How do we learn to work with intelligent machines? | Matt Beane

64,159 views ใƒป 2019-02-21

TED


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

ืชืจื’ื•ื: Talia Breuer ืขืจื™ื›ื”: Ido Dekkers
00:13
Itโ€™s 6:30 in the morning,
0
13292
1875
ื”ืฉืขื” 6:30 ื‘ื‘ื•ืงืจ,
00:15
and Kristen is wheeling her prostate patient into the OR.
1
15583
4875
ื•ื›ืจื™ืกื˜ื™ืŸ ืžืกื™ืขื” ืืช ื”ืžื˜ื•ืคืœ ืฉืœื” ืืฉืจ ื—ื•ืœื” ื‘ืกืจื˜ืŸ ื”ืขืจืžื•ื ื™ืช, ืืœ ื—ื“ืจ ื”ื ื™ืชื•ื—
00:21
She's a resident, a surgeon in training.
2
21500
2250
ื›ืจื™ืกื˜ื™ืŸ ื”ื™ื ืžืชืžื—ื”, ืžื ืชื—ืช ื‘ื”ื›ืฉืจื”.
00:24
Itโ€™s her job to learn.
3
24333
2167
ื”ืขื‘ื•ื“ื” ืฉืœื” ื”ื™ื ืœืœืžื•ื“.
00:27
Today, sheโ€™s really hoping to do some of the nerve-sparing,
4
27292
3351
ื”ื™ื•ื, ื”ื™ื ืžืงื•ื•ื” ืžืื•ื“ ืœื ืœืคื’ื•ืข ื‘ืขืฆื‘ื™ื ืฉืกื‘ื™ื‘ ื”ืจืงืžื•ืช,
00:30
extremely delicate dissection that can preserve erectile function.
5
30667
3875
ืชื”ืœื™ืš ืขื“ื™ืŸ ื‘ื™ื•ืชืจ, ืฉื‘ืืžืฆืขื•ืชื• ื ื™ืชืŸ ืœืฉืžืจ ืืช ืชืคืงื•ื“ ื”ื–ืงืคื”.
00:35
That'll be up to the attending surgeon, though, but he's not there yet.
6
35500
3338
ื”ืคืขื•ืœื” ื”ื–ื• ืชืชื‘ืฆืข ืขืœ ื™ื“ื™ ื”ืžื ืชื— ื”ื‘ื›ื™ืจ, ืืžื ื, ืืš ื”ื•ื ืœื ื”ื’ื™ืข ืขื“ื™ื™ืŸ.
00:39
She and the team put the patient under,
7
39625
2393
ื”ืžืชืžื—ื” ื•ื”ืฆื•ื•ืช ื”ืฉื›ื™ื‘ื• ืืช ื”ืžื˜ื•ืคืœ,
00:42
and she leads the initial eight-inch incision in the lower abdomen.
8
42042
3708
ื•ื”ื™ื ื”ื•ื‘ื™ืœื” ืืช ื”ื—ืชืš ื”ืจืืฉื•ื ื™, ื‘ืื•ืจืš 20 ืกื ื˜ื™ืžื˜ืจื™ื, ื‘ื‘ื˜ืŸ ื”ืชื—ืชื•ื ื”.
00:47
Once sheโ€™s got that clamped back, she tells the nurse to call the attending.
9
47042
3586
ื‘ืจื’ืข ืฉื”ื™ื ืžืกื™ื™ืžืช ืœื”ื“ืง ืื•ืชื•, ื”ื™ื ืžื‘ืงืฉืช ืžื”ืื—ื•ืช ืœืงืจื•ื ืœืžื ืชื— ื”ื‘ื›ื™ืจ.
00:51
He arrives, gowns up,
10
51583
2292
ื”ื‘ื›ื™ืจ ืžื’ื™ืข, ืžืชืœื‘ืฉ,
00:54
And from there on in, their four hands are mostly in that patient --
11
54458
5792
ื•ืžืจื’ืข ื–ื” ื•ืื™ืœืš, ืืจื‘ืขืช ื”ื™ื“ื™ื™ื ืฉืœื”ื ื‘ืขื™ืงืจ ื‘ืชื•ืš ื’ื•ืฃ ื”ื—ื•ืœื” --
01:00
with him guiding but Kristin leading the way.
12
60708
2917
ื›ืืฉืจ ื”ื‘ื›ื™ืจ ืžื“ืจื™ืš ืืš ื›ืจื™ืกื˜ื™ืŸ ืžื•ื‘ื™ืœื” ืืช ื”ื“ืจืš.
01:04
When the prostates out (and, yes, he let Kristen do a little nerve sparing),
13
64875
4643
ื›ืืฉืจ ื”ืขืจืžื•ื ื™ืช ื‘ื—ื•ืฅ (ืื›ืŸ ื›ืŸ, ื”ื•ื ื ืชืŸ ืœื›ืจื™ืกื˜ื™ืŸ ืœืขื‘ื•ื“ ืขืœ ื”ืขืฆื‘ื™ื),
01:09
he rips off his scrubs.
14
69542
1226
ื”ื•ื ืชื•ืœืฉ ืืช ื”ื‘ื’ื“ื™ื ื”ืกื˜ืจื™ืœื™ื™ื ืฉืœื•.
01:10
He starts to do paperwork.
15
70792
1375
ืžืชื—ื™ืœ ืœืขืกื•ืง ื‘ื ื™ื™ืจืช.
01:12
Kristen closes the patient by 8:15,
16
72833
5375
ื›ืจื™ืกื˜ื™ืŸ ืกื•ื’ืจืช ืืช ื”ืžื˜ื•ืคืœ ื‘ืฉืขื” 8:15,
01:18
with a junior resident looking over her shoulder.
17
78583
2435
ื›ืืฉืจ ืžืชืžื—ื” ืฆืขื™ืจ ืžืฆื™ืฅ ืžืขื‘ืจ ืœื›ืชืฃ ืฉืœื”.
01:21
And she lets him do the final line of sutures.
18
81042
3083
ื•ื”ื™ื ื ื•ืชื ืช ืœื• ืœื‘ืฆืข ืืช ืฉื•ืจืช ื”ืชืคืจื™ื ื”ืื—ืจื•ื ื”.
01:24
Kristen feels great.
19
84833
3042
ื›ืจื™ืกื˜ื™ืŸ ืžืจื’ื™ืฉื” ื ื”ื“ืจ.
01:28
Patientโ€™s going to be fine,
20
88250
1559
ื”ืžื˜ื•ืคืœ ื”ื•ืœืš ืœื”ื™ื•ืช ื‘ืกื“ืจ,
01:29
and no doubt sheโ€™s a better surgeon than she was at 6:30.
21
89833
3167
ื•ืื™ืŸ ืกืคืง ืฉื”ื™ื ืžื ืชื—ืช ื˜ื•ื‘ื” ื™ื•ืชืจ ืžืžื” ืฉื”ื™ื™ืชื” ื‘6:30.
01:34
Now this is extreme work.
22
94208
2834
ื–ื•ื”ื™ ืขื‘ื•ื“ื” ืงื™ืฆื•ื ื™ืช.
01:37
But Kristinโ€™s learning to do her job the way that most of us do:
23
97417
3833
ืืš ื›ืจื™ืกื˜ื™ืŸ ืœืžื“ื” ืœืขืฉื•ืช ืืช ื”ืขื‘ื•ื“ื” ืฉืœื” ื‘ืื•ืชื” ื“ืจืš ื‘ื” ืจื•ื‘ื ื• ืœื•ืžื“ื™ื:
01:41
watching an expert for a bit,
24
101625
1893
ืœืฆืคื•ืช ื‘ืžื•ืžื—ื” ืœื–ืžืŸ ืžื”,
01:43
getting involved in easy, safe parts of the work
25
103542
3142
ืœื”ืฉืชืœื‘ ื‘ืžืฉื™ืžื•ืช ืงืœื•ืช ืขื ื”ืฆืœื—ื” ื‘ื˜ื•ื—ื”
01:46
and progressing to riskier and harder tasks
26
106708
2185
ื•ืœื”ืชืงื“ื ืœืžืฉื™ืžื•ืช ืงืฉื•ืช ื‘ืขืœื•ืช ืกื™ื›ื•ืŸ ื’ื‘ื•ื” ื™ื•ืชืจ
01:48
as they guide and decide sheโ€™s ready.
27
108917
2333
ื›ืืฉืจ ืžื“ืจื™ื›ื™ื ืื•ืชื” ื•ืžื—ืœื™ื˜ื™ื ืžืชื™ ื”ื™ื ืžื•ื›ื ื”.
01:52
My whole life Iโ€™ve been fascinated by this kind of learning.
28
112042
2892
ื›ืœ ื—ื™ื™ ื”ื•ืงืกืžืชื™ ืžื”ืกื•ื’ ื”ื–ื” ืฉืœ ืœืžื™ื“ื”.
01:54
It feels elemental, part of what makes us human.
29
114958
3667
ื–ื” ืžืจื’ื™ืฉ ืžื•ื‘ืŸ ืžืืœื™ื•, ื—ืœืง ืžื”ื™ื•ืชื™ื ื• ื‘ื ื™ ืื“ื.
01:59
It has different names: apprenticeship, coaching, mentorship, on the job training.
30
119750
5417
ื•ื™ืฉ ืœื›ืš ืฉืžื•ืช ืฉื•ื ื™ื: ื—ื ื™ื›ื”, ืื™ืžื•ืŸ, ื”ื ื—ื™ื”, ืœืžื™ื“ื” ืžื”ืชื ืกื•ืช.
02:05
In surgery, itโ€™s called โ€œsee one, do one, teach one.โ€
31
125542
3291
ื‘ืขื•ืœื ื”ื ื™ืชื•ื—, ื–ื” ื ืงืจื "ืจืื” ืื—ื“, ืขืฉื” ืื—ื“, ืœืžื“ ืื—ื“."
02:09
But the process is the same,
32
129625
1344
ืืš ื”ืชื”ืœื™ืš ื–ื”ื”,
02:10
and itโ€™s been the main path to skill around the globe for thousands of years.
33
130993
4174
ื•ื–ื• ื”ื™ื™ืชื” ื”ื“ืจืš ื”ืขื™ืงืจื™ืช ืœืœืžื™ื“ื” ืžืกื‘ื™ื‘ ืœืขื•ืœื ื‘ืžืฉืš ืืœืคื™ ืฉื ื™ื.
02:16
Right now, weโ€™re handling AI in a way that blocks that path.
34
136333
4500
ื›ืขืช, ืื ื• ืžืชื ื”ื’ื™ื ื›ืœืคื™ ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ื‘ืฆื•ืจื” ืฉื—ื•ืกืžืช ืืช ื”ื“ืจืš ื”ื–ื•.
02:21
Weโ€™re sacrificing learning in our quest for productivity.
35
141625
2690
ืื ื• ืžืงืจื™ื‘ื™ื ืœืžื™ื“ื” ืขื‘ื•ืจ ื”ื”ืชื™ื™ืขืœื•ืช.
02:25
I found this first in surgery while I was at MIT,
36
145292
2809
ื”ื“ื‘ืจ ื”ืชื‘ื”ืจ ืœื™ ืœืจืืฉื•ื ื” ื‘ืขื•ืœื ื”ื ื™ืชื•ื— ื›ืืฉืจ ื”ื™ื™ืชื™ ื‘ืืž.ืื™ื™.ื˜ื™,
02:28
but now Iโ€™ve got evidence itโ€™s happening all over,
37
148125
2476
ืืš ื›ืขืช ื™ืฉ ืœื™ ืจืื™ื•ืช ืœื›ืš ืฉื–ื” ืงื•ืจื” ื‘ื›ืœ ืขื‘ืจ,
02:30
in very different industries and with very different kinds of AI.
38
150625
3875
ื‘ืชืขืฉื™ื•ืช ืฉื•ื ื•ืช ืœื—ืœื•ื˜ื™ืŸ ื•ืขื ืกื•ื’ื™ื ืฉื•ื ื™ื ืฉืœ ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช.
02:35
If we do nothing, millions of us are going to hit a brick wall
39
155083
5851
ืื ืœื ื ืขืฉื” ื“ื‘ืจ, ืžื™ืœื™ื•ื ื™ื ืžืื™ืชื ื• ื”ื•ืœื›ื™ื ืœื”ืชืงืœ ื‘ืžื‘ื•ื™ ืกืชื•ื
02:40
as we try to learn to deal with AI.
40
160958
2417
ื‘ื ืกื™ื•ืŸ ื”ื”ืชืžื•ื“ื“ื•ืช ืขื ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช.
02:45
Letโ€™s go back to surgery to see how.
41
165125
1772
ื‘ื•ืื• ื ื—ื–ื•ืจ ืœืขื•ืœื ื”ื ื™ืชื•ื— ืœืฉื ื”ื”ื“ื’ืžื”.
02:47
Fast forward six months.
42
167708
1935
ื”ืจืฆื” 6 ื—ื•ื“ืฉื™ื ืงื“ื™ืžื”.
02:49
Itโ€™s 6:30am again, and Kristen is wheeling another prostate patient in,
43
169667
5476
ื”ืฉืขื” ืฉื•ื‘ 6:30, ื•ื›ืจื™ืกื˜ื™ืŸ ืžื›ื ื™ืกื” ืžื˜ื•ืคืœ ื ื•ืกืฃ,
02:55
but this time to the robotic OR.
44
175167
3166
ืืš ื”ืคืขื ืืœ ื—ื“ืจ ื ื™ืชื•ื— ืจื•ื‘ื•ื˜ื™.
02:59
The attending leads attaching
45
179667
1684
ื”ืžื ืชื— ื”ื‘ื›ื™ืจ ืžื—ื‘ืจ
03:01
a four-armed, thousand-pound robot to the patient.
46
181375
2833
ืจื•ื‘ื•ื˜ ื‘ืžืฉืงืœ 1000 ืคืื•ื ื“, ื‘ืขืœ ืืจื‘ืข ื™ื“ื™ื™ื, ืืœ ื”ืžื˜ื•ืคืœ.
03:04
They both rip off their scrubs,
47
184750
2434
ืฉื ื™ื›ื ืžืกื™ืจื™ื ืืช ื”ื‘ื™ื’ื•ื“ ื”ืกื˜ืจื™ืœื™,
03:07
head to control consoles 10 or 15 feet away,
48
187208
3125
ื ื™ื’ืฉื™ื ืืช ืžื›ืฉื™ืจื™ ื”ืฉืœื™ื˜ื” ื‘ืžืจื—ืง 3 ืื• 5 ืžื˜ืจ ืžื”ืžื˜ื•ืคืœ,
03:11
and Kristen just watches.
49
191167
3750
ื•ื›ืจื™ืกื˜ื™ืŸ ืจืง ืฆื•ืคื”.
03:16
The robot allows the attending to do the whole procedure himself,
50
196375
3053
ื”ืจื•ื‘ื•ื˜ ืžืืคืฉืจ ืœืžื ืชื— ื”ื‘ื›ื™ืจ ืœื‘ืฆืข ืืช ื”ื”ืœื™ืš ื›ื•ืœื• ื‘ืขืฆืžื•,
03:19
so he basically does.
51
199452
1583
ื•ืœืžืขืฉื” ื›ืš ื”ื•ื ืคื•ืขืœ.
03:21
He knows she needs practice.
52
201917
2101
ื”ื•ื ื™ื•ื“ืข ืฉื”ื™ื ื–ืงื•ืงื” ืœืื™ืžื•ืŸ.
03:24
He wants to give her control.
53
204042
1583
ื”ื•ื ืจื•ืฆื” ืœื”ืขื‘ื™ืจ ืืœื™ื” ืืช ื”ืฉืœื™ื˜ื”.
03:26
But he also knows sheโ€™d be slower and make more mistakes,
54
206250
3393
ืืš ื”ื•ื ื’ื ื™ื•ื“ืข ืฉื”ื™ื ืชื”ื™ื” ืื™ื˜ื™ืช ื™ื•ืชืจ ื•ืชืขืฉื” ื™ื•ืชืจ ื˜ืขื•ื™ื•ืช,
03:29
and his patient comes first.
55
209667
1500
ื•ื”ืžื˜ื•ืคืœ ื‘ืขื“ื™ืคื•ืช ืขืœื™ื•ื ื”.
03:32
So Kristin has no hope of getting anywhere near those nerves during this rotation.
56
212250
4625
ืœื›ืŸ ืœื›ืจื™ืกื˜ื™ืŸ ืื™ืŸ ืชืงื•ื•ื” ืืคื™ืœื• ืœื”ืชืงืจื‘ ืœืขืฆื‘ื™ื ืฉืœ ื”ืžื˜ื•ืคืœ ื‘ืžื”ืœืš ื”ื ื™ืชื•ื—.
03:37
Sheโ€™ll be lucky if she operates more than 15 minutes during a four-hour procedure.
57
217417
4375
ื”ื™ื ืชื”ื™ื” ื‘ืจืช ืžื–ืœ ืื ืชื•ื›ืœ ืœืชืคืขืœ ืžืขืœ 15 ื“ืงื•ืช ื‘ืชื”ืœื™ืš ืฉืœ 4 ืฉืขื•ืช.
03:42
And she knows that when she slips up,
58
222250
2625
ื•ื”ื™ื ื™ื•ื“ืขืช ืฉืื ืชืขืฉื” ื˜ืขื•ืช,
03:45
heโ€™ll tap a touch screen, and sheโ€™ll be watching again,
59
225458
3042
ื”ื•ื ื™ืœื—ืฅ ืขืœ ื”ืžืกืš, ื•ื”ื™ื ืชื—ื–ื•ืจ ืœื”ื™ื•ืช ืฆื•ืคื”,
03:48
feeling like a kid in the corner with a dunce cap.
60
228917
2625
ืžืจื’ื™ืฉื” ื›ืžื• ื™ืœื“ ืฉื ืฉืœื— ืœืขืžื•ื“ ื‘ืคื™ื ื” ืขื ื›ื•ื‘ืข ื‘ื•ืฉื”.
03:53
Like all the studies of robots and work Iโ€™ve done in the last eight years,
61
233583
3501
ื›ืžื• ื‘ื›ืœ ื”ืžื—ืงืจื™ื ืขืœ ืจื•ื‘ื•ื˜ื™ื ื•ื”ืขื‘ื•ื“ื•ืช ืฉื‘ื™ืฆืขืชื™ ื‘8 ื”ืฉื ื™ื ื”ืื—ืจื•ื ื•ืช,
03:57
I started this one with a big, open question:
62
237108
2118
ื”ืชื—ืœืชื™ ืืช ื”ืžื—ืงืจ ื”ื–ื” ืขื ืฉืืœื” ื’ื“ื•ืœื”, ืคืชื•ื—ื”:
03:59
How do we learn to work with intelligent machines?
63
239250
2792
ื›ื™ืฆื“ ื ืœืžื“ ืœืขื‘ื•ื“ ืขื ืžื›ื•ื ื•ืช ื—ื›ืžื•ืช?
04:02
To find out, I spent two and a half years observing dozens of residents and surgeons
64
242792
5809
ื›ื“ื™ ืœืžืฆื•ื ืืช ื”ืชืฉื•ื‘ื”, ื‘ื™ืœื™ืชื™ ืฉื ืชื™ื™ื ื•ื—ืฆื™ ื‘ืชืฆืคื™ืช ืขืœ ืžืชืžื—ื™ื ื•ืžื ืชื—ื™ื
04:08
doing traditional and robotic surgery, interviewing them
65
248625
3476
ืžื‘ืฆืขื™ื ื ื™ืชื•ื— ืžืกื•ืจืชื™ ื•ื ื™ืชื•ื— ื‘ืืžืฆืขื•ืช ืจื•ื‘ื•ื˜ื™ื ืจืื™ื™ื ืชื™ ืื•ืชื
04:12
and in general hanging out with the residents as they tried to learn.
66
252125
3338
ื•ืคืฉื•ื˜ ื‘ื™ืœื™ืชื™ ื‘ื—ื‘ืจืช ืžืชืžื—ื™ื ื‘ืชื”ืœื™ืš ื”ืœืžื™ื“ื” ืฉืœื”ื.
04:16
I covered 18 of the top US teaching hospitals,
67
256250
3351
ืฉื”ื™ืชื™ ื‘ 18 ืžื‘ืชื™ ื”ื—ื•ืœื™ื ื”ืœื™ืžื•ื“ื™ื™ื ื”ื˜ื•ื‘ื™ื ื‘ื™ื•ืชืจ ื‘ืืจื”"ื‘,
04:19
and the story was the same.
68
259625
1458
ื•ื”ืกื™ืคื•ืจ ื‘ื›ื•ืœื ื”ื™ื” ื–ื”ื”.
04:21
Most residents were in Kristen's shoes.
69
261875
2542
ืจื•ื‘ ื”ืžืชืžื—ื™ื ื”ื™ื• ื‘ื ืขืœื™ื™ื ืฉืœ ื›ืจื™ืกื˜ื™ืŸ.
04:24
They got to โ€œsee oneโ€ plenty,
70
264958
1792
ื™ืฆื ืœื”ื "ืœืจืื•ืช ืื—ื“" ื”ืžื•ืŸ ืคืขืžื™ื,
04:27
but the โ€œdo oneโ€ was barely available.
71
267583
2292
ืืš ื”"ืœืขืฉื•ืช ืื—ื“" ืœื ื”ืชืืคืฉืจ ืœืจื•ื‘.
04:30
So they couldnโ€™t struggle, and they werenโ€™t learning.
72
270333
2528
ื›ืš ืฉืœื ื”ื™ื™ืชื” ืœื”ื ื”ื–ื“ืžื ื•ืช ืœื”ืชืžื•ื“ื“, ื•ื”ื ืœื ืœืžื“ื•.
04:33
This was important news for surgeons, but I needed to know how widespread it was:
73
273291
3810
ื–ื• ื”ื™ื™ืชื” ื™ื“ื™ืขื” ื—ืฉื•ื‘ื” ืขื‘ื•ืจ ื”ืžื ืชื—ื™ื, ืืš ื”ื™ื™ืชื™ ื—ื™ื™ื‘ ืœื”ื‘ื™ืŸ ืžื” ืกื“ืจ ื”ื’ื•ื“ืœ ืฉืœ ื”ื ื•ืฉื:
04:37
Where else was using AI blocking learning on the job?
74
277125
3833
ื‘ืื™ืœื• ืชื—ื•ืžื™ื ื ื•ืกืคื™ื, ื”ืฉื™ืžื•ืฉ ื‘ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ืžื•ื ืข ืœืžื™ื“ื” ืžื”ืชื ืกื•ืช?
04:42
To find out, Iโ€™ve connected with a small but growing group of young researchers
75
282208
4310
ืขืœ ืžื ืช ืœื’ืœื•ืช, ื—ื‘ืจืชื™ ืœืงื‘ื•ืฆื” ืงื˜ื ื” ื•ืžืชืคืชื—ืช ืฉืœ ื—ื•ืงืจื™ื ืฆืขื™ืจื™ื
04:46
whoโ€™ve done boots-on-the-ground studies of work involving AI
76
286542
3434
ืฉื‘ื™ืฆืขื• ืœืžื™ื“ื” ื‘ืฉื˜ื— ืขืœ ืžืงืฆื•ืขื•ืช ืžืฉื•ืœื‘ื™ ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช
04:50
in very diverse settings like start-ups, policing,
77
290000
2976
ื‘ืฉื“ื•ืช ืžื’ื•ื•ื ื™ื ืžืื•ื“ ื›ืžื• ืกื˜ืืจื˜ืืคื™ื, ืฉื™ื˜ื•ืจ,
04:53
investment banking and online education.
78
293000
2601
ื”ืฉืงืขื•ืช ื‘ื ืงืื™ื•ืช ื•ืœื™ืžื•ื“ ืื™ื˜ืจื ื˜ื™.
04:55
Like me, they spent at least a year and many hundreds of hours observing,
79
295625
5851
ื›ืžื•ื ื™, ื”ื ื‘ื™ืœื• ืœืคื—ื•ืช ืฉื ื” ื‘ืฉืขื•ืช ืชืฆืคื•ืช ืžืจื•ื‘ื•ืช,
05:01
interviewing and often working side-by-side with the people they studied.
80
301500
3917
ืจืื™ื™ื ื• ื•ืขื‘ื“ื• ืœืฆื“ ื”ืื ืฉื™ื ืื•ืชื ื—ืงืจื•.
05:06
We shared data, and I looked for patterns.
81
306458
2417
ื—ืœืงื ื• ืžื™ื“ืข, ื•ื—ื™ืคืฉืชื™ ืื—ืจ ื“ืคื•ืกื™ื.
05:09
No matter the industry, the work, the AI, the story was the same.
82
309917
5208
ื‘ืœื™ ืงืฉืจ ืœืชืขืฉื™ื™ื”, ืœืขื‘ื•ื“ื”, ืกื•ื’ ื”ื‘ื™ื ื” ื”ืžืœืื›ื•ืชื™ืช, ื”ืกื™ืคื•ืจ ื”ื™ื” ืื•ืชื• ื”ื“ื‘ืจ.
05:16
Organizations were trying harder and harder to get results from AI,
83
316042
3642
ืืจื’ื•ื ื™ื ื ื™ืกื• ื™ื•ืชืจ ื•ื™ื•ืชืจ ืœืงื‘ืœ ืชื•ืฆืื•ืช ืžื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช,
05:19
and they were peeling learners away from expert work as they did it.
84
319708
3542
ื•ื‘ื›ืš ื”ืจื—ื™ืงื• ืืช ื”ืžืชืœืžื“ื™ื ืžืขื‘ื•ื“ืช ื”ื”ืชืžื—ื•ืช ืฉืœื”ื.
05:24
Start-up managers were outsourcing their customer contact.
85
324333
2875
ืžื ื”ืœื™ ืกื˜ืืจื˜ืืคื™ื ื™ื™ืฆืื• ืืช ื”ืงืฉืจ ืขื ื”ืœืงื•ื—ื•ืช ืฉืœื”ื.
05:27
Cops had to learn to deal with crime forecasts without experts support.
86
327833
4042
ืฉื•ื˜ืจื™ื ื ืืœืฆื• ืœื”ืชืžื•ื“ื“ ืขื ืชื—ื–ื™ื•ืช ืคืฉื™ืขื” ืœืœื ืชืžื™ื›ื” ืฉืœ ืžื•ืžื—ื™ื.
05:32
Junior bankers were getting cut out of complex analysis,
87
332875
3250
ื‘ื ืงืื™ื ืฆืขื™ืจื™ื ื”ื•ืจื—ืงื• ืžื‘ื™ืฆื•ืข ื ื™ืชื•ื— ื ืชื•ื ื™ื ืžืขืžื™ืง,
05:36
and professors had to build online courses without help.
88
336500
3083
ื•ืคืจื•ืคืกื•ืจื™ื ื ืืœืฆื• ืœื‘ื ื•ืช ืงื•ืจืกื™ื ืื™ื ื˜ืจื ื˜ื™ื™ื ืœืœื ื›ืœ ืขื–ืจื”.
05:41
And the effect of all of this was the same as in surgery.
89
341125
3226
ื•ื”ื”ืฉืคืขื” ืฉืœ ื›ืœ ืืœื• ื”ื™ื ื›ืžื• ื‘ืจืคื•ืื”.
05:44
Learning on the job was getting much harder.
90
344375
2917
ืœืžื™ื“ื” ืชื•ืš ื›ื“ื™ ืขื‘ื•ื“ื” ื ืขืฉืชื” ืงืฉื” ื‘ื”ืจื‘ื”.
05:48
This canโ€™t last.
91
348958
1417
ื–ื” ืœื ื™ื•ื›ืœ ืœื”ื—ื–ื™ืง ืžืขืžื“.
05:51
McKinsey estimates that between half a billion and a billion of us
92
351542
4267
ืžืงื™ื ื–ื™ ืžืขืจื™ื›ื™ื ื›ื™ ื‘ื™ืŸ ื—ืฆื™ ืžื™ืœื™ืืจื“ ืœืžื™ืœื™ืืจื“ ืžืื™ืชื ื•
05:55
are going to have to adapt to AI in our daily work by 2030.
93
355833
4125
ื™ื™ืืœืฆื• ืœื”ืชืจื’ืœ ืœืฉื™ืœื•ื‘ ืฉืœ ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ื‘ืขื‘ื•ื“ื” ื”ื™ื•ืžื™ื•ืžื™ืช ืฉืœื ื• ืขื“ ืฉื ืช 2030.
06:01
And weโ€™re assuming that on-the-job learning
94
361000
2011
ื•ืื ื• ืžื ื™ื—ื™ื ื›ื™ ืœืžื™ื“ื” ื‘ื”ืชื ืกื•ืช
06:03
will be there for us as we try.
95
363035
1917
ืชื”ื™ื” ืฉื ืขื‘ื•ืจื ื• ื›ืฉื–ื” ื™ืงืจื”.
06:05
Accentureโ€™s latest workers survey showed that most workers learned key skills
96
365500
4268
ืกืงืจ ื”ืขื•ื‘ื“ื™ื ื”ืื—ืจื•ืŸ ืฉืœ ืืงืกื ืฆ'ื•ืจ ื”ืจืื” ืฉืจื•ื‘ ื”ืขื•ื‘ื“ื™ื ืœืžื“ื• ื›ื™ืฉื•ืจื™ื ื—ืฉื•ื‘ื™ื
06:09
on the job, not in formal training.
97
369792
2291
ืชื•ืš ื›ื“ื™ ืขื‘ื•ื“ื”, ื•ืœื ืžืœืžื™ื“ื” ืคื•ืจืžืœื™ืช.
06:13
So while we talk a lot about its potential future impact,
98
373292
3517
ืื– ืœืžืจื•ืช ืฉืื ื• ื“ื ื™ื ื”ืžื•ืŸ ืขืœ ื”ื”ืฉืคืขื” ื”ืคื•ื˜ื ืฆื™ืืœื™ืช ืฉืœื”,
06:16
the aspect of AI that may matter most right now
99
376833
3685
ื”ื”ื™ื‘ื˜ ืฉืœ ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ื‘ืขืœ ื”ืžืฉืžืขื•ืช ื”ื’ื“ื•ืœื” ื‘ื™ื•ืชืจ ื›ืจื’ืข
06:20
is that weโ€™re handling it in a way that blocks learning on the job
100
380542
3375
ื”ื•ื ืฉืื ื• ืžื ื”ืœื™ื ืื•ืชื” ื‘ืฆื•ืจื” ื›ื–ื• ืฉื”ื™ื ืžื•ื ืขืช ืžืื™ืชื ื• ืœืœืžื•ื“
06:24
just when we need it most.
101
384375
1625
ื‘ื“ื™ื•ืง ื›ืฉืื ื• ื–ืงื•ืงื™ื ืœื›ืš.
06:27
Now across all our sites, a small minority found a way to learn.
102
387458
6042
ื›ืขืช ื‘ื›ืœ ื”ืืชืจื™ื ืฉืœื ื• ืžื™ืขื•ื˜ ืงื˜ืŸ ืžืฆื ื“ืจืš ืœืœืžื•ื“.
06:35
They did it by breaking and bending rules.
103
395625
3042
ื”ื ืขืฉื• ื–ืืช ืขืœ ื™ื“ื™ ืฉื‘ื™ืจื” ื•ื›ื™ืคื•ืฃ ืฉืœ ื”ื—ื•ืงื™ื.
06:39
Approved methods werenโ€™t working, so they bent and broke rules
104
399083
4643
ืฉื™ื˜ื•ืช ืžื•ื›ืจื•ืช ืœื ืขื‘ื“ื•, ื•ืœื›ืŸ ื”ื ื›ื•ืคืคื• ื•ืฉื‘ืจื• ืืช ื”ื—ื•ืงื™ื
06:43
to get hands-on practice with experts.
105
403750
1976
ืขืœ ืžื ืช ืœื”ืฉื™ื’ ืื™ืžื•ืŸ ืžืขืฉื™ ืขื ืžื•ืžื—ื™ื.
06:45
In my setting, residents got involved in robotic surgery in medical school
106
405750
5601
ื‘ืžืงื•ื ื‘ื• ื”ื™ื™ืชื™, ื”ืžืชืžื—ื™ื ื”ืชืขืกืงื• ื‘ื ื™ืชื•ื— ืจื•ื‘ื•ื˜ื™ ื‘ื‘ื™ืช ืกืคืจ ืœืจืคื•ืื”
06:51
at the expense of their generalist education.
107
411375
3583
ืขืœ ื—ืฉื‘ื•ืŸ ื”ื”ืฉื›ืœื” ื”ื›ืœืœื™ืช ืฉืœื”ื.
06:56
And they spent hundreds of extra hours with simulators and recordings of surgery,
108
416417
5851
ื•ื”ื ื‘ื™ืœื• ืžืื•ืช ืฉืœ ืฉืขื•ืช ื ื•ืกืคื•ืช ืขื ืกื™ืžื•ืœื˜ื•ืจื™ื ื•ื”ืงืœื˜ื•ืช ืฉืœ ื ื™ืชื•ื—ื™ื,
07:02
when you were supposed to learn in the OR.
109
422292
2541
ื›ืืฉืจ ื”ื™ื• ืืžื•ืจื™ื ื‘ืขืฆื ืœืœืžื•ื“ ื‘ื—ื“ืจ ื”ื ื™ืชื•ื—.
07:05
And maybe most importantly, they found ways to struggle
110
425375
3476
ื•ืื•ืœื™ ื—ืฉื•ื‘ ืžื›ืš, ื”ื ืžืฆืื• ื“ืจื›ื™ื ืœื”ืชืžื•ื“ื“
07:08
in live procedures with limited expert supervision.
111
428875
3750
ื‘ื”ืœื™ื›ื™ื ืืžื™ืชื™ื™ื ืขื ื”ืฉื’ื—ืช ืžื•ืžื—ื” ืžื•ื’ื‘ืœืช.
07:13
I call all this โ€œshadow learning,โ€ because it bends the rules
112
433792
4309
ืื ื™ ืงื•ืจื ืœื›ืœ ื–ื” ืดืœืžื™ื“ื” ื‘ืฆืœืด, ื›ื™ื•ื•ืŸ ืฉื”ื™ื ืžื›ื•ืคืคืช ืืช ื”ื—ื•ืงื™ื
07:18
and learnerโ€™s do it out of the limelight.
113
438125
2000
ื•ื”ืชืœืžื™ื“ื™ื ืขื•ืฉื™ื ื–ืืช ืžืื—ื•ืจื™ ื”ืงืœืขื™ื.
07:21
And everyone turns a blind eye because it gets results.
114
441542
4101
ื•ื›ื•ืœื ืžืขืœื™ืžื™ื ืขื™ืŸ ื›ื™ื•ื•ืŸ ืฉื–ื” ืžื ื™ื‘ ืชื•ืฆืื•ืช.
07:25
Remember, these are the star pupils of the bunch.
115
445667
3166
ื–ื›ืจื•, ืืœื• ื”ื ื”ืดื›ื•ื›ื‘ื™ืืด ืฉื‘ื—ื‘ื•ืจื”.
07:29
Now, obviously, this is not OK, and itโ€™s not sustainable.
116
449792
3208
ืขื›ืฉื™ื•, ื›ืžื•ื‘ืŸ ืฉื–ื” ืœื ื‘ืกื“ืจ, ื•ื–ื•ื”ื™ ืœื ืฉื™ื˜ื” ืฉืชื—ื–ื™ืง ืœืื•ืจืš ื–ืžืŸ.
07:33
No one should have to risk getting fired
117
453708
2185
ืืฃ ืื—ื“ ืœื ืฆืจื™ืš ืœื”ืกืชื›ืŸ ื‘ืคื™ื˜ื•ืจื™ื
07:35
to learn the skills they need to do their job.
118
455917
2150
ื‘ืฉื‘ื™ืœ ืœืœืžื•ื“ ื›ื™ืฉื•ืจื™ ืขื‘ื•ื“ื”.
07:38
But we do need to learn from these people.
119
458792
2056
ืืš ืื ื• ื›ืŸ ืฆืจื™ื›ื™ื ืœืœืžื•ื“ ืžื”ืื ืฉื™ื ื”ืืœื•.
07:41
They took serious risks to learn.
120
461917
2250
ื”ื ืœืงื—ื• ืกื™ื›ื•ื ื™ื ืื“ื™ืจื™ื ื‘ืฉื‘ื™ืœ ืœืœืžื•ื“.
07:44
They understood they needed to protect struggle and challenge in their work
121
464792
4351
ื”ื ื”ื‘ื™ื ื• ืฉื”ื ืฆืจื™ื›ื™ื ืœื”ื’ืŸ ืขืœ ื”ืžืืžืฅ ื•ื”ืืชื’ืจ ื‘ืขื‘ื•ื“ื”
07:49
so that they could push themselves to tackle hard problems
122
469167
2892
ื›ื“ื™ ืฉื™ื•ื›ืœื• ืœื“ื—ื•ืฃ ืืช ืขืฆืžื ืœื”ืชืžื•ื“ื“ ืขื ื‘ืขื™ื•ืช ืงืฉื•ืช
07:52
right near the edge of their capacity.
123
472083
1959
ืขื“ ืงืฆื” ื’ื‘ื•ืœ ื”ื™ื›ื•ืœืช ืฉืœื”ื.
07:54
They also made sure there was an expert nearby
124
474458
2216
ื‘ื ื•ืกืฃ ื”ื ื•ื™ื“ืื• ืฉื™ืฉ ืžื•ืžื—ื” ื‘ืงืจื‘ืช ืžืงื•ื
07:56
to offer pointers and to backstop against catastrophe.
125
476698
3094
ืฉืฆื™ืข ื ืงื•ื“ื•ืช ืœืฉื™ืคื•ืจ ื•ื‘ืžื™ื“ืช ื”ืฆื•ืจืš ืœืžื ื•ืข ืืกื•ืŸ.
08:00
Letโ€™s build this combination of struggle and expert support
126
480875
3458
ื‘ื•ืื• ื ื‘ื ื” ืืช ื”ืฉื™ืœื•ื‘ ื‘ื™ืŸ ืžืืžืฅ ืœืชืžื™ื›ืช ืžื•ืžื—ื™ื
08:04
into each AI implementation.
127
484708
2750
ืœืชื•ืš ื›ืœ ื”ื˜ืžืขื” ืฉืœ ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช.
08:08
Hereโ€™s one clear example I could get of this on the ground.
128
488375
2828
ืœื”ืœืŸ ื“ื•ื’ืžื” ื‘ืจื•ืจื” ืฉื™ื›ืœืชื™ ืœืžืฆื•ื ืœื›ืš ื‘ืžืฆื™ืื•ืช.
08:12
Before robots,
129
492125
1226
ืœืคื ื™ ื”ืจื•ื‘ื•ื˜ื™ื,
08:13
if you were a bomb disposal technician, you dealt with an IED by walking up to it.
130
493375
4792
ืื ื”ื™ื™ืช ืžื”ื ื“ืก ื ื™ื˜ืจื•ืœ ืคืฆืฆื•ืช, ื”ืชืžื•ื“ื“ืช ืขื ืคืฆืฆื” ืขืœ ื™ื“ื™ ื›ืš ืฉื ื™ื’ืฉืช ืืœื™ื”.
08:19
A junior officer was hundreds of feet away,
131
499333
2143
ืฉื•ื˜ืจ ืฆืขื™ืจ ื”ื™ื” ื‘ืžืจื—ืง ืžืื•ืช ืจื’ืœ ืžืฉื,
08:21
so could only watch and help if you decided it was safe
132
501500
3309
ื›ื“ื™ ืœืชืฆืคืช ื•ืœืขื–ื•ืจ ืื ื”ื™ื™ืช ืžื—ืœื™ื˜ ืฉื”ืื–ื•ืจ ื‘ื˜ื•ื—
08:24
and invited them downrange.
133
504833
1417
ื•ืžื–ืžื™ืŸ ืื•ืชื• ืœื”ืชืงืจื‘.
08:27
Now you sit side-by-side in a bomb-proof truck.
134
507208
3893
ื›ืขืช ื”ื ื™ื•ืฉื‘ื™ื ื–ื” ืœืฆื“ ื–ื” ื‘ืžืฉืื™ืช ืžืฉื•ืจื™ื™ื ืช.
08:31
You both watched the video feed.
135
511125
1809
ืฉื ื™ื›ื ืจืื™ืชื ืืช ื”ืกืจื˜ื•ืŸ.
08:32
They control a distant robot, and you guide the work out loud.
136
512958
4310
ื”ื ืฉืœื˜ื• ื‘ืจื•ื‘ื•ื˜ ืžืจื—ื•ืง, ื•ืืชื” ืžื ื”ืœ ืืช ื”ืขื‘ื•ื“ื” ื‘ืงื•ืœ ืจื.
08:37
Trainees learn better than they did before robots.
137
517292
3208
ืžืชืœืžื“ื™ื ืœื•ืžื“ื™ื ื˜ื•ื‘ ื™ื•ืชืจ ืžืืฉืจ ืฉืœืžื“ื• ื˜ืจื ื”ืจื•ื‘ื•ื˜ื™ื.
08:41
We can scale this to surgery, start-ups, policing,
138
521125
3933
ื ื•ื›ืœ ืœื”ื“ื’ื™ื ื–ืืช ื’ื ืขืœ ื ื™ืชื•ื—, ืกื˜ืืจื˜ืืคื™ื, ืฉื™ื˜ื•ืจ,
08:45
investment banking, online education and beyond.
139
525082
2625
ื”ืฉืงืขื•ืช ื‘ื ืงืื™ื•ืช, ื”ืฉื›ืœื” ืื™ื ื˜ืจื ื˜ื™ืช ื•ืžืขื‘ืจ.
08:48
The good news is weโ€™ve got new tools to do it.
140
528375
2500
ื”ื—ื“ืฉื•ืช ื”ื˜ื•ื‘ื•ืช ื”ืŸ ืฉื™ืฉ ืœื ื• ื›ืœื™ื ืœืขืฉื•ืช ืืช ื–ื”.
08:51
The internet and the cloud mean we donโ€™t always need one expert for every trainee,
141
531750
4082
ื”ืื™ื ื˜ืจื ื˜ ื•ื”ืขื ืŸ ืžืฉืžืขื ืฉืื™ื ื ื• ืฆืจื™ื›ื™ื ืชืžื™ื“ ืžื•ืžื—ื” ืœื›ืœ ืžืชืœืžื“,
08:56
for them to be physically near each other or even to be in the same organization.
142
536167
4458
ืฉื™ื”ื™ื• ืคื™ื–ื™ืช ืงืจื•ื‘ื™ื ืื—ื“ ืœืฉื ื™ ืื• ืืคื™ืœื• ืฉื™ื”ื™ื• ื—ืœืง ืžืื•ืชื• ืืจื’ื•ืŸ.
09:01
And we can build AI to help:
143
541292
3041
ื•ืื ื• ื™ื›ื•ืœื™ื ืœื™ืฆื•ืจ ื‘ื™ื ื” ืžืœืื›ื•ืชื™ืช ืฉืชืขื–ื•ืจ:
09:05
to coach learners as they struggle, to coach experts as they coach
144
545167
5059
ืœื”ื“ืจื™ืš ืชืœืžื™ื“ื™ื ื›ืืฉืจ ื”ื ืžืชืงืฉื™ื, ืœื”ื“ืจื™ืš ืžื•ืžื—ื™ื ื›ืืฉืจ ื”ื ืžื“ืจื™ื›ื™ื
09:10
and to connect those two groups in smart ways.
145
550250
2542
ื•ืœื—ื‘ืจ ื‘ื™ืŸ ืฉืชื™ ื”ืงื‘ื•ืฆื•ืช ื”ืœืœื• ื‘ื“ืจื›ื™ื ื—ื›ืžื•ืช.
09:15
There are people at work on systems like this,
146
555375
2542
ื™ืฉ ืื ืฉื™ื ืฉืขื•ื‘ื“ื™ื ืขืœ ืžืขืจื›ื•ืช ืžืกื•ื’ ื–ื”,
09:18
but theyโ€™ve been mostly focused on formal training.
147
558333
2792
ืืš ื”ื ื”ืชืจื›ื–ื• ื‘ืขื™ืงืจ ื‘ืœื™ืžื•ื“ ืคื•ืจืžืœื™.
09:21
And the deeper crisis is in on-the-job learning.
148
561458
2584
ื•ื”ืžืฉื‘ืจ ื”ืขืžื•ืง ื™ื•ืชืจ ื”ื•ื ืœืžื™ื“ื” ืžื”ืชื ืกื•ืช.
09:24
We must do better.
149
564417
1851
ืื ื• ืžื•ื›ืจื—ื™ื ืœื”ืฉืชืคืจ.
09:26
Todayโ€™s problems demand we do better
150
566292
2583
ื”ื‘ืขื™ื•ืช ืฉืœ ื”ื™ื•ื ื“ื•ืจืฉื•ืช ืฉื ืฉืชืคืจ
09:29
to create work that takes full advantage of AIโ€™s amazing capabilities
151
569375
4875
ืฉื ื™ื™ืฆืจ ืขื‘ื•ื“ื” ืฉืžื ืฆืœืช ืืช ื”ื™ื›ื•ืœื•ืช ื”ืžื“ื”ื™ืžื•ืช ืฉืœ ื”ื‘ื™ื ื” ื”ืžืœืื›ื•ืชื™ืช ื‘ืžืœื•ืืŸ
09:35
while enhancing our skills as we do it.
152
575042
2750
ืชื•ืš ื›ื“ื™ ืฉื™ืคื•ืจ ื”ื›ื™ืฉื•ืจื™ื ืฉืœื ื•.
09:38
Thatโ€™s the kind of future I dreamed of as a kid.
153
578333
2750
ื–ื”ื• ื”ืขืชื™ื“ ืขืœื™ื• ื—ืœืžืชื™ ื›ื™ืœื“.
09:41
And the time to create it is now.
154
581458
2167
ื•ื”ื–ืžืŸ ืœื™ืฆื•ืจ ืื•ืชื• ื”ื•ื ืขื›ืฉื™ื•.
09:44
Thank you.
155
584333
1226
ืชื•ื“ื”.
09:45
(Applause)
156
585583
3625
(ืžื—ื™ืื•ืช ื›ืคื™ื™ื)
ื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