How AI can enhance our memory, work and social lives | Tom Gruber

198,137 views ใƒป 2017-08-21

TED


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

ืชืจื’ื•ื: dana menahem ืขืจื™ื›ื”: zeeva livshitz
ืื ื™ ื›ืืŸ ื›ื“ื™ ืœื”ืฆื™ืข ืœื›ื ื“ืจืš ื—ื“ืฉื” ืœื—ืฉื•ื‘ ืขืœ ื”ืชื—ื•ื ืฉืœื™.
ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช.
ืื ื™ ื—ื•ืฉื‘ ืฉื”ืžื˜ืจื” ืฉืœ ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช
00:12
I'm here to offer you a new way to think about my field,
0
12860
3751
ื”ื™ื ืœื”ืขืฆื™ื ื‘ื ื™ ืื“ื ื‘ืขื–ืจืช ืื™ื ื˜ืœื™ื’ื ืฆื™ืช ืžื›ื•ื ื”.
00:16
artificial intelligence.
1
16635
1768
ื•ื›ื›ืœ ืฉื”ืžื›ื•ื ื•ืช ื ืขืฉื•ืช ื—ื›ืžื•ืช ื™ื•ืชืจ,
00:18
I think the purpose of AI
2
18427
2014
ื›ืš ื’ื ืื ื—ื ื•.
00:20
is to empower humans with machine intelligence.
3
20465
3434
ืื ื™ ืงื•ืจื ืœื–ื” ืดื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ื”ื•ืžื ื™ืกื˜ื™ืชืด
ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ืžืขื•ืฆื‘ืช ืœืขื ื•ืช ืขืœ ืฆืจื›ื™ื ืื ื•ืฉื™ื™ื
00:25
And as machines get smarter,
4
25106
2450
00:27
we get smarter.
5
27580
1253
ืขืœ ื™ื“ื™ ืฉื™ืชื•ืฃ ืคืขื•ืœื” ื•ื”ื’ื“ืœืช ืื ืฉื™ื.
00:29
I call this "humanistic AI" --
6
29727
2512
ืขื›ืฉื™ื•, ื”ื™ื•ื ืื ื™ ืฉืžื— ืœืจืื•ืช
00:32
artificial intelligence designed to meet human needs
7
32263
3671
ืฉื”ืจืขื™ื•ืŸ ืฉืœ ืขื•ื–ืจ ืื™ื ื˜ืœื™ื’ื ื˜ื™
00:35
by collaborating and augmenting people.
8
35958
2931
ื”ื•ื ืžื™ื™ื ืกื˜ืจื™ื.
ื–ื• ื”ืžื˜ืืคื•ืจื” ื”ืžืงื•ื‘ืœืช ืขื‘ื•ืจ ื”ืžืžืฉืง ื‘ื™ืŸ ื‘ื ื™ ืื“ื ื•ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช.
00:39
Now, today I'm happy to see
9
39969
1910
00:41
that the idea of an intelligent assistant
10
41903
2177
ื•ื”ืื—ื“ ืฉืื ื™ ืขื–ืจืชื™ ืœื™ืฆื•ืจ ื ืงืจื "ืกื™ืจื™".
00:44
is mainstream.
11
44104
1469
00:45
It's the well-accepted metaphor for the interface between humans and AI.
12
45597
5060
ืืชื ืžื›ื™ืจื™ื ืืช ืกื™ืจื™.
ืกื™ืจื™ ื”ื™ื ื”ื“ื‘ืจ ืฉื™ื•ื“ืข ืืช ื”ื›ื•ื•ื ื” ืฉืœื›ื
ื•ืขื•ื–ืจ ืœื›ื ืœื‘ืฆืขื” ื‘ืฉื‘ื™ืœื›ื,
00:51
And the one I helped create is called Siri.
13
51376
2697
ืขื•ื–ืจ ืœื›ื ืœื‘ืฆืข ื“ื‘ืจื™ื.
ืื‘ืœ ืžื” ืฉืืชื ืื•ืœื™ ืœื ื™ื•ื“ืขื™ื ื–ื” ืฉืขื™ืฆื‘ื ื• ืืช ืกื™ืจื™
00:54
You know Siri.
14
54777
1152
00:55
Siri is the thing that knows your intent
15
55953
2645
ื‘ืชื•ืจ ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ื”ื•ืžื ื™ืกื˜ื™ืช,
00:58
and helps you do it for you,
16
58622
2042
ืœื”ืขืฆื™ื ืื ืฉื™ื ืขื ืžืžืฉืง ืฉื™ื—ื”
01:00
helps you get things done.
17
60688
1501
01:02
But what you might not know is that we designed Siri
18
62213
2474
ืฉืื™ืคืฉืจ ืœื”ื ืœื”ืฉืชืžืฉ ื‘ืžื—ืฉื•ื‘ ื ื™ื™ื“,
01:05
as humanistic AI,
19
65465
1966
ื‘ืœื™ ืงืฉืจ ืœืžื™ ื”ื ื”ื™ื• ื•ื”ื™ื›ื•ืœื•ืช ืฉืœื”ื.
01:07
to augment people with a conversational interface
20
67455
3171
01:10
that made it possible for them to use mobile computing,
21
70650
3453
ืขื›ืฉื™ื• ืœืจื•ื‘ื ื•,
ื”ื”ืฉืคืขื” ืฉืขืฉืชื” ื”ื˜ื›ื ื•ืœื•ื’ื™ื” ื”ื–ื•
01:14
regardless of who they were and their abilities.
22
74127
3284
ื”ื™ื ืœื”ืคื•ืš ื“ื‘ืจื™ื ืœืงืฆืช ื™ื•ืชืจ ืงืœื™ื ืœืฉื™ืžื•ืฉ.
ืื‘ืœ ื‘ืฉื‘ื™ืœ ื”ื—ื‘ืจ ืฉืœื™ ื“ื ื™ืืœ,
01:18
Now for most of us,
23
78634
1577
01:20
the impact of this technology
24
80235
2781
ื”ื”ืฉืคืขื” ืฉืœ ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ื‘ืžืขืจื›ื•ืช ื”ืืœื• ื”ื™ื ืžืฉื ืช ื—ื™ื™ื.
01:23
is to make things a little bit easier to use.
25
83040
2283
ืืชื ืžื‘ื™ื ื™ื, ื“ื ื™ืืœ ื”ื•ื ื‘ื—ื•ืจ ืžืื•ื“ ืžื™ื•ื—ื“,
01:25
But for my friend Daniel,
26
85736
2179
01:29
the impact of the AI in these systems is a life changer.
27
89172
3171
ื•ื”ื•ื ืขื™ื•ื•ืจ ื•ืžืฉื•ืชืง ื‘ืืจื‘ืข ื’ืคื™ื™ื.
ืžื” ืฉื”ื•ืคืš ืืช ื”ืฉื™ืžื•ืฉ ื‘ืžื›ืฉื™ืจื™ื ืฉื›ื•ืœื ื• ืœื•ืงื—ื™ื ื›ืžื•ื‘ื ื™ื ืžืืœื™ื”ื ืœืงืฉื”.
01:33
You see, Daniel is a really social guy,
28
93037
4833
ื‘ืคืขื ื”ืื—ืจื•ื ื” ืฉื”ื™ื™ืชื™ ื‘ื‘ื™ืช ืฉืœื•, ืื—ื™ื• ืืžืจ,
01:37
and he's blind and quadriplegic,
29
97894
1759
ืดื—ื›ื” ืฉื ื™ื”, ื“ื ื™ืืœ ืœื ืžื•ื›ืŸ.
01:40
which makes it hard to use those devices that we all take for granted.
30
100434
3859
ื”ื•ื ื‘ื˜ืœืคื•ืŸ ืžื“ื‘ืจ ืขื ืื™ืฉื” ืฉื”ื•ื ืคื’ืฉ ื‘ืื™ื ื˜ืจื ื˜.ืด
ืืžืจืชื™, ืดื–ื” ืžื’ื ื™ื‘, ืื™ืš ื”ื•ื ืขืฉื” ืืช ื–ื”?ืด
01:44
The last time I was at his house, his brother said,
31
104317
2442
ื˜ื•ื‘, ื“ื ื™ืืœ ืžืฉืชืžืฉ ื‘ืกื™ืจื™ ืœื ื™ื”ื•ืœ ื”ื—ื™ื™ื ื”ื—ื‘ืจืชื™ื™ื ืฉืœื•-
01:46
"Hang on a second, Daniel's not ready.
32
106783
1858
01:48
He's on the phone with a woman he met online."
33
108665
2342
ื”ืžื™ื™ืœ ืฉืœื•, ื”ื•ื“ืขื•ืช ื•ื˜ืœืคื•ืŸ-
01:52
I'm like, "That's cool, how'd he do it?"
34
112070
2010
ื‘ืœื™ ืœื”ื™ื•ืช ืชืœื•ื™ ื‘ืžื˜ืคืœื™ื ืฉืœื•.
01:54
Well, Daniel uses Siri to manage his own social life --
35
114104
3494
ื–ื” ื“ื™ ืžืขื ื™ื™ืŸ, ื ื›ื•ืŸ?
ื”ืื™ืจื•ื ื™ื” ืคื” ื’ื“ื•ืœื”.
01:58
his email, text and phone --
36
118551
1940
ื”ื ื” ื”ืื™ืฉ ืฉืžืขืจื›ืช ื™ื—ืกื™ื ืขื ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช
02:00
without depending on his caregivers.
37
120515
2476
ืขื•ื–ืจืช ืœื• ืœื ื”ืœ ืžืขืจื›ื•ืช ื™ื—ืกื™ื ืขื ื‘ื ื™ ืื“ื ืืžื™ืชื™ื™ื.
02:04
This is kind of interesting, right?
38
124321
1738
02:06
The irony here is great.
39
126083
1317
ื•ื–ื• ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ื”ื•ืžื ื™ืกื˜ื™ืช.
02:07
Here's the man whose relationship with AI
40
127424
2896
02:10
helps him have relationships with genuine human beings.
41
130344
3177
ืขื•ื“ ื“ื•ื’ืžื” ืขื ืชื•ืฆืื•ืช ืžืฉื ื•ืช ื—ื™ื™ื
ื”ื™ื ืื™ื‘ื—ื•ืŸ ืกืจื˜ืŸ.
02:14
And this is humanistic AI.
42
134625
3130
ื›ืฉืจื•ืคื/ื” ื—ื•ืฉื“/ืช ื‘ืกืจื˜ืŸ,
ื”ื ืœื•ืงื—ื™ื ื“ื’ื™ืžื” ื•ืฉื•ืœื—ื™ื ืื•ืชื” ืœืคืชื•ืœื•ื’/ื™ืช,
02:19
Another example with life-changing consequences
43
139224
2398
02:21
is diagnosing cancer.
44
141646
1257
ืฉืžืกืชื›ืœ/ืช ืขืœื™ื” ื‘ืžื™ืงืจื•ืกืงื•ืค.
ืขื›ืฉื™ื•, ืคืชื•ืœื•ื’ื™ื ื‘ื•ื“ืงื™ื ืืœืคื™ ืฉืงื•ืคื™ื•ืช
02:24
When a doctor suspects cancer,
45
144615
2155
02:26
they take a sample and send it to a pathologist,
46
146794
2703
ืฉืœ ืžื™ืœื™ื•ื ื™ ืชืื™ื ื‘ื›ืœ ื™ื•ื.
02:29
who looks at it under a microscope.
47
149521
1952
ืื– ื‘ืฉื‘ื™ืœ ืœืขื–ื•ืจ ื‘ืžืฉื™ืžื” ื”ื–ื•,
ื›ืžื” ื—ื•ืงืจื™ื ื™ืฆืจื• ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ืžืžื™ื™ื ืช.
02:31
Now, pathologists look at hundreds of slides
48
151965
2525
02:34
and millions of cells every day.
49
154514
2324
ืขื›ืฉื™ื•, ื”ืžืกื•ื•ื’ ืื•ืžืจ, ืดื”ืื ื–ื” ืกืจื˜ืŸ ืื• ืฉื–ื” ืœื ืกืจื˜ืŸ?
02:37
So to support this task,
50
157450
1387
02:38
some researchers made an AI classifier.
51
158861
3442
ื‘ื”ืชื‘ื•ื ื ื•ืช ื‘ืชืžื•ื ื•ืช.
ื”ืžืกื•ื•ื’ ื”ื™ื” ื“ื™ ื˜ื•ื‘,
02:43
Now, the classifier says, "Is this cancer or is this not cancer?"
52
163368
4553
ืื‘ืœ ืœื ื˜ื•ื‘ ื›ืžื• ื”ืื“ื,
02:47
looking at the pictures.
53
167945
1305
ืฉืฆื“ืง ืจื•ื‘ ื”ื–ืžืŸ.
ืื‘ืœ ื›ืฉืžืฉืœื‘ื™ื ืืช ื”ื™ื›ื•ืœื•ืช ืฉืœ ื”ืžื›ื•ื ื” ื•ืฉืœ ื”ืื“ื ื‘ื™ื—ื“,
02:50
The classifier was pretty good,
54
170175
1564
02:52
but not as good as the person,
55
172787
1685
02:55
who got it right most of the time.
56
175765
1768
ื”ื“ื™ื•ืง ืขื•ืœื” ืœ99.5 ืื—ื•ื–ื™ื.
02:58
But when they combine the ability of the machine and the human together,
57
178247
6180
ื”ื•ืกืคื” ืฉืœ ื”ื‘ื™ื ื” ื”ืžืœืื›ื•ืชื™ืช ืœืฉื•ืชืคื•ืช ื—ื™ืกืœื” 85 ืื—ื•ื–ื™ื ืžื”ื˜ืขื•ื™ื•ืช
03:04
accuracy went to 99.5 percent.
58
184451
2995
ืฉืคืชื•ืœื•ื’ื™ื ืื ื•ืฉื™ื™ื ื”ื™ื• ืขื•ืฉื™ื ื‘ืขื‘ื•ื“ื” ืœื‘ื“.
03:08
Adding that AI to a partnership eliminated 85 percent of the errors
59
188970
5001
ื–ื” ื”ืจื‘ื” ืกืจื˜ืŸ ืฉืื—ืจืช ืœื ื”ื™ื” ืžื˜ื•ืคืœ.
03:13
that the human pathologist would have made working alone.
60
193995
3686
ืขื›ืฉื™ื•, ืœืกืงืจื ื™ื, ืžืกืชื‘ืจ
ืฉื”ืื“ื ื”ื™ื” ื˜ื•ื‘ ื™ื•ืชืจ ื‘ืฉืœื™ืœืช ืชื•ืฆืื•ืช ื—ื™ื•ื‘ื™ื•ืช ืฉื’ื•ื™ื•ืช,
03:18
That's a lot of cancer that would have otherwise gone untreated.
61
198640
3159
ื•ื”ืžื›ื•ื ื” ื”ื™ื™ืชื” ื˜ื•ื‘ื” ื™ื•ืชืจ ื‘ื–ื™ื”ื•ื™ ื”ืžืงืจื™ื ืฉืงืฉื” ืœืžืฆื•ื.
03:22
Now, for the curious, it turns out
62
202917
1930
ืื‘ืœ ื”ืฉื™ืขื•ืจ ื›ืืŸ ื”ื•ื ืœื ืขืœ ืื™ื–ื” ืกื•ื›ืŸ ื˜ื•ื‘ ื™ื•ืชืจ
03:24
that the human was better at rejecting false positives,
63
204871
3474
ื‘ืžืฉื™ืžืช ืกื™ื•ื•ื’ ื”ืชืžื•ื ื•ืช ื”ื–ื•.
03:28
and the machine was better at recognizing those hard-to-spot cases.
64
208369
3572
ื”ื“ื‘ืจื™ื ื”ืืœื• ืžืฉืชื ื™ื ืžื™ื“ื™ ื™ื•ื.
03:31
But the lesson here isn't about which agent is better
65
211965
3237
ื”ืฉื™ืขื•ืจ ื›ืืŸ
ื”ื•ื ืฉืฉื™ืœื•ื‘ ื”ื™ื›ื•ืœื•ืช ืฉืœ ื”ืื“ื ื•ื”ืžื›ื•ื ื”,
03:35
at this image-classification task.
66
215226
2078
ื™ืฆืจ ืฉื•ืชืคื•ืช ืขื ื‘ื™ืฆื•ืขื™ื™ื ืขืœ ืื ื•ืฉื™ื™ื.
03:37
Those things are changing every day.
67
217708
1839
03:39
The lesson here
68
219571
1568
03:41
is that by combining the abilities of the human and machine,
69
221163
3360
ื•ื–ื•ื”ื™ ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ื”ื•ืžื ื™ืกื˜ื™ืช.
03:44
it created a partnership that had superhuman performance.
70
224547
4951
ืขื›ืฉื™ื• ื‘ื•ืื• ื ืกืชื›ืœ ืขืœ ื“ื•ื’ืžื” ื ื•ืกืคืช
ืขื ื‘ื™ืฆื•ืขื™ ื˜ืขื™ื ืช ื˜ื•ืจื‘ื•.
03:50
And that is humanistic AI.
71
230363
2940
ื–ื” ืขื™ืฆื•ื‘.
ืขื›ืฉื™ื•, ื‘ื•ืื• ื ื’ื™ื“ ืฉืืช/ื” ืžื”ื ื“ืก/ืช.
03:54
Now let's look at another example
72
234607
1986
ืืช/ื” ืจื•ืฆื” ืœืขืฆื‘ ืžืกื’ืจืช ื—ื“ืฉื” ืœืจื—ืคืŸ.
03:56
with turbocharging performance.
73
236617
2287
ืœื•ืงื—/ืช ืืช ื›ืœื™ ื”ืชื•ื›ื ื”, ืืช ื›ืœื™ ื”CAD ื”ืื”ื•ื‘ื™ื ืขืœื™ื™ืš,
03:58
This is design.
74
238928
1403
ื•ืืช/ื” ืžื–ื™ืŸ/ื ื” ืืช ื”ืฆื•ืจื” ื•ืืช ื”ื—ื•ืžืจื™ื, ื•ืื– ืืช/ื” ืžื ืชื—/ืช ื‘ื™ืฆื•ืขื™ื.
04:00
Now, let's say you're an engineer.
75
240355
2041
04:02
You want to design a new frame for a drone.
76
242420
2117
ื–ื” ื ื•ืชืŸ ืœืš ืขื™ืฆื•ื‘ ืื—ื“.
04:04
You get out your favorite software tools, CAD tools,
77
244561
3064
ืื ืชื™ืชื ื• ืืช ืื•ืชื ื”ื›ืœื™ื ืœื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช,
04:07
and you enter the form and the materials, and then you analyze performance.
78
247649
4215
ื”ื™ื ืชื•ื›ืœ ืœื™ืฆื•ืจ ืืœืคื™ ืขื™ืฆื•ื‘ื™ื.
04:11
That gives you one design.
79
251888
1344
04:13
If you give those same tools to an AI,
80
253982
3677
ื”ืกืจื˜ื•ืŸ ื”ื–ื” ืฉืœ Autodesk ื”ื•ื ืžื“ื”ื™ื.
ื–ื” ืืžื™ืชื™.
04:17
it can generate thousands of designs.
81
257683
3124
ืื– ื–ื” ืžืฉื ื” ืืช ืื™ืš ืฉืื ื—ื ื• ืžืขืฆื‘ื™ื.
ื”ืžื”ื ื“ืก/ืช ื”ืื ื•ืฉื™/ืช ืขื›ืฉื™ื•
04:22
This video by Autodesk is amazing.
82
262013
1934
ืื•ืžืจื™ื ืžื” ื”ืขื™ืฆื•ื‘ ืฉืฆืจื™ืš ืœื”ืฉื™ื’,
04:23
This is real stuff.
83
263971
1300
04:25
So this transforms how we do design.
84
265912
3071
ื•ื”ืžื›ื•ื ื” ืื•ืžืจืช,
ืดื”ื ื” ื”ืืคืฉืจื•ื™ื•ืช.ืด
04:29
The human engineer now
85
269007
1428
ืขื›ืฉื™ื• ื‘ืขื‘ื•ื“ื” ืฉืœื”, ืขื‘ื•ื“ืช ื”ืžื”ื ื“ืกืช
04:30
says what the design should achieve,
86
270459
3315
ื”ื™ื ืœื‘ื—ื•ืจ ืืช ื”ืืคืฉืจื•ืช ืฉืขื•ื ื” ืขืœ ืžื˜ืจื•ืช ื”ืขื™ืฆื•ื‘ ื‘ืฆื•ืจื” ื”ื›ื™ ื˜ื•ื‘ื”.
04:33
and the machine says,
87
273798
1156
04:34
"Here's the possibilities."
88
274978
1370
ืฉื”ื™ื ื‘ืชื•ืจ ื‘ืŸ ืื“ื ื™ื•ื“ืขืช ื˜ื•ื‘ ื™ื•ืชืจ ืžื›ืœ ืื—ื“ ืื—ืจ,
04:37
Now in her job, the engineer's job
89
277578
2940
ืชื•ืš ืฉื™ืžื•ืฉ ื‘ืฉื™ืคื•ื˜ ื•ืžื•ืžื—ื™ื•ืช ืื ื•ืฉื™ื™ื.
04:40
is to pick the one that best meets the goals of the design,
90
280542
3100
ื‘ืžืงืจื” ื”ื–ื”, ื”ืฆื•ืจื” ื”ืžื ืฆื—ืช
ื ืจืื™ืช ืงืฆืช ื›ืžื• ืžืฉื”ื• ืฉื”ื˜ื‘ืข ื”ื™ื” ืžืขืฆื‘,
04:43
which she knows as a human better than anyone else,
91
283666
3339
ืžื™ื ื•ืก ื›ืžื” ืžื™ืœื™ื•ื ื™ ืฉื ื•ืช ืื‘ื•ืœื•ืฆื™ื”
04:47
using human judgment and expertise.
92
287029
1794
ื•ื›ืœ ื”ืคืจื•ื•ื” ื”ืžื™ื•ืชืจืช ื”ื–ื•.
04:49
In this case, the winning form
93
289234
1580
04:50
looks kind of like something nature would have designed,
94
290838
2711
ืขื›ืฉื™ื• ื‘ื•ืื• ื ืจืื” ืœืืŸ ื”ืจืขื™ื•ืŸ ื”ื–ื” ืฉืœ ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ื”ื•ืžื ื™ืกื˜ื™ืช ืื•ืœื™ ื™ื•ื‘ื™ืœ ืื•ืชื ื•
04:53
minus a few million years of evolution
95
293573
2115
04:55
and all that unnecessary fur.
96
295712
2381
ืื ื ืขืงื•ื‘ ืื—ืจื™ ื–ื” ืœืกืคืงื•ืœืฆื™ื” ืฉืžืขื‘ืจ.
ืžื”ื™ ืกื•ื’ ื”ื”ื’ื“ืœื” ืฉื”ื™ื™ื ื• ื›ื•ืœื ื• ืจื•ืฆื™ื ืœืงื‘ืœ?
05:00
Now let's see where this idea of humanistic AI might lead us
97
300244
4672
ื˜ื•ื‘, ืžื” ืœื’ื‘ื™ ืฉื™ืคื•ืจ ืงื•ื’ื ื™ื˜ื™ื‘ื™?
05:04
if we follow it into the speculative beyond.
98
304940
2393
05:07
What's a kind of augmentation that we would all like to have?
99
307938
3004
ื‘ืžืงื•ื ืœืฉืื•ืœ,
ืดื›ืžื” ื—ื›ืžื•ืช ืืคืฉืจ ืœืขืฉื•ืช ืืช ื”ืžื›ื•ื ื•ืช ืฉืœื ื•?ืด
05:11
Well, how about cognitive enhancement?
100
311349
2906
ื‘ื•ืื• ื ืฉืืœ
ืดื›ืžื” ื—ื›ืžื™ื ื”ืžื›ื•ื ื•ืช ืฉืœื ื• ื™ื›ื•ืœื•ืช ืœืขืฉื•ืช ืื•ืชื ื•?ืด
05:15
Instead of asking,
101
315724
1479
ื›ืœื•ืžืจ, ืงื—ื• ืœื“ื•ื’ืžื” ืืช ื”ื–ื™ื›ืจื•ืŸ.
05:17
"How smart can we make our machines?"
102
317227
2061
05:19
let's ask
103
319312
1173
ื–ื™ื›ืจื•ืŸ ื”ื•ื ื”ื‘ืกื™ืก ืฉืœ ื—ื•ื›ืžื” ืื ื•ืฉื™ืช.
05:20
"How smart can our machines make us?"
104
320509
2705
ืื‘ืœ ื”ื–ื™ื›ืจื•ืŸ ื”ืื ื•ืฉื™ ื™ื“ื•ืข ื‘ื”ื™ื•ืชื• ื‘ืขืœ ืคื’ืžื™ื.
05:24
I mean, take memory for example.
105
324925
1780
05:27
Memory is the foundation of human intelligence.
106
327331
2946
ืื ื—ื ื• ืžืขื•ืœื™ื ื‘ืœืกืคืจ ืกื™ืคื•ืจื™ื,
ืื‘ืœ ื’ื ื‘ืœื˜ืขื•ืช ืœื’ื‘ื™ ื”ืคืจื˜ื™ื ืฉืœื”ื.
05:31
But human memory is famously flawed.
107
331587
2971
ื•ื”ื–ื™ื›ืจื•ื ื•ืช ืฉืœื ื•- ื“ื•ืขื›ื™ื ืขื ื”ื–ืžืŸ.
ื›ืœื•ืžืจ, ื›ืื™ืœื•, ืœืื™ืคื” ื”ืœื›ื• ืฉื ื•ืช ื”60, ื•ื”ืื ืื ื™ ื™ื›ื•ืœ ืœืœื›ืช ืœืฉื ื’ื?
05:35
We're great at telling stories,
108
335361
2074
05:37
but not getting the details right.
109
337459
1951
(ืฆื—ื•ืง)
05:39
And our memories -- they decay over time.
110
339434
3206
ืื‘ืœ ืžื” ืื ื™ื›ื•ืœืชื ืฉื™ื”ื™ื” ืœื›ื ื–ื™ื›ืจื•ืŸ ื˜ื•ื‘ ื›ืžื• ื–ื™ื›ืจื•ืŸ ืฉืœ ืžื—ืฉื‘,
05:42
I mean, like, where did the '60s go, and can I go there, too?
111
342664
3232
05:45
(Laughter)
112
345920
1467
ืฉื”ื™ื” ืงืฉื•ืจ ืœื—ื™ื™ื ืฉืœื›ื?
05:48
But what if you could have a memory that was as good as computer memory,
113
348468
4304
ืžื” ืื ื™ื›ื•ืœืชื ืœื–ื›ื•ืจ ื›ืœ ืื“ื ืฉืื™ ืคืขื ืคื’ืฉืชื,
ืื™ืš ืœื‘ื˜ื ืืช ื”ืฉืžื•ืช ืฉืœื”ื,
05:53
and was about your life?
114
353978
2859
ื”ืคืจื˜ื™ื ื”ืžืฉืคื—ืชื™ื™ื ืฉืœื”ื, ื”ืกืคื•ืจื˜ ื”ืื”ื•ื‘ ืขืœื™ื”ื,
ื”ืฉื™ื—ื” ื”ืื—ืจื•ื ื” ืฉื”ื™ื™ืชื” ืœื›ื ืื™ืชื?
05:57
What if you could remember every person you ever met,
115
357534
2789
ืื ื”ื™ื” ืœื›ื ืืช ื”ื–ื™ื›ืจื•ืŸ ื”ื–ื” ื›ืœ ื—ื™ื™ืš,
06:00
how to pronounce their name,
116
360347
1946
ื”ื™ื™ืชื” ืœื›ื ืืช ื ืงื•ื“ืช ื”ืžื‘ื˜ ืฉืœ ื”ื‘ื™ื ื” ื”ืžืœืื›ื•ืชื™ืช ืขืœ ื›ืœ ื”ืื™ื ื˜ืจืงืฆื™ื•ืช
06:02
their family details, their favorite sports,
117
362317
2092
06:04
the last conversation you had with them?
118
364433
1963
ืฉื”ื™ื• ืœื›ื ืขื ืื ืฉื™ื ื‘ืžืฉืš ื”ื–ืžืŸ
ื•ืขื–ืจื” ื‘ืœื”ืจื”ืจ ืขืœ ื”ืงืฉืช ื”ืืจื•ื›ื” ืฉืœ ืžืขืจื›ื•ืช ื”ื™ื—ืกื™ื ืฉืœื›ื.
06:06
If you had this memory all your life,
119
366420
2655
06:09
you could have the AI look at all the interactions
120
369099
3298
ืžื” ืื ื™ื›ื•ืœืชื ืœื‘ืงืฉ ืžื”ื‘ื™ื ื” ื”ืžืœืื›ื•ืชื™ืช ืœืงืจื•ื ืืช ื›ืœ ืžื” ืฉืื™ ืคืขื ืงืจืืชื
06:12
you had with people over time
121
372421
1417
06:13
and help you reflect on the long arc of your relationships.
122
373862
3793
ื•ืœื”ืงืฉื™ื‘ ืœื›ืœ ืฉื™ืจ ืฉืื™ ืคืขื ืฉืžืขืชื?
06:18
What if you could have the AI read everything you've ever read
123
378350
5066
ืžื”ืจืžื– ื”ื›ื™ ืงื˜ืŸ, ื–ื” ื™ื›ื•ืœ ืœืขื–ื•ืจ ืœื›ื ืœื”ื—ื–ื™ืจ
ื›ืœ ื“ื‘ืจ ืฉืื™ ืคืขื ืจืื™ืชื ืื• ืฉืžืขืชื ืœืคื ื™ ื›ืŸ.
06:23
and listen to every song you've ever heard?
124
383440
2514
ื“ืžื™ื™ื ื• ืžื” ื–ื” ื”ื™ื” ื™ื›ื•ืœ ืœืขืฉื•ืช ืœื™ื›ื•ืœืช ืœื™ืฆื•ืจ ืงืฉืจื™ื ื—ื“ืฉื™ื
06:27
From the tiniest clue, it could help you retrieve
125
387164
3739
ื•ืœืคืชื— ืจืขื™ื•ื ื•ืช ื—ื“ืฉื™ื.
06:30
anything you've ever seen or heard before.
126
390927
2113
ื•ืžื” ืœื’ื‘ื™ ื”ื’ื•ืฃ ืฉืœื ื•?
06:33
Imagine what that would do for the ability to make new connections
127
393751
3250
ืžื” ืื ื”ื™ื™ื ื• ื™ื›ื•ืœื™ื ืœื–ื›ื•ืจ ืืช ื”ื”ืฉืœื›ื•ืช ืฉืœ ื›ืœ ืื•ื›ืœ ืฉืื ื—ื ื• ืื•ื›ืœื™ื,
06:37
and form new ideas.
128
397025
1580
ื›ืœ ื’ืœื•ืœื” ืฉืื ื—ื ื• ืœื•ืงื—ื™ื.
06:38
And what about our bodies?
129
398629
1988
ื›ืœ ืœื™ืœื” ืœื‘ืŸ ืฉืขืฉื™ื ื•?
06:41
What if we could remember the consequences of every food we eat,
130
401552
4278
ื”ื™ื™ื ื• ื™ื›ื•ืœื™ื ืœืขืฉื•ืช ืžื“ืข ืžืฉืœื ื• ืขืœ ื”ื ืชื•ื ื™ื ืฉืœ ืขืฆืžื ื•
06:45
every pill we take,
131
405854
1501
ืขืœ ืžื” ื’ื•ืจื ืœื ื• ืœื”ืจื’ื™ืฉ ื˜ื•ื‘ ื•ืœื”ื™ืฉืืจ ื‘ืจื™ืื™ื.
06:48
every all-nighter we pull?
132
408305
1835
ื•ืชืืจื• ืœืขืฆืžื›ื ืื™ืš ื–ื” ื™ื›ื•ืœ ืœื”ื™ื•ืช ืžื”ืคื›ื ื™
06:50
We could do our own science on our own data
133
410868
2710
ื‘ื“ืจืš ืฉื‘ื” ืื ื—ื ื• ืžืชื ื”ืœื™ื ืขื ืืœืจื’ื™ื•ืช ื•ืžื—ืœื•ืช ื›ืจื•ื ื™ื•ืช.
06:53
about what makes us feel good and stay healthy.
134
413602
3860
ืื ื™ ืžืืžื™ืŸ ืฉื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ืชื”ืคื•ืš ืฉื™ืคื•ืจ ื–ื™ื›ืจื•ืŸ ืื™ืฉื™ ืœืžืฆื™ืื•ืช.
06:57
And imagine how this could revolutionize
135
417486
2476
06:59
the way we manage allergies and chronic disease.
136
419986
2690
ืื ื™ ืœื ื™ื›ื•ืœ ืœื”ื’ื™ื“ ืžืชื™ ืื• ืื™ืœื• ื’ื•ืจืžื™ ืฆื•ืจื” ืงืฉื•ืจื™ื,
07:04
I believe that AI will make personal memory enhancement a reality.
137
424729
4974
ืื‘ืœ ืื ื™ ื—ื•ืฉื‘ ืฉื–ื” ื‘ืœืชื™ ื ืžื ืข.
ื‘ื’ืœืœ ืฉืื•ืชื ื”ื“ื‘ืจื™ื ืฉืขื•ืฉื™ื ืืช ื”ื‘ื™ื ื” ื”ืžืœืื›ื•ืชื™ืช ืžื•ืฆืœื—ืช ื›ื™ื•ื-
07:10
I can't say when or what form factors are involved,
138
430942
2824
07:14
but I think it's inevitable,
139
434447
1963
ื”ื–ืžื™ื ื•ืช ืฉืœ ื ืชื•ื ื™ื ืžืงื™ืคื™ื
ื•ื”ื™ื›ื•ืœืช ืฉืœ ืžื›ื•ื ื•ืช ืœื”ื‘ื™ืŸ ืืช ื”ื ืชื•ื ื™ื ื”ืืœื•-
07:17
because the very things that make AI successful today --
140
437743
4827
ื ื™ืชื ื™ื ืœื™ื™ืฉื•ื ืขืœ ื”ื ืชื•ื ื™ื ืฉืœ ื—ื™ื™ื ื•.
07:22
the availability of comprehensive data
141
442594
2493
ื•ื”ื ืชื•ื ื™ื ื”ืืœื• ื ืžืฆืื™ื ื›ืืŸ ื”ื™ื•ื, ื–ืžื™ื ื™ื ืœื›ื•ืœื ื•.
07:25
and the ability for machines to make sense of that data --
142
445111
3548
ื›ื™ ืื ื—ื ื• ืžื ื”ืœื™ื ื—ื™ื™ื ื‘ืชื™ื•ื•ืš ื“ื™ื’ื™ื˜ืœื™,
07:28
can be applied to the data of our lives.
143
448683
3059
07:32
And those data are here today, available for all of us,
144
452325
3487
ื‘ืžื•ื‘ื™ื™ืœ ื•ื‘ืื™ื ื˜ืจื ื˜.
07:35
because we lead digitally mediated lives,
145
455836
3868
ื‘ื”ืกืชื›ืœื•ืช ืฉืœื™, ื–ื™ื›ืจื•ืŸ ืื™ืฉื™, ื”ื•ื ื–ื™ื›ืจื•ืŸ ืคืจื˜ื™.
ืื ื—ื ื• ื™ื›ื•ืœื™ื ืœื‘ื—ื•ืจ ืžื” ื ื–ื›ืจ ื•ื ืฉืžืจ ื•ืžื” ืœื.
07:40
in mobile and online.
146
460390
2135
07:43
In my view, a personal memory is a private memory.
147
463653
3890
ื–ื” ื—ื™ื•ื ื™ ืœื—ืœื•ื˜ื™ืŸ ืฉื–ื” ื™ืฉืžืจ ืžืื•ื“ ื‘ื˜ื•ื—.
07:47
We get to choose what is and is not recalled and retained.
148
467567
3937
ืขื›ืฉื™ื• ื‘ืฉื‘ื™ืœ ืจื•ื‘ื ื•,
ื”ื”ืฉืคืขื” ืฉืœ ื–ื™ื›ืจื•ืŸ ืื™ืฉื™ ืžื•ื’ื‘ืจ
07:52
It's absolutely essential that this be kept very secure.
149
472337
3664
ืชื”ื™ื” ืจื•ื•ื— ืžื ื˜ืœื™ ืžืฉื•ืคืจ ื™ื•ืชืจ,
ืื•ืœื™, ื‘ืชืงื•ื•ื”, ืงืฆืช ื™ื•ืชืจ ื—ืกื“ ื—ื‘ืจืชื™.
07:57
Now for most of us,
150
477114
1462
07:58
the impact of augmented personal memory
151
478600
2893
ืื‘ืœ ืœืžื™ืœื™ื•ื ื™ื ืฉืกื•ื‘ืœื™ื ืžืืœืฆื”ื™ื™ืžืจ ื•ื“ืžื ืฆื™ื”,
08:01
will be a more improved mental gain,
152
481517
2735
08:04
maybe, hopefully, a bit more social grace.
153
484276
2455
ื”ื”ื‘ื“ืœ ืฉื–ื™ื›ืจื•ืŸ ืžื•ื’ื‘ืจ ื™ื•ื›ืœ ืœืขืฉื•ืช
08:08
But for the millions who suffer from Alzheimer's and dementia,
154
488478
4647
ื”ื•ื ื”ื‘ื“ืœ ื‘ื™ืŸ ื—ื™ื™ื ืฉืœ ื‘ื“ื™ื“ื•ืช
ืœื—ื™ื™ื ืฉืœ ื›ื‘ื•ื“ ื•ื—ื™ื‘ื•ืจ.
08:15
the difference that augmented memory could make
155
495029
2203
ืื ื—ื ื• ื›ืจื’ืข ื‘ืขื™ืฆื•ืžื• ืฉืœ ืจื ืกื ืก ื‘ืชื—ื•ื ื”ื‘ื™ื ื” ื”ืžืœืื›ื•ืชื™ืช.
08:17
is a difference between a life of isolation
156
497256
2461
08:20
and a life of dignity and connection.
157
500836
1779
ื›ืœื•ืžืจ, ืจืง ื‘ื›ืžื” ืฉื ื™ื ื”ืื—ืจื•ื ื•ืช,
08:23
We are in the middle of a renaissance in artificial intelligence right now.
158
503986
5647
ืื ื—ื ื• ืžืชื—ื™ืœื™ื ืœืจืื•ืช ืคืชืจื•ื ื•ืช ืœื‘ืขื™ื•ืช ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช
ืื™ืชืŸ ื ืื‘ืงื ื• ืžืžืฉ ื‘ืžืฉืš ืฉื ื™ื:
08:30
I mean, in just the past few years,
159
510472
1971
08:32
we're beginning to see solutions to AI problems
160
512467
3631
ื”ื‘ื ืช ื“ื™ื‘ื•ืจ, ื”ื‘ื ืช ื˜ืงืกื˜,
ื”ื‘ื ืช ืชืžื•ื ื”.
08:36
that we have struggled with literally for decades:
161
516122
3970
ื™ืฉ ืœื ื• ื‘ื—ื™ืจื” ื‘ืื™ืš ืื ื—ื ื• ืžืฉืชืžืฉื™ื ื‘ื˜ื›ื ื•ืœื•ื’ื™ื” ื”ื—ื–ืงื” ื”ื–ื•.
08:41
speech understanding, text understanding,
162
521046
2667
ืื ื—ื ื• ื™ื›ื•ืœื™ื ืœื‘ื—ื•ืจ ืœื”ืฉืชืžืฉ ื‘ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ืœืื•ื˜ื•ืžืฆื™ื” ื•ืœื”ืชื—ืจื•ืช ื‘ื ื•
08:43
image understanding.
163
523737
1320
08:45
We have a choice in how we use this powerful technology.
164
525997
4075
ืื• ืฉืื ื—ื ื• ื™ื›ื•ืœื™ื ืœื”ืฉืชืžืฉ ื‘ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ืœื”ืขืฆืžื” ื•ืฉื™ืชื•ืฃ ืคืขื•ืœื” ืื™ืชื ื•,
08:51
We can choose to use AI to automate and compete with us,
165
531088
4864
ืœื”ืชื’ื‘ืจ ืขืœ ื”ืžื’ื‘ืœื•ืช ื”ืงื•ื’ื ื™ื˜ื™ื‘ื™ื•ืช ืฉืœื ื•
ื•ืœืขื–ื•ืจ ืœื ื• ืœืขืฉื•ืช ืืช ืžื” ืฉืื ื—ื ื• ืจื•ืฆื™ื ืœืขืฉื•ืช,
08:55
or we can use AI to augment and collaborate with us,
166
535976
3977
ืจืง ื˜ื•ื‘ ื™ื•ืชืจ.
08:59
to overcome our cognitive limitations
167
539977
3071
ื•ืชื•ืš ืฉืื ื—ื ื• ืžื’ืœื™ื ื“ืจื›ื™ื ื—ื“ืฉื•ืช ืœืชืช ืœืžื›ื•ื ื•ืช ืื™ื ื˜ืœื™ื’ื ืฆื™ื”,
09:03
and to help us do what we want to do,
168
543072
3162
09:06
only better.
169
546258
1291
ืื ื—ื ื• ื™ื›ื•ืœื™ื ืœื”ืคื™ืฅ ืืช ื”ืื™ื ื˜ืœื™ื’ื ืฆื™ื” ื”ื–ื•
09:08
And as we discover new ways to give machines intelligence,
170
548159
6504
ืœื›ืœ ืขื•ื–ืจื™ ื”ื‘ื™ื ื” ื”ืžืœืื›ื•ืชื™ืช ื‘ืขื•ืœื,
ื•ื‘ื›ืš ืœื›ืœ ืื“ื, ื‘ืœื™ ืงืฉืจ ืœื ืกื™ื‘ื•ืช.
09:14
we can distribute that intelligence
171
554687
3069
ื•ื–ื•ื”ื™ ื”ืกื™ื‘ื”,
09:17
to all of the AI assistants in the world,
172
557780
2381
ืฉื‘ื›ืœ ืคืขื ืฉืžื›ื•ื ื” ื ื”ื™ื™ืช ื—ื›ืžื” ื™ื•ืชืจ,
09:20
and therefore to every person, regardless of circumstance.
173
560185
3534
ืื ื—ื ื• ื ื”ื™ื™ื ื—ื›ืžื™ื ื™ื•ืชืจ.
ื–ื•ื”ื™ ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ืฉืฉื•ื•ื” ืœื”ืคื™ืฅ.
09:25
And that is why,
174
565009
1529
09:26
every time a machine gets smarter,
175
566562
3161
ืชื•ื“ื” ืœื›ื.
(ืžื—ื™ืื•ืช ื›ืคื™ื™ื)
09:30
we get smarter.
176
570448
1404
09:32
That is an AI worth spreading.
177
572848
3733
09:36
Thank you.
178
576605
1182
09:37
(Applause)
179
577811
3973
ืขืœ ืืชืจ ื–ื”

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

https://forms.gle/WvT1wiN1qDtmnspy7