Shyam Sankar: The rise of human-computer cooperation

62,464 views ・ 2012-09-06

TED


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

00:00
Translator: Joseph Geni Reviewer: Morton Bast
0
0
7000
00:15
I'd like to tell you about two games of chess.
1
15772
2556
00:18
The first happened in 1997, in which Garry Kasparov,
2
18328
3864
00:22
a human, lost to Deep Blue, a machine.
3
22192
3716
00:25
To many, this was the dawn of a new era,
4
25908
2240
00:28
one where man would be dominated by machine.
5
28148
2779
00:30
But here we are, 20 years on, and the greatest change
6
30927
3334
00:34
in how we relate to computers is the iPad,
7
34261
2690
00:36
not HAL.
8
36951
2045
00:38
The second game was a freestyle chess tournament
9
38996
2648
00:41
in 2005, in which man and machine could enter together
10
41644
2969
00:44
as partners, rather than adversaries, if they so chose.
11
44613
4666
00:49
At first, the results were predictable.
12
49279
1851
00:51
Even a supercomputer was beaten by a grandmaster
13
51130
2497
00:53
with a relatively weak laptop.
14
53627
2312
00:55
The surprise came at the end. Who won?
15
55939
2985
00:58
Not a grandmaster with a supercomputer,
16
58924
2776
01:01
but actually two American amateurs
17
61700
1493
01:03
using three relatively weak laptops.
18
63193
3822
01:07
Their ability to coach and manipulate their computers
19
67015
2596
01:09
to deeply explore specific positions
20
69611
2435
01:12
effectively counteracted the superior chess knowledge
21
72046
2390
01:14
of the grandmasters and the superior computational power
22
74436
2609
01:17
of other adversaries.
23
77045
1909
01:18
This is an astonishing result: average men,
24
78954
2905
01:21
average machines beating the best man, the best machine.
25
81859
4081
01:25
And anyways, isn't it supposed to be man versus machine?
26
85940
3199
01:29
Instead, it's about cooperation, and the right type of cooperation.
27
89139
4152
01:33
We've been paying a lot of attention to Marvin Minsky's
28
93291
2857
01:36
vision for artificial intelligence over the last 50 years.
29
96148
3242
01:39
It's a sexy vision, for sure. Many have embraced it.
30
99390
2262
01:41
It's become the dominant school of thought in computer science.
31
101652
2753
01:44
But as we enter the era of big data, of network systems,
32
104405
3072
01:47
of open platforms, and embedded technology,
33
107477
2698
01:50
I'd like to suggest it's time to reevaluate an alternative vision
34
110175
3392
01:53
that was actually developed around the same time.
35
113567
3070
01:56
I'm talking about J.C.R. Licklider's human-computer symbiosis,
36
116637
3332
01:59
perhaps better termed "intelligence augmentation," I.A.
37
119969
3808
02:03
Licklider was a computer science titan who had a profound
38
123777
2640
02:06
effect on the development of technology and the Internet.
39
126417
3006
02:09
His vision was to enable man and machine to cooperate
40
129423
2868
02:12
in making decisions, controlling complex situations
41
132291
3590
02:15
without the inflexible dependence
42
135881
1770
02:17
on predetermined programs.
43
137651
2533
02:20
Note that word "cooperate."
44
140184
2498
02:22
Licklider encourages us not to take a toaster
45
142682
2747
02:25
and make it Data from "Star Trek,"
46
145429
2284
02:27
but to take a human and make her more capable.
47
147713
3535
02:31
Humans are so amazing -- how we think,
48
151248
1911
02:33
our non-linear approaches, our creativity,
49
153159
2618
02:35
iterative hypotheses, all very difficult if possible at all
50
155777
2131
02:37
for computers to do.
51
157908
1345
02:39
Licklider intuitively realized this, contemplating humans
52
159253
2452
02:41
setting the goals, formulating the hypotheses,
53
161705
2327
02:44
determining the criteria, and performing the evaluation.
54
164032
3376
02:47
Of course, in other ways, humans are so limited.
55
167408
1775
02:49
We're terrible at scale, computation and volume.
56
169183
3235
02:52
We require high-end talent management
57
172418
1836
02:54
to keep the rock band together and playing.
58
174254
2064
02:56
Licklider foresaw computers doing all the routinizable work
59
176318
2204
02:58
that was required to prepare the way for insights and decision making.
60
178522
3276
03:01
Silently, without much fanfare,
61
181798
2224
03:04
this approach has been compiling victories beyond chess.
62
184022
3354
03:07
Protein folding, a topic that shares the incredible expansiveness of chess —
63
187376
3356
03:10
there are more ways of folding a protein than there are atoms in the universe.
64
190732
3042
03:13
This is a world-changing problem with huge implications
65
193774
2353
03:16
for our ability to understand and treat disease.
66
196127
2308
03:18
And for this task, supercomputer field brute force simply isn't enough.
67
198435
4248
03:22
Foldit, a game created by computer scientists,
68
202683
2384
03:25
illustrates the value of the approach.
69
205067
2502
03:27
Non-technical, non-biologist amateurs play a video game
70
207569
3041
03:30
in which they visually rearrange the structure of the protein,
71
210610
3073
03:33
allowing the computer to manage the atomic forces
72
213683
1499
03:35
and interactions and identify structural issues.
73
215182
2957
03:38
This approach beat supercomputers 50 percent of the time
74
218139
3023
03:41
and tied 30 percent of the time.
75
221162
2584
03:43
Foldit recently made a notable and major scientific discovery
76
223746
3137
03:46
by deciphering the structure of the Mason-Pfizer monkey virus.
77
226883
3160
03:50
A protease that had eluded determination for over 10 years
78
230043
3015
03:53
was solved was by three players in a matter of days,
79
233058
2626
03:55
perhaps the first major scientific advance
80
235684
2025
03:57
to come from playing a video game.
81
237709
2323
04:00
Last year, on the site of the Twin Towers,
82
240032
2181
04:02
the 9/11 memorial opened.
83
242213
1473
04:03
It displays the names of the thousands of victims
84
243686
2721
04:06
using a beautiful concept called "meaningful adjacency."
85
246407
3063
04:09
It places the names next to each other based on their
86
249470
2166
04:11
relationships to one another: friends, families, coworkers.
87
251636
2213
04:13
When you put it all together, it's quite a computational
88
253849
3028
04:16
challenge: 3,500 victims, 1,800 adjacency requests,
89
256877
4223
04:21
the importance of the overall physical specifications
90
261100
3092
04:24
and the final aesthetics.
91
264192
2137
04:26
When first reported by the media, full credit for such a feat
92
266329
2615
04:28
was given to an algorithm from the New York City
93
268944
1892
04:30
design firm Local Projects. The truth is a bit more nuanced.
94
270836
4001
04:34
While an algorithm was used to develop the underlying framework,
95
274837
2871
04:37
humans used that framework to design the final result.
96
277708
3008
04:40
So in this case, a computer had evaluated millions
97
280716
2225
04:42
of possible layouts, managed a complex relational system,
98
282941
3335
04:46
and kept track of a very large set of measurements
99
286276
2414
04:48
and variables, allowing the humans to focus
100
288690
2410
04:51
on design and compositional choices.
101
291100
2802
04:53
So the more you look around you,
102
293902
1036
04:54
the more you see Licklider's vision everywhere.
103
294938
1962
04:56
Whether it's augmented reality in your iPhone or GPS in your car,
104
296900
3304
05:00
human-computer symbiosis is making us more capable.
105
300204
2970
05:03
So if you want to improve human-computer symbiosis,
106
303174
1655
05:04
what can you do?
107
304829
1429
05:06
You can start by designing the human into the process.
108
306258
2452
05:08
Instead of thinking about what a computer will do to solve the problem,
109
308710
2204
05:10
design the solution around what the human will do as well.
110
310914
3869
05:14
When you do this, you'll quickly realize that you spent
111
314783
1937
05:16
all of your time on the interface between man and machine,
112
316720
2879
05:19
specifically on designing away the friction in the interaction.
113
319599
3099
05:22
In fact, this friction is more important than the power
114
322698
2766
05:25
of the man or the power of the machine
115
325464
2052
05:27
in determining overall capability.
116
327516
1931
05:29
That's why two amateurs with a few laptops
117
329447
1977
05:31
handily beat a supercomputer and a grandmaster.
118
331424
2456
05:33
What Kasparov calls process is a byproduct of friction.
119
333880
3005
05:36
The better the process, the less the friction.
120
336885
2401
05:39
And minimizing friction turns out to be the decisive variable.
121
339286
4256
05:43
Or take another example: big data.
122
343542
2243
05:45
Every interaction we have in the world is recorded
123
345785
1906
05:47
by an ever growing array of sensors: your phone,
124
347691
3059
05:50
your credit card, your computer. The result is big data,
125
350750
2373
05:53
and it actually presents us with an opportunity
126
353123
1742
05:54
to more deeply understand the human condition.
127
354865
2662
05:57
The major emphasis of most approaches to big data
128
357527
2305
05:59
focus on, "How do I store this data? How do I search
129
359832
2215
06:02
this data? How do I process this data?"
130
362047
2276
06:04
These are necessary but insufficient questions.
131
364323
2204
06:06
The imperative is not to figure out how to compute,
132
366527
2471
06:08
but what to compute. How do you impose human intuition
133
368998
2184
06:11
on data at this scale?
134
371182
1791
06:12
Again, we start by designing the human into the process.
135
372973
3499
06:16
When PayPal was first starting as a business, their biggest
136
376472
2812
06:19
challenge was not, "How do I send money back and forth online?"
137
379284
2804
06:22
It was, "How do I do that without being defrauded by organized crime?"
138
382088
3872
06:25
Why so challenging? Because while computers can learn
139
385960
2088
06:28
to detect and identify fraud based on patterns,
140
388048
3144
06:31
they can't learn to do that based on patterns
141
391192
1479
06:32
they've never seen before, and organized crime
142
392671
2116
06:34
has a lot in common with this audience: brilliant people,
143
394787
2709
06:37
relentlessly resourceful, entrepreneurial spirit — (Laughter) —
144
397496
3640
06:41
and one huge and important difference: purpose.
145
401136
2712
06:43
And so while computers alone can catch all but the cleverest
146
403848
2832
06:46
fraudsters, catching the cleverest is the difference
147
406680
2253
06:48
between success and failure.
148
408933
2545
06:51
There's a whole class of problems like this, ones with
149
411478
2221
06:53
adaptive adversaries. They rarely if ever present with a
150
413699
2575
06:56
repeatable pattern that's discernable to computers.
151
416274
2736
06:59
Instead, there's some inherent component of innovation or disruption,
152
419010
3993
07:03
and increasingly these problems are buried in big data.
153
423003
2735
07:05
For example, terrorism. Terrorists are always adapting
154
425738
2500
07:08
in minor and major ways to new circumstances, and despite
155
428238
2052
07:10
what you might see on TV, these adaptations,
156
430290
3094
07:13
and the detection of them, are fundamentally human.
157
433384
2293
07:15
Computers don't detect novel patterns and new behaviors,
158
435677
3117
07:18
but humans do. Humans, using technology, testing hypotheses,
159
438794
3235
07:22
searching for insight by asking machines to do things for them.
160
442029
4620
07:26
Osama bin Laden was not caught by artificial intelligence.
161
446649
2320
07:28
He was caught by dedicated, resourceful, brilliant people
162
448969
2553
07:31
in partnerships with various technologies.
163
451522
4269
07:35
As appealing as it might sound, you cannot algorithmically
164
455791
2818
07:38
data mine your way to the answer.
165
458609
1601
07:40
There is no "Find Terrorist" button, and the more data
166
460210
2855
07:43
we integrate from a vast variety of sources
167
463065
2302
07:45
across a wide variety of data formats from very
168
465367
2133
07:47
disparate systems, the less effective data mining can be.
169
467500
3309
07:50
Instead, people will have to look at data
170
470809
2024
07:52
and search for insight, and as Licklider foresaw long ago,
171
472833
3456
07:56
the key to great results here is the right type of cooperation,
172
476289
2685
07:58
and as Kasparov realized,
173
478974
1524
08:00
that means minimizing friction at the interface.
174
480498
3031
08:03
Now this approach makes possible things like combing
175
483529
2758
08:06
through all available data from very different sources,
176
486287
3386
08:09
identifying key relationships and putting them in one place,
177
489673
2792
08:12
something that's been nearly impossible to do before.
178
492465
2928
08:15
To some, this has terrifying privacy and civil liberties
179
495393
1942
08:17
implications. To others it foretells of an era of greater
180
497335
3410
08:20
privacy and civil liberties protections,
181
500745
1909
08:22
but privacy and civil liberties are of fundamental importance.
182
502654
2936
08:25
That must be acknowledged, and they can't be swept aside,
183
505590
2193
08:27
even with the best of intents.
184
507783
2530
08:30
So let's explore, through a couple of examples, the impact
185
510313
2518
08:32
that technologies built to drive human-computer symbiosis
186
512831
2406
08:35
have had in recent time.
187
515237
2919
08:38
In October, 2007, U.S. and coalition forces raided
188
518156
3416
08:41
an al Qaeda safe house in the city of Sinjar
189
521572
2416
08:43
on the Syrian border of Iraq.
190
523988
1934
08:45
They found a treasure trove of documents:
191
525922
2376
08:48
700 biographical sketches of foreign fighters.
192
528298
2335
08:50
These foreign fighters had left their families in the Gulf,
193
530633
2584
08:53
the Levant and North Africa to join al Qaeda in Iraq.
194
533217
3146
08:56
These records were human resource forms.
195
536363
1616
08:57
The foreign fighters filled them out as they joined the organization.
196
537979
2855
09:00
It turns out that al Qaeda, too,
197
540834
1211
09:02
is not without its bureaucracy. (Laughter)
198
542045
2597
09:04
They answered questions like, "Who recruited you?"
199
544642
2098
09:06
"What's your hometown?" "What occupation do you seek?"
200
546740
2854
09:09
In that last question, a surprising insight was revealed.
201
549594
3169
09:12
The vast majority of foreign fighters
202
552763
2400
09:15
were seeking to become suicide bombers for martyrdom --
203
555163
2400
09:17
hugely important, since between 2003 and 2007, Iraq
204
557563
4338
09:21
had 1,382 suicide bombings, a major source of instability.
205
561901
4244
09:26
Analyzing this data was hard. The originals were sheets
206
566145
2058
09:28
of paper in Arabic that had to be scanned and translated.
207
568203
2742
09:30
The friction in the process did not allow for meaningful
208
570945
2192
09:33
results in an operational time frame using humans, PDFs
209
573137
3350
09:36
and tenacity alone.
210
576487
2218
09:38
The researchers had to lever up their human minds
211
578705
1953
09:40
with technology to dive deeper, to explore non-obvious
212
580658
2345
09:43
hypotheses, and in fact, insights emerged.
213
583003
3218
09:46
Twenty percent of the foreign fighters were from Libya,
214
586221
2644
09:48
50 percent of those from a single town in Libya,
215
588865
2968
09:51
hugely important since prior statistics put that figure at
216
591833
2450
09:54
three percent. It also helped to hone in on a figure
217
594283
2383
09:56
of rising importance in al Qaeda, Abu Yahya al-Libi,
218
596666
2977
09:59
a senior cleric in the Libyan Islamic fighting group.
219
599643
2631
10:02
In March of 2007, he gave a speech, after which there was
220
602274
2664
10:04
a surge in participation amongst Libyan foreign fighters.
221
604938
3466
10:08
Perhaps most clever of all, though, and least obvious,
222
608404
3106
10:11
by flipping the data on its head, the researchers were
223
611510
2073
10:13
able to deeply explore the coordination networks in Syria
224
613583
2900
10:16
that were ultimately responsible for receiving and
225
616483
2517
10:19
transporting the foreign fighters to the border.
226
619000
2464
10:21
These were networks of mercenaries, not ideologues,
227
621464
2633
10:24
who were in the coordination business for profit.
228
624097
2398
10:26
For example, they charged Saudi foreign fighters
229
626495
1904
10:28
substantially more than Libyans, money that would have
230
628399
2199
10:30
otherwise gone to al Qaeda.
231
630598
2320
10:32
Perhaps the adversary would disrupt their own network
232
632918
2045
10:34
if they knew they cheating would-be jihadists.
233
634963
3035
10:37
In January, 2010, a devastating 7.0 earthquake struck Haiti,
234
637998
3745
10:41
third deadliest earthquake of all time, left one million people,
235
641743
2916
10:44
10 percent of the population, homeless.
236
644659
2584
10:47
One seemingly small aspect of the overall relief effort
237
647243
3137
10:50
became increasingly important as the delivery of food
238
650380
2176
10:52
and water started rolling.
239
652556
2160
10:54
January and February are the dry months in Haiti,
240
654716
1458
10:56
yet many of the camps had developed standing water.
241
656174
2942
10:59
The only institution with detailed knowledge of Haiti's
242
659116
2122
11:01
floodplains had been leveled
243
661238
1297
11:02
in the earthquake, leadership inside.
244
662535
3008
11:05
So the question is, which camps are at risk,
245
665543
2575
11:08
how many people are in these camps, what's the
246
668118
1921
11:10
timeline for flooding, and given very limited resources
247
670039
2311
11:12
and infrastructure, how do we prioritize the relocation?
248
672350
3384
11:15
The data was incredibly disparate. The U.S. Army had
249
675734
2344
11:18
detailed knowledge for only a small section of the country.
250
678078
2929
11:21
There was data online from a 2006 environmental risk
251
681007
2511
11:23
conference, other geospatial data, none of it integrated.
252
683518
2664
11:26
The human goal here was to identify camps for relocation
253
686182
2958
11:29
based on priority need.
254
689140
2395
11:31
The computer had to integrate a vast amount of geospacial
255
691535
2440
11:33
information, social media data and relief organization
256
693975
2584
11:36
information to answer this question.
257
696559
3480
11:40
By implementing a superior process, what was otherwise
258
700039
2415
11:42
a task for 40 people over three months became
259
702454
2608
11:45
a simple job for three people in 40 hours,
260
705062
3176
11:48
all victories for human-computer symbiosis.
261
708238
2628
11:50
We're more than 50 years into Licklider's vision
262
710866
2054
11:52
for the future, and the data suggests that we should be
263
712920
2242
11:55
quite excited about tackling this century's hardest problems,
264
715162
3030
11:58
man and machine in cooperation together.
265
718192
2947
12:01
Thank you. (Applause)
266
721139
2197
12:03
(Applause)
267
723336
2505
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