprofit organization, the Sapling Foundation, with the hope of finding new ways to tackle tough global issues through media, technology, entrepreneurship and, most of all, ideas. In 2001, the foundation acquired the TED Conference, then an annual meeting of luminaries in the fields of Technology, Entertainment and Design held in Monterey, California, and Chris left Future to work full time on TED.

He expanded the conference's remit to cover all topics, including science, business and key global issues, while adding a Fellows program, which now has some 300 alumni, and the TED Prize, which grants its recipients "one wish to change the world." The TED stage has become a place for thinkers and doers from all fields to share their ideas and their work, capturing imaginations, sparking conversation and encouraging discovery along the way.

In 2006, TED experimented with posting some of its talks on the Internet. Their viral success encouraged Chris to begin positioning the organization as a global media initiative devoted to 'ideas worth spreading,' part of a new era of information dissemination using the power of online video. In June 2015, the organization posted its 2,000th talk online. The talks are free to view, and they have been translated into more than 100 languages with the help of volunteers from around the world. Viewership has grown to approximately one billion views per year.

Continuing a strategy of 'radical openness,' in 2009 Chris introduced the TEDx initiative, allowing free licenses to local organizers who wished to organize their own TED-like events. More than 8,000 such events have been held, generating an archive of 60,000 TEDx talks. And three years later, the TED-Ed program was launched, offering free educational videos and tools to students and teachers.

More profile about the speaker
Chris Anderson | Speaker | TED.com