TED2018
Kai-Fu Lee: How AI can save our humanity
Filmed:
Readability: 4.7
3,583,197 views
AI is massively transforming our world, but there's one thing it cannot do: love. In a visionary talk, computer scientist Kai-Fu Lee details how the US and China are driving a deep learning revolution -- and shares a blueprint for how humans can thrive in the age of AI by harnessing compassion and creativity. "AI is serendipity," Lee says. "It is here to liberate us from routine jobs, and it is here to remind us what it is that makes us human."
Kai-Fu Lee - Investor, computer scientist
At Sinovation Ventures, Kai-Fu Lee invests in the next generation of Chinese high-tech companies. Full bio
At Sinovation Ventures, Kai-Fu Lee invests in the next generation of Chinese high-tech companies. Full bio
Double-click the English transcript below to play the video.
00:12
I'm going to talk about
how AI and mankind can coexist,
how AI and mankind can coexist,
0
640
4536
00:17
but first, we have to rethink
about our human values.
about our human values.
1
5200
3816
00:21
So let me first make a confession
about my errors in my values.
about my errors in my values.
2
9040
4320
00:25
It was 11 o'clock, December 16, 1991.
3
13920
4256
00:30
I was about to become a father
for the first time.
for the first time.
4
18200
2680
00:33
My wife, Shen-Ling,
lay in the hospital bed
lay in the hospital bed
5
21480
2856
00:36
going through a very difficult
12-hour labor.
12-hour labor.
6
24360
3080
00:40
I sat by her bedside
7
28280
2296
00:42
but looked anxiously at my watch,
8
30600
2376
00:45
and I knew something that she didn't.
9
33000
1800
00:47
I knew that if in one hour,
10
35440
1800
00:50
our child didn't come,
11
38360
1936
00:52
I was going to leave her there
12
40320
1936
00:54
and go back to work
13
42280
1656
00:55
and make a presentation about AI
14
43960
2576
00:58
to my boss, Apple's CEO.
15
46560
2200
01:03
Fortunately, my daughter
was born at 11:30 --
was born at 11:30 --
16
51160
3896
01:07
(Laughter)
17
55080
1936
01:09
(Applause)
18
57040
1936
01:11
sparing me from doing the unthinkable,
19
59000
4336
01:15
and to this day, I am so sorry
20
63360
2496
01:17
for letting my work ethic
take precedence over love for my family.
take precedence over love for my family.
21
65880
4656
01:22
(Applause)
22
70560
5736
01:28
My AI talk, however, went off brilliantly.
23
76320
2616
01:30
(Laughter)
24
78960
2736
01:33
Apple loved my work
and decided to announce it
and decided to announce it
25
81720
3136
01:36
at TED1992,
26
84880
2696
01:39
26 years ago on this very stage.
27
87600
3016
01:42
I thought I had made one of the biggest,
most important discoveries in AI,
most important discoveries in AI,
28
90640
5056
01:47
and so did the "Wall Street Journal"
on the following day.
on the following day.
29
95720
2920
01:51
But as far as discoveries went,
30
99440
2536
01:54
it turned out,
31
102000
1336
01:55
I didn't discover India, or America.
32
103360
2776
01:58
Perhaps I discovered
a little island off of Portugal.
a little island off of Portugal.
33
106160
3080
02:02
But the AI era of discovery continued,
34
110720
3256
02:06
and more scientists
poured their souls into it.
poured their souls into it.
35
114000
3216
02:09
About 10 years ago, the grand AI discovery
36
117240
2616
02:11
was made by three
North American scientists,
North American scientists,
37
119880
3176
02:15
and it's known as deep learning.
38
123080
1600
02:17
Deep learning is a technology
that can take a huge amount of data
that can take a huge amount of data
39
125400
3696
02:21
within one single domain
40
129120
1736
02:22
and learn to predict or decide
at superhuman accuracy.
at superhuman accuracy.
41
130880
4936
02:27
For example, if we show
the deep learning network
the deep learning network
42
135840
2776
02:30
a massive number of food photos,
43
138640
2816
02:33
it can recognize food
44
141480
1576
02:35
such as hot dog or no hot dog.
45
143080
3296
02:38
(Applause)
46
146400
3136
02:41
Or if we show it many pictures
and videos and sensor data
and videos and sensor data
47
149560
5016
02:46
from driving on the highway,
48
154600
2536
02:49
it can actually drive a car
as well as a human being
as well as a human being
49
157160
3280
02:53
on the highway.
50
161360
1656
02:55
And what if we showed
this deep learning network
this deep learning network
51
163040
2816
02:57
all the speeches made by President Trump?
52
165880
2400
03:01
Then this artificially
intelligent President Trump,
intelligent President Trump,
53
169680
3000
03:05
actually the network --
54
173600
2216
03:07
(Laughter)
55
175840
1656
03:09
can --
56
177520
1216
03:10
(Applause)
57
178760
4056
03:14
You like double oxymorons, huh?
58
182840
2336
03:17
(Laughter)
59
185200
3856
03:21
(Applause)
60
189080
6096
03:27
So this network, if given the request
to make a speech about AI,
to make a speech about AI,
61
195200
5256
03:32
he, or it, might say --
62
200480
2640
03:36
(Recording) Donald Trump:
It's a great thing
It's a great thing
63
204280
2096
03:38
to build a better world
with artificial intelligence.
with artificial intelligence.
64
206400
2936
03:41
Kai-Fu Lee: And maybe in another language?
65
209360
2016
03:43
DT: (Speaking Chinese)
66
211400
1816
03:45
(Laughter)
67
213240
1496
03:46
KFL: You didn't know
he knew Chinese, did you?
he knew Chinese, did you?
68
214760
2160
03:50
So deep learning has become the core
in the era of AI discovery,
in the era of AI discovery,
69
218120
5016
03:55
and that's led by the US.
70
223160
1816
03:57
But we're now in the era
of implementation,
of implementation,
71
225000
3256
04:00
where what really matters is execution,
product quality, speed and data.
product quality, speed and data.
72
228280
5536
04:05
And that's where China comes in.
73
233840
2096
04:07
Chinese entrepreneurs,
74
235960
1576
04:09
who I fund as a venture capitalist,
75
237560
1896
04:11
are incredible workers,
76
239480
1736
04:13
amazing work ethic.
77
241240
1936
04:15
My example in the delivery room is nothing
compared to how hard people work in China.
compared to how hard people work in China.
78
243200
5296
04:20
As an example, one startup
tried to claim work-life balance:
tried to claim work-life balance:
79
248520
3696
04:24
"Come work for us because we are 996."
80
252240
3696
04:27
And what does that mean?
81
255960
1256
04:29
It means the work hours
of 9am to 9pm, six days a week.
of 9am to 9pm, six days a week.
82
257240
4400
04:34
That's contrasted
with other startups that do 997.
with other startups that do 997.
83
262960
3280
04:39
And the Chinese product quality
has consistently gone up
has consistently gone up
84
267320
3096
04:42
in the past decade,
85
270440
1696
04:44
and that's because of
a fiercely competitive environment.
a fiercely competitive environment.
86
272160
4120
04:48
In Silicon Valley, entrepreneurs
compete in a very gentlemanly fashion,
compete in a very gentlemanly fashion,
87
276760
5896
04:54
sort of like in old wars
in which each side took turns
in which each side took turns
88
282680
3896
04:58
to fire at each other.
89
286600
1256
04:59
(Laughter)
90
287880
1056
05:00
But in the Chinese environment,
91
288960
2016
05:03
it's truly a gladiatorial
fight to the death.
fight to the death.
92
291000
2880
05:06
In such a brutal environment,
entrepreneurs learn to grow very rapidly,
entrepreneurs learn to grow very rapidly,
93
294920
6096
05:13
they learn to make their products
better at lightning speed,
better at lightning speed,
94
301040
3936
05:17
and they learn
to hone their business models
to hone their business models
95
305000
2336
05:19
until they're impregnable.
96
307360
1400
05:21
As a result, great Chinese products
like WeChat and Weibo
like WeChat and Weibo
97
309400
3856
05:25
are arguably better
98
313280
1416
05:26
than the equivalent American products
from Facebook and Twitter.
from Facebook and Twitter.
99
314720
3480
05:31
And the Chinese market
embraces this change
embraces this change
100
319920
3096
05:35
and accelerated change
and paradigm shifts.
and paradigm shifts.
101
323040
3016
05:38
As an example, if any of you go to China,
102
326080
2096
05:40
you will see it's almost cashless
and credit card-less,
and credit card-less,
103
328200
4016
05:44
because that thing that we all
talk about, mobile payment,
talk about, mobile payment,
104
332240
2736
05:47
has become the reality in China.
105
335000
2416
05:49
In the last year,
106
337440
1296
05:50
18.8 trillion US dollars
were transacted on mobile internet,
were transacted on mobile internet,
107
338760
5936
05:56
and that's because
of very robust technologies
of very robust technologies
108
344720
2936
05:59
built behind it.
109
347680
1256
06:00
It's even bigger than the China GDP.
110
348960
2440
06:04
And this technology, you can say,
how can it be bigger than the GDP?
how can it be bigger than the GDP?
111
352120
3616
06:07
Because it includes all transactions:
112
355760
1976
06:09
wholesale, channels,
retail, online, offline,
retail, online, offline,
113
357760
3776
06:13
going into a shopping mall
or going into a farmers market like this.
or going into a farmers market like this.
114
361560
5056
06:18
The technology is used
by 700 million people
by 700 million people
115
366640
3376
06:22
to pay each other, not just merchants,
116
370040
2056
06:24
so it's peer to peer,
117
372120
1416
06:25
and it's almost transaction-fee-free.
118
373560
2680
06:29
And it's instantaneous,
119
377720
2376
06:32
and it's used everywhere.
120
380120
1440
06:34
And finally, the China market is enormous.
121
382320
3376
06:37
This market is large,
122
385720
1976
06:39
which helps give entrepreneurs
more users, more revenue,
more users, more revenue,
123
387720
4496
06:44
more investment, but most importantly,
124
392240
2336
06:46
it gives the entrepreneurs a chance
to collect a huge amount of data
to collect a huge amount of data
125
394600
4536
06:51
which becomes rocket fuel
for the AI engine.
for the AI engine.
126
399160
2960
06:54
So as a result, the Chinese AI companies
127
402960
2936
06:57
have leaped ahead
128
405920
1736
06:59
so that today, the most valuable companies
129
407680
3096
07:02
in computer vision, speech recognition,
130
410800
2616
07:05
speech synthesis,
machine translation and drones
machine translation and drones
131
413440
3416
07:08
are all Chinese companies.
132
416880
1920
07:11
So with the US leading
the era of discovery
the era of discovery
133
419400
3216
07:14
and China leading
the era of implementation,
the era of implementation,
134
422640
3136
07:17
we are now in an amazing age
135
425800
2256
07:20
where the dual engine
of the two superpowers
of the two superpowers
136
428080
3256
07:23
are working together
137
431360
1680
07:25
to drive the fastest
revolution in technology
revolution in technology
138
433960
3656
07:29
that we have ever seen as humans.
139
437640
2360
07:32
And this will bring tremendous wealth,
140
440640
2496
07:35
unprecedented wealth:
141
443160
1776
07:36
16 trillion dollars, according to PwC,
142
444960
3896
07:40
in terms of added GDP
to the worldwide GDP by 2030.
to the worldwide GDP by 2030.
143
448880
5376
07:46
It will also bring immense challenges
144
454280
2616
07:48
in terms of potential job replacements.
145
456920
3576
07:52
Whereas in the Industrial Age
146
460520
2160
07:55
it created more jobs
147
463880
1536
07:57
because craftsman jobs were being
decomposed into jobs in the assembly line,
decomposed into jobs in the assembly line,
148
465440
5776
08:03
so more jobs were created.
149
471240
1976
08:05
But AI completely replaces
the individual jobs
the individual jobs
150
473240
4936
08:10
in the assembly line with robots.
151
478200
2256
08:12
And it's not just in factories,
152
480480
1936
08:14
but truckers, drivers
153
482440
2056
08:16
and even jobs like telesales,
customer service
customer service
154
484520
4096
08:20
and hematologists as well as radiologists
155
488640
2976
08:23
over the next 15 years
156
491640
2096
08:25
are going to be gradually replaced
157
493760
2536
08:28
by artificial intelligence.
158
496320
1440
08:30
And only the creative jobs --
159
498360
2056
08:32
(Laughter)
160
500440
1976
08:34
I have to make myself safe, right?
161
502440
2200
08:38
Really, the creative jobs
are the ones that are protected,
are the ones that are protected,
162
506960
2976
08:41
because AI can optimize but not create.
163
509960
3040
08:45
But what's more serious
than the loss of jobs
than the loss of jobs
164
513960
3575
08:49
is the loss of meaning,
165
517559
1777
08:51
because the work ethic
in the Industrial Age
in the Industrial Age
166
519360
3136
08:54
has brainwashed us into thinking
that work is the reason we exist,
that work is the reason we exist,
167
522520
5616
09:00
that work defined
the meaning of our lives.
the meaning of our lives.
168
528160
2936
09:03
And I was a prime and willing victim
to that type of workaholic thinking.
to that type of workaholic thinking.
169
531120
6296
09:09
I worked incredibly hard.
170
537440
1616
09:11
That's why I almost left
my wife in the delivery room,
my wife in the delivery room,
171
539080
3576
09:14
that's why I worked 996
alongside my entrepreneurs.
alongside my entrepreneurs.
172
542680
4176
09:18
And that obsession that I had with work
173
546880
4416
09:23
ended abruptly a few years ago
174
551320
3056
09:26
when I was diagnosed
with fourth stage lymphoma.
with fourth stage lymphoma.
175
554400
3920
09:31
The PET scan here shows
over 20 malignant tumors
over 20 malignant tumors
176
559480
4136
09:35
jumping out like fireballs,
177
563640
2176
09:37
melting away my ambition.
178
565840
2576
09:40
But more importantly,
179
568440
1456
09:41
it helped me reexamine my life.
180
569920
2736
09:44
Knowing that I may only have
a few months to live
a few months to live
181
572680
3096
09:47
caused me to see how foolish it was
182
575800
2936
09:50
for me to base my entire self-worth
183
578760
3576
09:54
on how hard I worked
and the accomplishments from hard work.
and the accomplishments from hard work.
184
582360
4040
09:59
My priorities were
completely out of order.
completely out of order.
185
587320
2976
10:02
I neglected my family.
186
590320
1600
10:05
My father had passed away,
187
593000
1416
10:06
and I never had a chance
to tell him I loved him.
to tell him I loved him.
188
594440
2800
10:10
My mother had dementia
and no longer recognized me,
and no longer recognized me,
189
598120
3736
10:13
and my children had grown up.
190
601880
1920
10:16
During my chemotherapy,
191
604400
1656
10:18
I read a book by Bronnie Ware
192
606080
2456
10:20
who talked about dying wishes and regrets
of the people in the deathbed.
of the people in the deathbed.
193
608560
5496
10:26
She found that facing death,
194
614080
2256
10:28
nobody regretted that they didn't
work hard enough in this life.
work hard enough in this life.
195
616360
3600
10:32
They only regretted that they didn't
spend enough time with their loved ones
spend enough time with their loved ones
196
620880
5176
10:38
and that they didn't spread their love.
197
626080
2760
10:42
So I am fortunately today in remission.
198
630400
4536
10:46
(Applause)
199
634960
6856
10:53
So I can be back at TED again
200
641840
1976
10:55
to share with you
that I have changed my ways.
that I have changed my ways.
201
643840
3376
10:59
I now only work 965 --
202
647240
2760
11:03
occasionally 996, but usually 965.
203
651200
3976
11:07
I moved closer to my mother,
204
655200
1896
11:09
my wife usually travels with me,
205
657120
2456
11:11
and when my kids have vacation,
if they don't come home, I go to them.
if they don't come home, I go to them.
206
659600
3816
11:15
So it's a new form of life
207
663440
2456
11:17
that helped me recognize
208
665920
1816
11:19
how important it is that love is for me,
209
667760
3176
11:22
and facing death
helped me change my life,
helped me change my life,
210
670960
3416
11:26
but it also helped me see a new way
211
674400
2456
11:28
of how AI should impact mankind
212
676880
3776
11:32
and work and coexist with mankind,
213
680680
2480
11:36
that really, AI is taking away
a lot of routine jobs,
a lot of routine jobs,
214
684680
4496
11:41
but routine jobs are not what we're about.
215
689200
3656
11:44
Why we exist is love.
216
692880
2256
11:47
When we hold our newborn baby,
217
695160
2096
11:49
love at first sight,
218
697280
1496
11:50
or when we help someone in need,
219
698800
1776
11:52
humans are uniquely able
to give and receive love,
to give and receive love,
220
700600
4256
11:56
and that's what differentiates us from AI.
221
704880
2800
12:00
Despite what science fiction may portray,
222
708800
2696
12:03
I can responsibly tell you
that AI has no love.
that AI has no love.
223
711520
3736
12:07
When AlphaGo defeated
the world champion Ke Jie,
the world champion Ke Jie,
224
715280
3536
12:10
while Ke Jie was crying
and loving the game of go,
and loving the game of go,
225
718840
3096
12:13
AlphaGo felt no happiness from winning
226
721960
3176
12:17
and certainly no desire
to hug a loved one.
to hug a loved one.
227
725160
4480
12:23
So how do we differentiate ourselves
228
731600
2656
12:26
as humans in the age of AI?
229
734280
2536
12:28
We talked about the axis of creativity,
230
736840
3096
12:31
and certainly that is one possibility,
231
739960
2856
12:34
and now we introduce a new axis
232
742840
2296
12:37
that we can call compassion,
love, or empathy.
love, or empathy.
233
745160
3616
12:40
Those are things that AI cannot do.
234
748800
2576
12:43
So as AI takes away the routine jobs,
235
751400
2816
12:46
I like to think we can, we should
and we must create jobs of compassion.
and we must create jobs of compassion.
236
754240
4960
12:51
You might ask how many of those there are,
237
759800
2336
12:54
but I would ask you:
238
762160
1616
12:55
Do you not think that we are going
to need a lot of social workers
to need a lot of social workers
239
763800
3816
12:59
to help us make this transition?
240
767640
1600
13:01
Do you not think we need
a lot of compassionate caregivers
a lot of compassionate caregivers
241
769960
3256
13:05
to give more medical care to more people?
242
773240
2696
13:07
Do you not think we're going to need
10 times more teachers
10 times more teachers
243
775960
3256
13:11
to help our children find their way
244
779240
2776
13:14
to survive and thrive
in this brave new world?
in this brave new world?
245
782040
3256
13:17
And with all the newfound wealth,
246
785320
2440
13:19
should we not also make
labors of love into careers
labors of love into careers
247
787800
4536
13:24
and let elderly accompaniment
248
792360
2696
13:27
or homeschooling become careers also?
249
795080
3496
13:30
(Applause)
250
798600
5280
13:36
This graph is surely not perfect,
251
804800
2256
13:39
but it points at four ways
that we can work with AI.
that we can work with AI.
252
807080
3536
13:42
AI will come and take away
the routine jobs
the routine jobs
253
810640
3576
13:46
and in due time, we will be thankful.
254
814240
2040
13:49
AI will become great tools
for the creatives
for the creatives
255
817000
3096
13:52
so that scientists, artists,
musicians and writers
musicians and writers
256
820120
3656
13:55
can be even more creative.
257
823800
1600
13:58
AI will work with humans
as analytical tools
as analytical tools
258
826120
5376
14:03
that humans can wrap their warmth around
259
831520
2696
14:06
for the high-compassion jobs.
260
834240
1896
14:08
And we can always differentiate ourselves
261
836160
2656
14:10
with the uniquely capable jobs
262
838840
1816
14:12
that are both compassionate and creative,
263
840680
3616
14:16
using and leveraging
our irreplaceable brains and hearts.
our irreplaceable brains and hearts.
264
844320
5176
14:21
So there you have it:
265
849520
1296
14:22
a blueprint of coexistence
for humans and AI.
for humans and AI.
266
850840
3640
14:27
AI is serendipity.
267
855400
1816
14:29
It is here to liberate us
from routine jobs,
from routine jobs,
268
857240
2976
14:32
and it is here to remind us
what it is that makes us human.
what it is that makes us human.
269
860240
3680
14:36
So let us choose to embrace AI
and to love one another.
and to love one another.
270
864440
3976
14:40
Thank you.
271
868440
1216
14:41
(Applause)
272
869680
6160
ABOUT THE SPEAKER
Kai-Fu Lee - Investor, computer scientistAt Sinovation Ventures, Kai-Fu Lee invests in the next generation of Chinese high-tech companies.
Why you should listen
Dr. Kai-Fu Lee has a unique perspective on the global technology industry, having worked extensively between the US and China, and has been in artificial intelligence research, development and investment for more than 30 years. He is one of China's top technology investors, undertaking early pioneering work in the field of artificial intelligence and working with multiple US technology giants.
Lee is the Chairman and CEO of Sinovation Ventures, managing a $1.7 billion dual-currency investment fund. Sinovation is a leading technology-savvy investment firm focusing on developing Chinese high-tech companies. Lee also serves as president of the Sinovation Ventures Artificial Intelligence Institute.
Prior to founding Sinovation in 2009, Lee was the President of Google China. Previously, he held executive positions at Microsoft, SGI and Apple. He is a Fellow of the Institute of Electrical and Electronics Engineers (IEEE), and he was Vice Chairman of the Committee of 100 and named one of the 100 most influential people in the world by TIME Magazine. He is the author of seven best-selling books in Chinese, launching his new book AI Superpowers internationally in fall 2018, and has over 50 million followers on social media.
In the field of artificial intelligence, Lee founded Microsoft Research China, which was named as the hottest research lab by MIT Technology Review. Later renamed Microsoft Research Asia, this institute has trained the great majority of AI leaders in China, including CTOs or AI heads at Baidu, Tencent, Alibaba, Lenovo, Huawei and Haier. While with Apple, Lee led AI projects in speech and natural language, which appeared on Good Morning America on ABC Television and the front page of The Wall Street Journal. He has authored 10 US patents, and more than 100 journal and conference papers.
Kai-Fu Lee | Speaker | TED.com