0
00:00:00,010 --> 00:00:05,777
SUB BY : DENI AUROR@
https://aurorarental.blogspot.com/
1
00:00:06,406 --> 00:00:12,045
Steve Austin, astronaut, a man barely alive.
2
00:00:12,078 --> 00:00:15,047
We can rebuild him. We have the technology.
3
00:00:15,081 --> 00:00:18,084
We can make him better than he was.
4
00:00:18,117 --> 00:00:19,552
Better...
5
00:00:19,586 --> 00:00:21,420
stronger...
6
00:00:21,453 --> 00:00:22,821
faster.
7
00:00:22,855 --> 00:00:24,356
Man, when I was a kid,
8
00:00:24,390 --> 00:00:27,060
the Six Million Dollar Man was all the rage.
9
00:00:27,093 --> 00:00:30,763
It's a show about a guy who gets rebuilt with robotic machine parts.
10
00:00:30,797 --> 00:00:32,131
God, I loved it.
11
00:00:32,165 --> 00:00:35,001
And then 35 years later, I got to play a similar role,
12
00:00:35,034 --> 00:00:39,372
a character who enhances himself via technology and engineering.
13
00:00:39,406 --> 00:00:42,474
Now, the term "bionics" goes back to the 1950s,
14
00:00:42,508 --> 00:00:46,312
but the idea of enhancement actually dates back much further,
15
00:00:46,345 --> 00:00:49,682
to Greek mythology, Aztec gods, and even ancient Hinduism.
16
00:00:49,716 --> 00:00:53,119
So these next stories are about augmenting our human abilities,
17
00:00:53,152 --> 00:00:57,457
everything from a bionic limb that behaves naturally and understands intent,
18
00:00:57,490 --> 00:00:59,225
to data that improves performance,
19
00:00:59,258 --> 00:01:00,627
and vision enhancement
20
00:01:00,660 --> 00:01:03,329
that saves people in actual life-threatening situations.
21
00:01:04,130 --> 00:01:06,966
It seems with A.I...
22
00:01:07,000 --> 00:01:09,001
anything's possible.
23
00:01:09,035 --> 00:01:10,469
So it raises the question,
24
00:01:10,503 --> 00:01:12,071
do we even want to be superhuman,
25
00:01:12,104 --> 00:01:15,208
or is imperfection what makes life interesting in the first place?
26
00:01:15,241 --> 00:01:18,177
Whatevs. I gotta get back to the gym.
27
00:01:18,210 --> 00:01:20,747
Normally I'd jog, but I got a wonky knee.
28
00:01:20,780 --> 00:01:23,216
I should probably switch it out.
29
00:01:23,249 --> 00:01:25,718
We can rebuild me.
30
00:01:25,751 --> 00:01:27,053
Better.
31
00:01:27,086 --> 00:01:28,954
Faster.
32
00:01:28,987 --> 00:01:31,057
Stronger.
33
00:01:31,090 --> 00:01:32,925
Seven miles an hour...
34
00:01:32,958 --> 00:01:34,327
full out!
35
00:01:37,330 --> 00:01:41,967
Designers within the field of bionics,
36
00:01:42,001 --> 00:01:45,137
they don't view the human body itself
37
00:01:45,171 --> 00:01:46,840
as designable media.
38
00:01:48,607 --> 00:01:52,177
We now have sophistication in Artificial Intelligence,
39
00:01:52,211 --> 00:01:55,047
in motor technology,
40
00:01:55,081 --> 00:01:57,283
in material science,
41
00:01:57,316 --> 00:01:59,885
in how to talk to the nervous system,
42
00:01:59,918 --> 00:02:01,620
setting the foundation...
43
00:02:03,289 --> 00:02:05,892
...for the end of human disability.
44
00:02:09,061 --> 00:02:11,364
- Do you wanna do an omelette?
- Sure.
45
00:02:11,397 --> 00:02:13,366
You probably don't want too much onion,
46
00:02:13,399 --> 00:02:15,368
because you'll have bad breath all day.
47
00:02:15,401 --> 00:02:17,437
Oh, thanks, yeah.
48
00:02:17,470 --> 00:02:19,338
Anyway, what were we talking about?
49
00:02:19,371 --> 00:02:20,640
26 years, 29 years--
50
00:02:20,673 --> 00:02:24,077
Twenty six years, almost 29 years together, yup.
51
00:02:24,110 --> 00:02:28,481
Our first date was, uh, uh... 2000, wasn't it?
52
00:02:28,514 --> 00:02:31,384
- 1990.
- 1990, oh my God.
53
00:02:31,417 --> 00:02:34,320
I could tell when I first met Jim
54
00:02:34,353 --> 00:02:35,788
that he's highly intelligent,
55
00:02:35,821 --> 00:02:39,024
and he said, "Would you like to go rock climbing sometime?"
56
00:02:39,058 --> 00:02:41,494
and I said, "Sure!"
57
00:02:41,527 --> 00:02:45,297
I started rock climbing in my very early teens,
58
00:02:45,330 --> 00:02:49,068
and I consider climbing to be...
59
00:02:49,101 --> 00:02:50,803
it's more than a passion for me.
60
00:02:50,836 --> 00:02:51,905
It's my lifestyle.
61
00:02:54,106 --> 00:02:56,075
In 2014,
62
00:02:56,109 --> 00:03:00,013
my family and I traveled to the Cayman Islands.
63
00:03:00,046 --> 00:03:01,881
We went rock climbing.
64
00:03:01,914 --> 00:03:05,217
I set up the ropes for the day,
65
00:03:05,250 --> 00:03:09,455
and we'd done a few climbs with no problems.
66
00:03:10,489 --> 00:03:12,759
I started up the final section,
67
00:03:12,792 --> 00:03:16,262
and... shifted my feet and slipped off...
68
00:03:19,198 --> 00:03:21,534
...50 feet to the ground.
69
00:03:24,170 --> 00:03:26,372
When I saw him at the hospital,
70
00:03:26,405 --> 00:03:30,243
I have never felt so helpless in my entire life.
71
00:03:30,910 --> 00:03:32,679
It was horrible.
72
00:03:33,780 --> 00:03:35,948
The front and back of my pelvis
73
00:03:35,982 --> 00:03:37,483
were completely shattered.
74
00:03:38,184 --> 00:03:41,253
My left wrist was shattered.
75
00:03:41,287 --> 00:03:45,792
My left ankle was broken into two or three chunks,
76
00:03:45,825 --> 00:03:49,462
but the rest of it was kind of pulverized.
77
00:03:49,495 --> 00:03:51,931
It slowly started to dawn on me
78
00:03:51,964 --> 00:03:55,401
that this was something that was going to be life-changing.
79
00:03:56,235 --> 00:03:58,003
You're at the hospital.
80
00:03:59,238 --> 00:04:01,206
We're not looking at the photos right now.
81
00:04:01,240 --> 00:04:02,742
I'm looking at 'em.
82
00:04:05,344 --> 00:04:08,481
You look all you want, baby.
83
00:04:09,748 --> 00:04:12,685
After a year, everything else seemed to heal well,
84
00:04:12,718 --> 00:04:15,888
but the ankle continued to be a problem.
85
00:04:15,921 --> 00:04:18,223
The bone was mostly dead,
86
00:04:18,257 --> 00:04:21,560
and the main fracture was still there.
87
00:04:21,593 --> 00:04:23,963
He was in severe pain all the time,
88
00:04:23,996 --> 00:04:26,666
and he just became so depressed.
89
00:04:26,699 --> 00:04:30,270
I couldn't do the things that I love.
90
00:04:30,303 --> 00:04:33,005
I could barely walk down the street without pain,
91
00:04:33,038 --> 00:04:35,408
never mind go rock climbing.
92
00:04:36,508 --> 00:04:39,512
Rock climbing is his passion.
93
00:04:39,545 --> 00:04:42,515
I mean, I just see him withering if he could not climb.
94
00:04:45,685 --> 00:04:48,654
Jim didn't know what to do,
95
00:04:48,687 --> 00:04:51,857
and then, in a truly incredible stroke of luck,
96
00:04:51,891 --> 00:04:54,994
his past came back to help decide his future.
97
00:04:55,027 --> 00:04:58,364
I began mountain climbing at the tender age of seven.
98
00:04:58,398 --> 00:05:01,634
At the age of 17, I was in a mountain-climbing accident,
99
00:05:01,667 --> 00:05:04,403
and I suffered severe frostbite,
100
00:05:04,436 --> 00:05:07,507
and my legs were amputated.
101
00:05:09,575 --> 00:05:11,510
I really dedicated myself
102
00:05:11,544 --> 00:05:14,546
to redesigning first my own legs,
103
00:05:14,580 --> 00:05:18,484
and then the legs of many, many people around the world.
104
00:05:18,518 --> 00:05:21,353
A few decades, a couple of M.I.T. degrees,
105
00:05:21,387 --> 00:05:23,856
and a single-minded focus to innovate later,
106
00:05:23,889 --> 00:05:26,258
Hugh launched the prosthetics industry
107
00:05:26,291 --> 00:05:27,526
into the bionic future.
108
00:05:27,560 --> 00:05:30,296
I'm getting a tremendous amount of energy,
109
00:05:30,329 --> 00:05:31,463
power from the ankles,
110
00:05:31,496 --> 00:05:33,999
which enables me to walk uphill
111
00:05:34,032 --> 00:05:36,002
with a perfectly erect posture.
112
00:05:37,236 --> 00:05:39,505
My legs, they have the brain.
113
00:05:39,538 --> 00:05:41,974
It's a small computer the size of your thumbnail,
114
00:05:42,008 --> 00:05:44,810
and that brain receives sensory information
115
00:05:44,844 --> 00:05:47,413
from sensors on the bionic limb,
116
00:05:47,446 --> 00:05:49,348
and then it runs algorithms
117
00:05:49,382 --> 00:05:52,785
and makes decisions on how to actuate itself.
118
00:05:52,818 --> 00:05:57,657
And machine learning is used as part of those algorithms.
119
00:05:57,690 --> 00:05:59,791
So, machine learning
120
00:05:59,825 --> 00:06:03,262
is what's called a subset field of Artificial Intelligence.
121
00:06:03,295 --> 00:06:05,531
We learn from experience.
122
00:06:05,565 --> 00:06:08,767
Machine learning is basically learning from the experience,
123
00:06:08,801 --> 00:06:10,269
where the experience is the data.
124
00:06:10,302 --> 00:06:12,538
It takes input from the world,
125
00:06:12,571 --> 00:06:14,840
and the input could be text in books,
126
00:06:14,873 --> 00:06:17,843
it can be camera images from a car,
127
00:06:17,876 --> 00:06:21,046
it applies a very complex mathematical function,
128
00:06:21,080 --> 00:06:24,083
and then has an output, which is a decision.
129
00:06:24,116 --> 00:06:26,985
The bionic legs allow Hugh to walk, run,
130
00:06:27,019 --> 00:06:28,153
even climb,
131
00:06:28,187 --> 00:06:29,321
but for him,
132
00:06:29,354 --> 00:06:30,589
there was still something missing.
133
00:06:30,622 --> 00:06:32,658
Because I can't feel my legs,
134
00:06:32,691 --> 00:06:35,895
they... they remain tool-like to me,
135
00:06:35,928 --> 00:06:38,430
and I believe if I could feel my limbs,
136
00:06:38,464 --> 00:06:41,400
they would become part of me, part of self,
137
00:06:41,434 --> 00:06:44,537
and fundamentally change my relationship
138
00:06:44,570 --> 00:06:47,039
to the synthetic part of my body.
139
00:06:48,941 --> 00:06:51,878
I would describe it as just this amazing,
140
00:06:51,911 --> 00:06:53,478
lucky coincidence
141
00:06:53,512 --> 00:06:57,116
that Hugh and I were roommates 34 years ago.
142
00:06:58,117 --> 00:07:01,587
We were teenagers rock-climbing together
143
00:07:01,620 --> 00:07:05,123
and living like dirtbags, living like bums,
144
00:07:05,157 --> 00:07:06,426
climbing every day.
145
00:07:07,627 --> 00:07:09,962
So I decided I was going to look up Hugh
146
00:07:09,995 --> 00:07:13,466
and talk to him about what my options might be.
147
00:07:14,399 --> 00:07:16,602
Jim was in excruciating pain,
148
00:07:16,635 --> 00:07:19,638
and he asked me if I or my colleagues could help him.
149
00:07:19,671 --> 00:07:21,373
What I really was hoping
150
00:07:21,407 --> 00:07:23,108
was that he could put me in touch
151
00:07:23,141 --> 00:07:26,078
with a reconstructive surgeon that could rebuild my ankle.
152
00:07:26,112 --> 00:07:27,746
Right, so this is your X-ray...
153
00:07:27,780 --> 00:07:31,750
Jim was evaluated, and was provided, uh, options
154
00:07:31,783 --> 00:07:34,754
of either maintaining the biological limb
155
00:07:34,787 --> 00:07:36,154
and doing certain procedures
156
00:07:36,188 --> 00:07:40,626
to try to improve its function and to reduce the pain,
157
00:07:40,659 --> 00:07:44,797
or... to amputate the limb.
158
00:07:44,830 --> 00:07:48,600
The thought of amputation was just so big.
159
00:07:48,634 --> 00:07:49,802
What's life gonna be like for me
160
00:07:49,835 --> 00:07:52,572
if I choose to amputate this foot?
161
00:07:54,307 --> 00:07:56,876
But it hurt so bad.
162
00:07:57,777 --> 00:08:02,448
I spoke with Cathy and Maxine about it.
163
00:08:02,481 --> 00:08:04,450
They were behind me 100%.
164
00:08:04,483 --> 00:08:06,819
Whatever I needed to do.
165
00:08:08,154 --> 00:08:10,756
How do you make a decision like that?
166
00:08:11,724 --> 00:08:12,858
A few months later,
167
00:08:12,891 --> 00:08:14,994
he agreed to have his leg amputated
168
00:08:15,027 --> 00:08:16,295
and be the first person
169
00:08:16,328 --> 00:08:19,265
to try his friend's bold, but experimental procedure.
170
00:08:19,298 --> 00:08:20,665
It's gonna be good.
171
00:08:20,699 --> 00:08:23,202
- Yeah.
- Gonna be good.
172
00:08:23,235 --> 00:08:24,236
Love you.
173
00:08:24,269 --> 00:08:25,505
You too.
174
00:08:29,775 --> 00:08:31,376
Bye, honey, love you.
175
00:08:31,410 --> 00:08:32,811
Love you, too.
176
00:08:32,845 --> 00:08:34,980
The way in which limbs are amputated
177
00:08:35,013 --> 00:08:37,650
has not fundamentally changed since the U.S. Civil War...
178
00:08:46,558 --> 00:08:48,060
...but here at M.I.T.,
179
00:08:48,093 --> 00:08:51,664
we were developing a novel way of amputating limbs.
180
00:08:51,697 --> 00:08:54,566
We actually create little biological joints
181
00:08:54,599 --> 00:08:57,302
by linking muscles together in pairs,
182
00:08:57,335 --> 00:08:58,770
so when a person thinks
183
00:08:58,804 --> 00:09:01,373
and moves the limb that's been amputated away,
184
00:09:01,406 --> 00:09:03,642
these muscles move and send sensations
185
00:09:03,675 --> 00:09:06,012
that we can directly link to a bionic limb
186
00:09:06,045 --> 00:09:07,479
in a bi-directional way.
187
00:09:07,512 --> 00:09:09,281
So not only can the person think
188
00:09:09,315 --> 00:09:11,684
and actuate the synthetic limb,
189
00:09:11,717 --> 00:09:14,486
but they can actually feel those synthetic movements
190
00:09:14,519 --> 00:09:15,821
within their own nervous system.
191
00:09:15,854 --> 00:09:17,189
Until recently,
192
00:09:17,223 --> 00:09:20,859
creating a bionic limb that a person can actually feel
193
00:09:20,893 --> 00:09:23,129
has been more science fiction than reality.
194
00:09:23,162 --> 00:09:25,197
Now machine learning
195
00:09:25,231 --> 00:09:28,133
is revolutionizing the way we think about medicine.
196
00:09:28,166 --> 00:09:30,803
If anything can solve the hard problems in medicine,
197
00:09:30,836 --> 00:09:31,670
it's A.I.
198
00:09:31,703 --> 00:09:34,606
Let's take an example, heart disease.
199
00:09:34,640 --> 00:09:36,808
No single human being can have in their head
200
00:09:36,842 --> 00:09:39,478
all the knowledge that it takes to understand heart disease,
201
00:09:39,512 --> 00:09:41,546
but a computer can.
202
00:09:41,580 --> 00:09:43,716
Things like radiology, pathology.
203
00:09:43,749 --> 00:09:45,083
You have an X-ray,
204
00:09:45,117 --> 00:09:46,852
and you wanna see, like, is there cancer in this lung,
205
00:09:46,885 --> 00:09:49,388
and can you pinpoint where the tumor is or not?
206
00:09:49,421 --> 00:09:50,689
A.I.s can, actually, at this point,
207
00:09:50,722 --> 00:09:52,624
do this better than highly trained humans.
208
00:09:52,658 --> 00:09:56,095
Can we get the, uh, Esmarch, please?
209
00:10:00,099 --> 00:10:02,568
It's so light!
210
00:10:02,601 --> 00:10:04,236
It's weird.
211
00:10:05,370 --> 00:10:07,139
It really looks good.
212
00:10:07,173 --> 00:10:08,106
I'm super happy with this,
213
00:10:08,139 --> 00:10:10,743
and you've got a nice degree of padding here.
214
00:10:10,776 --> 00:10:12,744
Right after the surgery,
215
00:10:12,778 --> 00:10:15,647
the incredible, deep, in-the-bone pain
216
00:10:15,681 --> 00:10:17,850
that I had been experiencing for the past year
217
00:10:17,883 --> 00:10:18,717
was gone.
218
00:10:18,750 --> 00:10:21,353
So to me, that was...
219
00:10:21,387 --> 00:10:24,123
that was a success right there.
220
00:10:25,557 --> 00:10:27,326
Never mind whether or not
221
00:10:27,359 --> 00:10:30,996
the experimental part of the amputation was a success,
222
00:10:31,030 --> 00:10:35,334
I was glad to be free of the painful ankle.
223
00:10:35,367 --> 00:10:37,570
He was happy. It was done.
224
00:10:37,603 --> 00:10:39,071
He was ready to move on.
225
00:10:43,042 --> 00:10:44,643
Hey, guys.
226
00:10:44,677 --> 00:10:46,912
How's it looking? How's progress?
227
00:10:46,945 --> 00:10:48,013
Things look good.
228
00:10:48,047 --> 00:10:49,814
I'm gonna go ahead and get you wired up.
229
00:10:49,848 --> 00:10:51,583
The first time I went to Hugh's lab,
230
00:10:51,617 --> 00:10:53,719
Hugh started talking about
231
00:10:53,752 --> 00:10:56,988
"What do you think about a climbing robot ankle?
232
00:10:57,022 --> 00:10:58,391
Would you want us to make one of those?"
233
00:10:58,424 --> 00:11:01,060
I'm gonna go ahead and get us started here...
234
00:11:01,093 --> 00:11:02,928
It's a specifically designed limb
235
00:11:02,961 --> 00:11:05,364
that Jim can control with his mind,
236
00:11:05,397 --> 00:11:08,466
and actually feel the movements within his nervous system.
237
00:11:08,500 --> 00:11:10,202
I still need the evert.
238
00:11:10,235 --> 00:11:12,204
Yup, that's the one we're missing.
239
00:11:12,237 --> 00:11:15,574
All right, so we have you wired to the leg now.
240
00:11:15,607 --> 00:11:16,976
You're driving.
241
00:11:31,357 --> 00:11:32,324
Cool.
242
00:11:32,358 --> 00:11:34,226
Yeah. Can you give me a fast up-down?
243
00:11:37,429 --> 00:11:39,365
And slow, controlled?
244
00:11:41,166 --> 00:11:44,036
This freakin' blows me away every time you do it.
245
00:11:44,070 --> 00:11:45,304
It's so good.
246
00:11:47,205 --> 00:11:51,543
When we link Jim's nerves in that bi-directional way,
247
00:11:51,576 --> 00:11:53,412
we're able to create natural dynamics,
248
00:11:53,445 --> 00:11:56,949
so even though the limb is made of synthetic materials,
249
00:11:56,982 --> 00:11:59,551
it moves as if it's made of flesh and bone.
250
00:11:59,584 --> 00:12:01,019
There.
251
00:12:01,053 --> 00:12:04,390
Now it's neurally and mechanically connected.
252
00:12:04,423 --> 00:12:05,891
How does it feel different?
253
00:12:05,925 --> 00:12:09,528
Now it feels like it's my natural foot, somewhat.
254
00:12:09,561 --> 00:12:11,296
Like, I don't have the skin sensation,
255
00:12:11,330 --> 00:12:15,301
but all the motions make sense to my brain.
256
00:12:15,334 --> 00:12:16,502
In the algorithm,
257
00:12:16,535 --> 00:12:20,705
we make a virtual model of his missing biological limb,
258
00:12:20,739 --> 00:12:23,041
so when he fires his muscles with his brain,
259
00:12:23,075 --> 00:12:25,677
we use an electrode to measure that signal,
260
00:12:25,711 --> 00:12:28,113
and then that drives the virtual muscle
261
00:12:28,147 --> 00:12:30,415
and sends sensations to the brain
262
00:12:30,448 --> 00:12:32,585
about the position and dynamics.
263
00:12:32,618 --> 00:12:34,920
It kind of instantly felt part of me,
264
00:12:34,954 --> 00:12:37,957
almost as good as having a natural foot.
265
00:12:42,928 --> 00:12:45,197
Cathedral Ledge, New Hampshire.
266
00:12:45,230 --> 00:12:48,467
Seven hundred feet of awe-inspiring granite
267
00:12:48,500 --> 00:12:50,802
and climbing routes with names like "Thin Air,"
268
00:12:50,835 --> 00:12:53,739
"Nutcracker," and "They Died Laughing"
269
00:12:53,772 --> 00:12:55,240
make it one hell of a challenge
270
00:12:55,273 --> 00:12:57,242
for even the most serious climbers
271
00:12:57,276 --> 00:12:58,310
on a good day.
272
00:12:59,645 --> 00:13:01,413
For Jim, it's a test
273
00:13:01,447 --> 00:13:05,183
to see if what works at M.I.T. can work out on the mountain.
274
00:13:05,217 --> 00:13:08,520
I've been climbing here for 40 years,
275
00:13:08,553 --> 00:13:11,657
and I've probably spent more time on Cathedral Ledge
276
00:13:11,690 --> 00:13:14,827
than any place else on the planet.
277
00:13:17,863 --> 00:13:20,032
This climb is actually at the upper limit
278
00:13:20,065 --> 00:13:22,334
of my ability at the moment.
279
00:13:22,367 --> 00:13:23,869
I'm not worried at all.
280
00:13:24,803 --> 00:13:27,606
What could possibly go wrong?
281
00:13:27,639 --> 00:13:29,874
The mountain has no mercy,
282
00:13:29,908 --> 00:13:31,910
and no margin for error,
283
00:13:31,943 --> 00:13:33,178
and Jim's about to find out
284
00:13:33,212 --> 00:13:35,847
if his bionic leg can help him overcome
285
00:13:35,881 --> 00:13:37,215
and scale heights
286
00:13:37,249 --> 00:13:39,618
most people wouldn't dare try in the first place.
287
00:13:39,651 --> 00:13:41,019
That's me.
288
00:13:41,053 --> 00:13:43,856
Can machine learning take us even further?
289
00:13:44,956 --> 00:13:46,825
Replace not just what was lost,
290
00:13:46,859 --> 00:13:48,594
but enhance what we already have?
291
00:13:48,627 --> 00:13:50,896
Okay, stay close. I'll lead.
292
00:13:50,929 --> 00:13:52,598
This is insane!
293
00:13:53,531 --> 00:13:54,633
Augment performance
294
00:13:54,666 --> 00:13:57,168
beyond the limits of our natural human ability?
295
00:13:57,202 --> 00:14:00,639
Make strong, smart, fast people...
296
00:14:00,672 --> 00:14:02,607
stronger, smarter, and faster?
297
00:14:04,342 --> 00:14:06,378
In many ways, sports has been on the leading edge of prediction systems,
298
00:14:06,411 --> 00:14:08,881
and now, every serious sports contender
299
00:14:08,914 --> 00:14:11,217
uses sports analytics...
300
00:14:12,117 --> 00:14:14,386
but the big opportunity going forward
301
00:14:14,420 --> 00:14:16,154
is embedding devices
302
00:14:16,188 --> 00:14:18,290
that can collect real-time data
303
00:14:18,324 --> 00:14:19,591
to update strategies
304
00:14:19,625 --> 00:14:21,393
to take advantage of that learning.
305
00:14:21,426 --> 00:14:23,895
There's a revolution going on in sports,
306
00:14:23,928 --> 00:14:25,531
and machine learning is at the core of it.
307
00:14:32,071 --> 00:14:33,472
The first night race of the season,
308
00:14:33,505 --> 00:14:36,174
I'm sure you're ready to finally get behind the wheel.
309
00:14:36,207 --> 00:14:38,109
For sure. Just fired up to get going.
310
00:14:38,143 --> 00:14:40,379
Triple-A car's been pretty good this weekend,
311
00:14:40,412 --> 00:14:42,747
and we're pumped to get this thing started.
312
00:14:42,780 --> 00:14:45,751
Drivers, start your engines!
313
00:14:47,753 --> 00:14:51,122
The race track's a fairly hostile environment.
314
00:14:51,156 --> 00:14:53,491
The way I describe racing, and the way I live it,
315
00:14:53,525 --> 00:14:54,926
it's like war.
316
00:14:54,960 --> 00:14:57,195
Folks, get on your feet. Let's send these guys off!
317
00:14:57,229 --> 00:14:59,899
Boogity boogity boogity! Let's go racing, boys!
318
00:15:01,166 --> 00:15:03,602
You're trying to take your race car,
319
00:15:03,635 --> 00:15:05,037
your team, your driver,
320
00:15:05,070 --> 00:15:07,139
and beat the other drivers at all costs.
321
00:15:08,607 --> 00:15:09,874
Austin Dillon,
322
00:15:09,908 --> 00:15:13,879
stuck in the middle of a three-wide.
323
00:15:13,912 --> 00:15:15,581
This kind of race
324
00:15:15,614 --> 00:15:18,283
produces a lot of strategy,
325
00:15:18,317 --> 00:15:20,786
and that's where we have to use all of our tools
326
00:15:20,819 --> 00:15:23,589
to help us make those strategic decisions.
327
00:15:23,622 --> 00:15:26,024
When it comes to superhuman ability,
328
00:15:26,058 --> 00:15:28,160
you may think of people like LeBron James,
329
00:15:28,193 --> 00:15:30,963
Michael Phelps, or Serena Williams...
330
00:15:32,097 --> 00:15:35,266
but it's not just the body that can be enhanced.
331
00:15:35,300 --> 00:15:38,870
Sometimes it's something less tangible,
332
00:15:38,903 --> 00:15:41,940
like human intuition.
333
00:15:41,973 --> 00:15:43,976
What a battle going on here.
334
00:15:44,009 --> 00:15:46,044
You gotta be real careful here in the early stages
335
00:15:46,078 --> 00:15:47,379
making contact with somebody.
336
00:15:47,412 --> 00:15:49,915
Information is the next battleground.
337
00:15:52,484 --> 00:15:55,587
Every decision you make can have a big impact.
338
00:15:55,620 --> 00:15:56,622
Back in the day,
339
00:15:56,655 --> 00:15:59,357
intuition used to play a big part in sports.
340
00:15:59,391 --> 00:16:01,626
Athletes and coaches relied on their gut
341
00:16:01,660 --> 00:16:03,195
to make decisions.
342
00:16:03,228 --> 00:16:06,064
Now some competitors are leaning more and more
343
00:16:06,097 --> 00:16:07,399
on machine learning,
344
00:16:07,433 --> 00:16:10,035
looking to gain whatever extra edge they can.
345
00:16:10,835 --> 00:16:12,470
We use the A.I. tools
346
00:16:12,504 --> 00:16:15,641
to predict what the future not only is,
347
00:16:15,674 --> 00:16:17,109
but what it should be.
348
00:16:17,142 --> 00:16:19,244
We'll go behind the 20. You just start finish line...
349
00:16:19,278 --> 00:16:20,845
The strength of these A.I. systems
350
00:16:20,879 --> 00:16:23,448
come in having access to a ton of data
351
00:16:23,481 --> 00:16:25,951
and being able to find patterns in that data,
352
00:16:25,984 --> 00:16:28,186
generating insights and inferences
353
00:16:28,219 --> 00:16:30,989
that maybe people may not be aware of,
354
00:16:31,022 --> 00:16:33,125
and then augmenting people's abilities
355
00:16:33,158 --> 00:16:35,694
to make decisions based on that data.
356
00:16:35,727 --> 00:16:36,996
Machine learning
357
00:16:37,029 --> 00:16:40,032
is transforming many industries and applications,
358
00:16:40,065 --> 00:16:42,500
especially in areas where there's a lot of data,
359
00:16:42,534 --> 00:16:45,637
and predicting outcomes can have a big payoff.
360
00:16:45,671 --> 00:16:46,805
Finance,
361
00:16:46,838 --> 00:16:48,840
sports,
362
00:16:48,873 --> 00:16:50,442
or medicine come to mind.
363
00:16:51,643 --> 00:16:54,713
Using an emerging technology like machine learning
364
00:16:54,746 --> 00:16:57,383
in a classic old-school sport like stock car racing
365
00:16:57,416 --> 00:17:00,585
doesn't necessarily sit well with everybody...
366
00:17:00,618 --> 00:17:04,056
which may or may not explain why this guy's doing it...
367
00:17:06,157 --> 00:17:08,693
in a nerve center 250 miles away.
368
00:17:08,726 --> 00:17:11,663
Clear, clear, hit the marks, drive off, man.
369
00:17:12,831 --> 00:17:15,567
My role there really is looking at the data.
370
00:17:15,600 --> 00:17:18,103
How do you use data you can acquire at the racetrack
371
00:17:18,136 --> 00:17:19,237
to get these machines
372
00:17:19,271 --> 00:17:20,572
to be right on the limit of performance?
373
00:17:20,605 --> 00:17:23,341
His front rotors are really glowing.
374
00:17:23,375 --> 00:17:26,478
We get the braking, steering, throttle,
375
00:17:26,511 --> 00:17:29,448
all the acceleration off of every car in the field,
376
00:17:29,481 --> 00:17:30,549
real time...
377
00:17:30,582 --> 00:17:31,784
All this data
378
00:17:31,817 --> 00:17:35,086
is being fed into an A.I. program called "Pit Rho."
379
00:17:37,055 --> 00:17:38,490
Sensors in every car
380
00:17:38,524 --> 00:17:42,227
measure speed, throttle, braking, and steering.
381
00:17:42,260 --> 00:17:45,596
Advanced GPS tracks the car's position on the track.
382
00:17:45,630 --> 00:17:47,366
Watch your middle, watch your middle.
383
00:17:47,399 --> 00:17:51,236
All this data is made available to every team.
384
00:17:51,269 --> 00:17:54,540
This is where the power of A.I. comes in.
385
00:17:54,573 --> 00:17:55,507
So, our tool basically
386
00:17:55,540 --> 00:17:59,077
is analyzing the optimum strategy call
387
00:17:59,110 --> 00:18:01,279
of every car in the field, real time.
388
00:18:01,312 --> 00:18:04,315
Not just our car, but every car.
389
00:18:04,349 --> 00:18:06,618
It's almost threatening.
390
00:18:06,651 --> 00:18:08,487
I was a crew chief for Dale Earnhardt Sr.
391
00:18:08,520 --> 00:18:09,921
Comin' to ya.
392
00:18:09,954 --> 00:18:11,089
10-4.
393
00:18:11,123 --> 00:18:12,457
I would sit up on the box
394
00:18:12,491 --> 00:18:15,494
and intuitively kinda figure all these things.
395
00:18:15,527 --> 00:18:17,996
You kinda just make that gut call,
396
00:18:18,030 --> 00:18:19,364
"Bring him in now."
397
00:18:19,397 --> 00:18:21,232
Until now, many key decisions,
398
00:18:21,266 --> 00:18:23,502
like when to pit for tires or fuel,
399
00:18:23,535 --> 00:18:25,370
were made by the drivers and the crew chief
400
00:18:25,404 --> 00:18:27,839
using experience and intuition.
401
00:18:27,873 --> 00:18:29,441
Now, we've got artificial intelligence
402
00:18:29,474 --> 00:18:31,042
that's making all these calculations
403
00:18:31,076 --> 00:18:32,344
in real time.
404
00:18:32,377 --> 00:18:33,578
Some of the crew chiefs
405
00:18:33,611 --> 00:18:35,347
that have done what I've done over the years,
406
00:18:35,380 --> 00:18:37,916
sometimes it's hard for us to embrace it.
407
00:18:40,352 --> 00:18:43,020
They're trying to get through traffic as fast as they can
408
00:18:43,054 --> 00:18:44,122
so they don't get a lap down,
409
00:18:44,156 --> 00:18:46,624
but that's gonna use up those tires.
410
00:18:46,657 --> 00:18:48,793
You can go at this track on fuel
411
00:18:48,826 --> 00:18:52,230
probably 120 laps, but your tires will be shot way before then.
412
00:18:52,263 --> 00:18:54,066
In a NASCAR race,
413
00:18:54,099 --> 00:18:56,501
pit stops are the key to a winning strategy.
414
00:18:56,535 --> 00:18:57,769
You're trying to decide
415
00:18:57,803 --> 00:19:00,038
when in that cycle is the best to make that stop,
416
00:19:00,071 --> 00:19:03,308
because you lose a lot of time when you come off the track and you have to stop,
417
00:19:03,341 --> 00:19:06,010
but then you gain a lot of speed when you put new tires on.
418
00:19:06,044 --> 00:19:07,512
This is the first time
419
00:19:07,545 --> 00:19:09,615
we're facing, like, a strategy call here.
420
00:19:09,648 --> 00:19:11,616
The Pit Rho A.I. interface
421
00:19:11,649 --> 00:19:13,652
displays one of four suggestions...
422
00:19:15,053 --> 00:19:16,688
stay out on the track,
423
00:19:16,722 --> 00:19:18,156
pit for fuel only,
424
00:19:18,190 --> 00:19:19,658
pit for two tires,
425
00:19:19,691 --> 00:19:21,826
or pit for four tires.
426
00:19:21,860 --> 00:19:22,961
So right now,
427
00:19:22,994 --> 00:19:24,362
it's telling him to take four tires.
428
00:19:24,396 --> 00:19:26,098
Eric relays the message
429
00:19:26,131 --> 00:19:28,400
to the Childress team at the track.
430
00:19:28,433 --> 00:19:30,401
The final decision on when to pit
431
00:19:30,435 --> 00:19:32,871
will be up to the crew chief.
432
00:19:34,105 --> 00:19:35,774
When the pits are open,
433
00:19:35,807 --> 00:19:38,309
it'll be four tires here, four tires.
434
00:19:38,343 --> 00:19:40,078
For the first pit stop,
435
00:19:40,111 --> 00:19:42,247
the crew chief follows the A.I.'s advice.
436
00:19:42,280 --> 00:19:45,918
Five, four, three, two, one.
437
00:19:45,951 --> 00:19:47,853
Put on the brakes, wheels lift.
438
00:19:47,886 --> 00:19:50,756
The crew has to change all four tires
439
00:19:50,789 --> 00:19:52,791
in as little time as possible.
440
00:19:54,359 --> 00:19:57,963
This usually takes between 12 and 14 seconds.
441
00:20:01,800 --> 00:20:04,036
All the way, all the way!
442
00:20:04,069 --> 00:20:05,069
That's a good stop.
443
00:20:05,103 --> 00:20:06,604
Really good stop.
444
00:20:06,637 --> 00:20:08,973
Sometimes, what happens is, over the course of a race,
445
00:20:09,007 --> 00:20:10,742
those little bit better decisions
446
00:20:10,775 --> 00:20:13,478
puts you in a spot, and it puts you in an opportunity
447
00:20:13,511 --> 00:20:16,682
at the end of the race to be able to win the race.
448
00:20:22,754 --> 00:20:25,289
Every lap, it's analyzing the field,
449
00:20:25,323 --> 00:20:26,892
updating its models.
450
00:20:26,925 --> 00:20:30,528
As the race goes on, the prediction gets more and more accurate.
451
00:20:30,562 --> 00:20:32,998
They're using an A.I. technique
452
00:20:33,031 --> 00:20:34,900
called "reinforcement learning,"
453
00:20:34,933 --> 00:20:36,635
which is, basically,
454
00:20:36,668 --> 00:20:39,371
when the computer is given the rules of the game,
455
00:20:39,404 --> 00:20:40,772
plays it over and over
456
00:20:40,805 --> 00:20:43,341
till it learns every possible move and outcome,
457
00:20:43,374 --> 00:20:44,943
and then through trial and error,
458
00:20:44,976 --> 00:20:47,779
and patience that no human could possibly have...
459
00:20:47,812 --> 00:20:50,448
I wonder if we have a resignation here.
460
00:20:50,481 --> 00:20:53,752
...becomes amazing.
461
00:20:53,785 --> 00:20:56,087
Congratulations to AlphaGo
462
00:20:56,120 --> 00:20:57,622
and to the entire team.
463
00:20:57,655 --> 00:20:59,491
It's what Google's DeepMind did
464
00:20:59,524 --> 00:21:01,926
to become a world champ at Go.
465
00:21:01,960 --> 00:21:04,662
Here we go!
466
00:21:04,696 --> 00:21:06,598
It's what Open A.I. did
467
00:21:06,631 --> 00:21:09,100
to conquer the video game Dota 2...
468
00:21:09,134 --> 00:21:10,334
He's dominating.
469
00:21:10,368 --> 00:21:11,569
Are you scared of a bot here?
470
00:21:11,603 --> 00:21:13,939
...and build a robotic hand
471
00:21:13,972 --> 00:21:16,541
with near-human dexterity.
472
00:21:16,575 --> 00:21:20,312
It's what Eric's hoping to do to get the checkered flag.
473
00:21:21,513 --> 00:21:24,983
You can see he's on his way to the top 10.
474
00:21:25,016 --> 00:21:27,719
Yeah, we got through, Andrew, focus here.
475
00:21:29,287 --> 00:21:31,222
And you go up a few cars,
476
00:21:31,255 --> 00:21:33,057
you'll find the 3 of Austin Dillon
477
00:21:33,091 --> 00:21:34,258
up in sixth place,
478
00:21:34,292 --> 00:21:36,862
making up time on the race leader.
479
00:21:44,436 --> 00:21:48,373
So the recommendation is pit on lap 327.
480
00:21:51,075 --> 00:21:54,946
What my fear is is that they'll pit with the leaders
481
00:21:54,979 --> 00:21:58,016
instead of actually running to the strategy.
482
00:21:58,049 --> 00:22:00,352
Going to the pits when the leaders do
483
00:22:00,385 --> 00:22:02,721
is the safe play in the end stages of a race...
484
00:22:02,754 --> 00:22:04,155
Sparks, you got me?
485
00:22:04,188 --> 00:22:07,292
...but the A.I. tool is recommending a riskier plan
486
00:22:07,325 --> 00:22:10,929
that might gain them valuable seconds.
487
00:22:13,698 --> 00:22:15,233
By pitting later,
488
00:22:15,267 --> 00:22:17,202
Austin Dillon will have faster tires
489
00:22:17,235 --> 00:22:19,338
for the closing laps of the race,
490
00:22:19,371 --> 00:22:21,940
but he risks falling further behind the leaders
491
00:22:21,973 --> 00:22:24,943
once they come out of the pits with their fresh tires.
492
00:22:29,313 --> 00:22:32,350
A lot of times when our Pit Rho technology tells us,
493
00:22:32,383 --> 00:22:36,621
"This is the time to pit," or, "This is how to do it," it doesn't feel right.
494
00:22:38,256 --> 00:22:40,091
Are you sure you wanna do that now?
495
00:22:40,125 --> 00:22:41,993
Sometimes you might be sitting out there
496
00:22:42,026 --> 00:22:43,394
running laps on older tires,
497
00:22:43,428 --> 00:22:45,030
where everybody else is pitted,
498
00:22:45,063 --> 00:22:47,966
and it's like, it doesn't feel right for the driver.
499
00:22:52,370 --> 00:22:54,171
He's gonna want to pit,
500
00:22:54,205 --> 00:22:57,308
and you gotta convince him, "Stay, make good laps. Trust us, it's gonna work."
501
00:22:57,342 --> 00:23:00,712
Some leaders are gonna pit right here, and we need to run.
502
00:23:03,515 --> 00:23:05,216
Looks like the 22
503
00:23:05,250 --> 00:23:06,985
is gonna choose to come down pit road.
504
00:23:07,018 --> 00:23:10,288
So, all the front four came in on the same lap
505
00:23:10,321 --> 00:23:11,456
with 82 laps to go.
506
00:23:11,489 --> 00:23:13,591
On lap 318,
507
00:23:13,624 --> 00:23:17,095
the top four cars enter the pits.
508
00:23:26,103 --> 00:23:29,207
Here's where the faith in the tool ends up happening.
509
00:23:29,240 --> 00:23:30,575
When they all pit,
510
00:23:30,608 --> 00:23:32,443
it takes a lot of faith to just stay out there
511
00:23:32,477 --> 00:23:33,712
and run to your lap.
512
00:23:39,717 --> 00:23:42,420
Austin Dillon breaks from the A.I. strategy
513
00:23:42,453 --> 00:23:44,455
and follows the leaders into the pits.
514
00:23:46,591 --> 00:23:47,926
That's not good news.
515
00:23:47,959 --> 00:23:50,362
Three, two, one.
516
00:23:50,395 --> 00:23:52,697
Put on the brakes, wheels lift.
517
00:23:52,731 --> 00:23:54,599
To maintain their position,
518
00:23:54,632 --> 00:23:57,903
the team needs a flawless pit stop.
519
00:24:09,247 --> 00:24:11,216
Son of a bitch!
520
00:24:14,986 --> 00:24:17,088
We're not gonna be nowhere near 'em.
521
00:24:17,121 --> 00:24:18,356
Got killed on pit road.
522
00:24:18,389 --> 00:24:20,825
It's pretty disastrous.
523
00:24:20,858 --> 00:24:23,995
Prior to the pit stop, we were about 4.6 seconds back,
524
00:24:24,029 --> 00:24:26,298
but when we came out, we were nine seconds back,
525
00:24:26,331 --> 00:24:28,766
so we lost about four and a half seconds
526
00:24:28,799 --> 00:24:31,336
on that-- in that exchange. That's hard to get back.
527
00:24:31,369 --> 00:24:34,038
Let's go to work on him. This won't be easy.
528
00:24:34,071 --> 00:24:35,206
Just fight hard here.
529
00:24:35,240 --> 00:24:38,776
They've dropped from 6th place to 12th...
530
00:24:38,810 --> 00:24:40,712
and Austin Dillon has very little time left
531
00:24:40,745 --> 00:24:42,814
to fight his way back to the leaders.
532
00:24:42,847 --> 00:24:44,649
Come on, Austin, get him.
533
00:24:46,584 --> 00:24:48,186
...but the new tires give him an edge...
534
00:24:51,522 --> 00:24:53,124
The white flag waves,
535
00:24:53,158 --> 00:24:54,559
one lap to go.
536
00:25:01,465 --> 00:25:05,236
Get it, get it, get it!
537
00:25:05,270 --> 00:25:09,507
Short track win number one for Martin Truex!
538
00:25:09,540 --> 00:25:11,009
Sixth place is awesome.
539
00:25:11,042 --> 00:25:13,478
...and he ends up finishing sixth.
540
00:25:13,511 --> 00:25:16,948
Hell of a freakin' drive, Austin Dillon.
541
00:25:16,982 --> 00:25:20,084
Hey, nice work tonight, man, way to fight hard there.
542
00:25:20,118 --> 00:25:22,187
Hell of a job, boys.
543
00:25:23,989 --> 00:25:25,890
Hey, good job, guys.
544
00:25:25,923 --> 00:25:27,559
Progressing through the race
545
00:25:27,592 --> 00:25:29,560
definitely the cars have gotten faster,
546
00:25:29,593 --> 00:25:30,828
so, you know, we'll see good things
547
00:25:30,862 --> 00:25:32,430
that we'll take back next time we go to Richmond.
548
00:25:32,464 --> 00:25:34,465
Hell of a job this weekend, boys.
549
00:25:34,499 --> 00:25:36,200
The hardest thing
550
00:25:36,234 --> 00:25:38,970
as we've incorporated more A.I.-based tools
551
00:25:39,003 --> 00:25:40,005
is trust.
552
00:25:41,973 --> 00:25:44,509
Sometimes we're the ones that get in the way, right?
553
00:25:44,543 --> 00:25:47,612
There's still times when it's counterintuitive,
554
00:25:47,645 --> 00:25:48,813
and everybody's like,
555
00:25:48,847 --> 00:25:50,415
"It's the wrong call, it's the wrong call,"
556
00:25:50,448 --> 00:25:52,149
and over time, we have these battles
557
00:25:52,183 --> 00:25:55,820
because most of the time, the A.I. tools is right.
558
00:25:55,854 --> 00:25:58,723
Nine times out of ten, or even more, it's the right call.
559
00:25:58,756 --> 00:26:01,525
Andy and Eric's team were using A.I.,
560
00:26:01,559 --> 00:26:03,728
and on track for a strong finish,
561
00:26:03,762 --> 00:26:05,297
but they fell behind
562
00:26:05,330 --> 00:26:06,764
when the team ignored the machine
563
00:26:06,797 --> 00:26:08,233
and went with their intuition.
564
00:26:09,334 --> 00:26:11,236
That'll do it.
565
00:26:11,269 --> 00:26:12,703
Convincing humans
566
00:26:12,737 --> 00:26:13,805
that machines know what they're doing
567
00:26:13,838 --> 00:26:14,805
is the central difficulty
568
00:26:14,839 --> 00:26:16,907
in deploying A.I. out in society,
569
00:26:16,941 --> 00:26:19,410
whether it's the pit boss in car racing,
570
00:26:19,443 --> 00:26:22,013
or even astronauts flying to the moon.
571
00:26:22,046 --> 00:26:25,550
Do we trust the A.I. to make decisions for us?
572
00:26:25,583 --> 00:26:27,284
We already do with GPS maps.
573
00:26:27,318 --> 00:26:30,588
Perhaps here, the team just didn't have enough experience with it
574
00:26:30,621 --> 00:26:33,325
to override their own intuition,
575
00:26:33,358 --> 00:26:35,327
but what about other situations?
576
00:26:37,228 --> 00:26:39,597
At what point do we start trusting A.I.
577
00:26:39,631 --> 00:26:41,399
in more serious matters?
578
00:26:42,600 --> 00:26:45,136
Matters of life and death?
579
00:26:45,170 --> 00:26:48,773
It was a smoldering fire that filled the whole house with smoke,
580
00:26:48,806 --> 00:26:51,342
and you couldn't see your hand in front of your face.
581
00:26:51,376 --> 00:26:54,145
You literally had to feel your way up the stairs.
582
00:26:54,178 --> 00:26:56,314
Totally blind search.
583
00:26:56,347 --> 00:26:59,350
Yeah. Sometimes that's the best thing we can do.
584
00:26:59,383 --> 00:27:00,851
Yeah.
585
00:27:00,885 --> 00:27:02,887
Every two hours and 45 minutes,
586
00:27:02,921 --> 00:27:06,824
a U.S. citizen dies by fire in their own home.
587
00:27:06,858 --> 00:27:08,692
We've lost more than 3,000 a year
588
00:27:08,726 --> 00:27:10,729
consistently for 30 years.
589
00:27:10,762 --> 00:27:13,231
The Worcester fire.
590
00:27:13,264 --> 00:27:17,368
Three guys go in. They all get disoriented and get lost.
591
00:27:17,401 --> 00:27:18,770
Two more go in to find them.
592
00:27:18,803 --> 00:27:20,739
They get lost. Two more go in.
593
00:27:20,772 --> 00:27:23,508
I mean, before you know it, they finally had to, "Okay!
594
00:27:23,541 --> 00:27:25,610
We're not sending any more guys in there,
595
00:27:25,643 --> 00:27:27,411
'cause they're all friggin' lost."
596
00:27:27,444 --> 00:27:31,582
On his radio, a commanding officer heard two firefighters
597
00:27:31,615 --> 00:27:33,217
desperately crying out for help.
598
00:27:33,251 --> 00:27:36,120
"Mayday, mayday. We're running out of air.
599
00:27:36,153 --> 00:27:38,790
Come to the door so we can see where you are,"
600
00:27:38,823 --> 00:27:41,492
and then, we did that, and we went beyond the door,
601
00:27:41,526 --> 00:27:43,260
and we yelled, and we had lights,
602
00:27:43,294 --> 00:27:44,395
and they were...
603
00:27:44,428 --> 00:27:48,265
they were inside somewhere that they couldn't see us.
604
00:27:48,299 --> 00:27:52,037
All those guys who died in that...
605
00:27:56,440 --> 00:28:00,211
When we go into a structure that's dark and smoky,
606
00:28:00,245 --> 00:28:03,814
the biggest challenge is the visibility.
607
00:28:03,848 --> 00:28:07,585
The ability to navigate is a... is a challenge,
608
00:28:07,619 --> 00:28:11,055
and often firefighters have become disoriented,
609
00:28:11,088 --> 00:28:13,725
and then they run out of air.
610
00:28:17,295 --> 00:28:20,231
With the challenge of smoke and having no vision,
611
00:28:20,265 --> 00:28:23,835
I knew that there was a possibility of changing that.
612
00:28:26,604 --> 00:28:29,006
That's when I finally met the C-THRU team.
613
00:28:29,040 --> 00:28:31,508
Okay, is the system turning on?
614
00:28:31,542 --> 00:28:33,777
Let's see.
615
00:28:33,811 --> 00:28:35,913
I'm gonna unplug that one.
616
00:28:35,947 --> 00:28:38,015
I guess the best way to describe myself
617
00:28:38,049 --> 00:28:40,151
is I'm infinitely curious.
618
00:28:41,285 --> 00:28:42,987
I like to solve problems,
619
00:28:43,020 --> 00:28:44,789
look at things through a new lens.
620
00:28:44,822 --> 00:28:47,491
All right...
621
00:28:47,525 --> 00:28:51,128
I was in disbelief that firefighting in a smoked-out building
622
00:28:51,162 --> 00:28:52,897
involves training their personnel
623
00:28:52,930 --> 00:28:55,533
to revert back to feeling around the room.
624
00:28:55,566 --> 00:28:56,967
How's the battery level doing?
625
00:28:57,001 --> 00:28:58,903
That was really the inspiration
626
00:28:58,936 --> 00:29:00,805
behind creating C-THRU.
627
00:29:05,643 --> 00:29:08,746
Sam Cossman saw the light when he jumped into a volcano.
628
00:29:08,779 --> 00:29:09,981
Line!
629
00:29:10,014 --> 00:29:12,249
Fire!
630
00:29:12,283 --> 00:29:14,852
Part globetrotting adrenaline junkie,
631
00:29:14,885 --> 00:29:16,421
part computer engineer,
632
00:29:16,454 --> 00:29:19,390
the self-proclaimed Indiana Jones of tech
633
00:29:19,423 --> 00:29:22,260
envisioned a tool that would help firefighters
634
00:29:22,293 --> 00:29:24,328
and save lives,
635
00:29:24,361 --> 00:29:26,331
a kind of X-ray vision.
636
00:29:28,999 --> 00:29:31,702
The problem that C-THRU is trying to solve
637
00:29:31,735 --> 00:29:33,471
is really flipping the lights on
638
00:29:33,504 --> 00:29:37,074
for people operating in zero-visibility conditions.
639
00:29:37,107 --> 00:29:39,644
The concept of C-THRU
640
00:29:39,677 --> 00:29:42,547
was the helmet that had enhanced audio,
641
00:29:42,580 --> 00:29:43,748
enhanced vision...
642
00:29:43,781 --> 00:29:46,918
I see you! I'm on my way.
643
00:29:46,951 --> 00:29:50,587
...outlines their surrounding geometry
644
00:29:50,621 --> 00:29:52,189
so that they can navigate faster.
645
00:29:52,223 --> 00:29:55,693
So is it this plane right here that... that changed recently?
646
00:29:55,726 --> 00:29:59,697
Yes, basically like a simpler design that can achieve more.
647
00:29:59,730 --> 00:30:01,065
We have a mask,
648
00:30:01,098 --> 00:30:02,633
and we have a thermal-imaging device
649
00:30:02,666 --> 00:30:04,401
that sits on the side of that mask,
650
00:30:04,435 --> 00:30:08,272
and we process that image through a small computer.
651
00:30:08,306 --> 00:30:10,174
Sam and Omer created a mask
652
00:30:10,207 --> 00:30:12,243
with special glasses clipped inside
653
00:30:12,276 --> 00:30:15,780
which allows firefighters to see edges as green lines
654
00:30:15,813 --> 00:30:18,783
in an augmented reality overlay.
655
00:30:18,816 --> 00:30:20,818
How's the alignment look on that one?
656
00:30:20,852 --> 00:30:22,086
It's not bad.
657
00:30:22,119 --> 00:30:24,521
We need to calibrate it a little bit more.
658
00:30:24,555 --> 00:30:27,425
Omer and I have been working on refining the prototypes
659
00:30:27,458 --> 00:30:28,893
for the last couple of years,
660
00:30:28,926 --> 00:30:30,728
just trying to MacGyver some of these problems
661
00:30:30,761 --> 00:30:34,031
with off-the-shelf parts, you know, duct tape and bubble gum.
662
00:30:34,064 --> 00:30:36,568
Move your hand around a little bit.
663
00:30:36,601 --> 00:30:38,235
- Okay.
- Other hand, like that one.
664
00:30:38,269 --> 00:30:40,972
Yeah, this is definitely better.
665
00:30:41,005 --> 00:30:43,975
It may look like old Tron-era night vision,
666
00:30:44,008 --> 00:30:45,509
but there's actually
667
00:30:45,542 --> 00:30:48,679
some pretty slick artificial intelligence at work here.
668
00:30:48,712 --> 00:30:50,214
Thermal imaging cameras
669
00:30:50,247 --> 00:30:52,416
stream video from the firefighter's helmet
670
00:30:52,449 --> 00:30:54,485
into an A.I. processor.
671
00:30:54,518 --> 00:30:56,286
Using infrared light
672
00:30:56,320 --> 00:30:59,123
and a powerful edge-detection algorithm,
673
00:30:59,157 --> 00:31:02,227
the mask detects subtle changes in brightness
674
00:31:02,260 --> 00:31:05,663
to predict shapes invisible to the human eye,
675
00:31:05,696 --> 00:31:07,965
like a wall hidden by smoke,
676
00:31:07,998 --> 00:31:10,568
or a kid hiding under a bed.
677
00:31:12,870 --> 00:31:14,839
There you go, take this mask.
678
00:31:14,872 --> 00:31:16,340
Sam and Omer
679
00:31:16,373 --> 00:31:19,109
are now at a familiar point in the innovator's journey...
680
00:31:19,143 --> 00:31:22,513
get out of the garage and into the real world
681
00:31:22,547 --> 00:31:25,450
to see if their invention can take the heat.
682
00:31:27,084 --> 00:31:29,586
Fire Dispatch, Medic 71 arrived on scene,
683
00:31:29,620 --> 00:31:30,922
have report of smoke showing.
684
00:31:30,955 --> 00:31:32,489
Fire Dispatch, copy.
685
00:31:32,523 --> 00:31:35,059
One of the most important things any fire department does
686
00:31:35,092 --> 00:31:37,896
is regular hands-on training.
687
00:31:38,729 --> 00:31:39,697
There he is.
688
00:31:39,730 --> 00:31:40,998
How you doing, Captain?
689
00:31:41,031 --> 00:31:42,566
Good to see you, brother.
690
00:31:42,600 --> 00:31:45,436
We're gonna give the C-THRU solution a hard run...
691
00:31:45,469 --> 00:31:47,939
We've got a prototype fresh off the print.
692
00:31:47,972 --> 00:31:50,174
...and we're gonna put it in fire and smoke,
693
00:31:50,207 --> 00:31:53,678
and we're gonna see how it acts while crews are working with it.
694
00:31:53,711 --> 00:31:56,314
- Shall we get him inside the smoke?
- Let's do it!
695
00:31:57,715 --> 00:31:59,083
...cleared for dispatch.
696
00:31:59,717 --> 00:32:01,352
I am a Cyborg.
697
00:32:01,386 --> 00:32:05,155
Okay, we're ready.
698
00:32:05,189 --> 00:32:06,991
Crews will be doing live fire drills
699
00:32:07,024 --> 00:32:08,293
in our training tower.
700
00:32:10,027 --> 00:32:12,296
It is active, real fire
701
00:32:12,329 --> 00:32:15,566
with temperatures at the ceiling at 1,200 degrees.
702
00:32:22,539 --> 00:32:23,975
Anybody over here?
703
00:32:25,342 --> 00:32:27,511
Firefighters are in a hurry,
704
00:32:27,545 --> 00:32:28,679
looking for victims.
705
00:32:28,713 --> 00:32:32,483
Visibility will be limited at best.
706
00:32:35,053 --> 00:32:37,689
Often, firefighters will be able to see nothing.
707
00:33:02,413 --> 00:33:04,048
C-THRU's maiden voyage
708
00:33:04,081 --> 00:33:06,350
is cut short by a malfunction.
709
00:33:06,384 --> 00:33:08,151
In an active firefight,
710
00:33:08,185 --> 00:33:10,954
it's critical that things work.
711
00:33:10,988 --> 00:33:12,757
It's life and death.
712
00:33:12,790 --> 00:33:17,195
Uh, at first, it was good. I got through, went down to the floor,
713
00:33:17,228 --> 00:33:19,463
and I looked, and I could see everything clear.
714
00:33:19,496 --> 00:33:20,798
Yeah.
715
00:33:20,831 --> 00:33:23,468
Really well, everything was lined out. Once I started working...
716
00:33:23,501 --> 00:33:24,902
- Yeah?
- I lost it.
717
00:33:24,936 --> 00:33:27,238
- Yeah, the signal went out.
- Signal went out.
718
00:33:27,271 --> 00:33:29,807
I'm not sure what that was, but we're gonna figure it out.
719
00:33:29,841 --> 00:33:32,643
There was a lot of interference, or maybe a cable issue.
720
00:33:32,676 --> 00:33:35,313
We did encounter some challenges, the biggest of them
721
00:33:35,346 --> 00:33:37,948
was some wi-fi interference that we've encountered
722
00:33:37,982 --> 00:33:39,750
where the system would just shut down.
723
00:33:39,783 --> 00:33:42,653
Yeah, it's actually like over here with the connections,
724
00:33:42,686 --> 00:33:45,089
- like, this pin, you know?
- That's what's...
725
00:33:45,122 --> 00:33:49,260
Yeah, the pin connections here, and here, actually.
726
00:33:49,293 --> 00:33:52,663
We should just shield the cables as best we can
727
00:33:52,696 --> 00:33:54,465
and give it another go.
728
00:33:56,333 --> 00:33:58,068
Battalion Ten, Fire Dispatch,
729
00:33:58,102 --> 00:33:59,937
uh, we got a caller on the second floor
730
00:33:59,971 --> 00:34:01,372
trapped in the bathroom.
731
00:34:01,406 --> 00:34:04,008
So, if you wanna go ahead and try it on for a fit,
732
00:34:04,041 --> 00:34:05,943
we'll see how it goes.
733
00:34:05,976 --> 00:34:09,813
-Fire Dispatch, Battalion Ten...
734
00:34:09,847 --> 00:34:14,118
Engine 72 arrived on scene, reporting of heavy smoke showing
735
00:34:14,151 --> 00:34:16,287
from the first and second floor.
736
00:34:20,991 --> 00:34:22,460
We've got smoke showing
737
00:34:22,493 --> 00:34:24,562
from the first and second floors.
738
00:34:29,033 --> 00:34:30,268
Engine 7-1,
739
00:34:30,301 --> 00:34:31,803
you're gonna be taking fire attack.
740
00:34:46,250 --> 00:34:49,219
71, who is on scene,
741
00:34:49,253 --> 00:34:52,457
smoke showing from the second and first floor.
742
00:35:05,802 --> 00:35:09,506
Command copies, one victim coming out of the second window, you need EMS.
743
00:35:09,540 --> 00:35:12,376
Medic 72, you're gonna have to take patient care.
744
00:35:12,409 --> 00:35:16,247
As soon as I got in, I could see the outline of the room.
745
00:35:16,280 --> 00:35:18,816
As I stepped in, I just kinda took a look around,
746
00:35:18,849 --> 00:35:22,052
I could see where the victim was and an outline of the door.
747
00:35:22,085 --> 00:35:24,989
- I mean, hands free, you know?
- Yeah.
748
00:35:27,391 --> 00:35:30,060
It is kind of like, I mean, like Iron Man, you know,
749
00:35:30,094 --> 00:35:31,362
being able to see through the smoke,
750
00:35:31,395 --> 00:35:35,199
and having everything so clear-cut, um...
751
00:35:35,232 --> 00:35:37,735
It's... it's pretty cool.
752
00:35:37,768 --> 00:35:41,105
What we're working on is really a game-changing tool
753
00:35:41,138 --> 00:35:44,074
that completely has the potential to transform
754
00:35:44,108 --> 00:35:45,575
how the work here is done.
755
00:35:45,609 --> 00:35:47,344
This is, uh, some of the videos
756
00:35:47,377 --> 00:35:49,247
of the C-THRU mask, okay?
757
00:35:49,280 --> 00:35:51,516
That is way crisper than I've seen.
758
00:35:51,549 --> 00:35:53,184
That is insane.
759
00:35:53,217 --> 00:35:57,187
Over the 30 years that I've been at this, I've seen a lot of changes.
760
00:35:57,220 --> 00:35:59,390
We have mobile data computers,
761
00:35:59,423 --> 00:36:01,525
we have computer-aided dispatch systems...
762
00:36:01,558 --> 00:36:03,026
- No, that's gonna...
- Wow.
763
00:36:03,060 --> 00:36:04,962
That's gonna be a game-changer.
764
00:36:04,995 --> 00:36:09,700
...and now we have the possibility with machine learning and A. I.
765
00:36:09,733 --> 00:36:11,268
to progress to a place
766
00:36:11,301 --> 00:36:13,837
just a couple of years ago we couldn't have imagined.
767
00:36:13,871 --> 00:36:16,440
- Is that completely pitch dark in there?
- That recording--
768
00:36:16,473 --> 00:36:18,242
Oh, gotta go!
769
00:36:18,275 --> 00:36:20,878
That is actually what you see in the mask.
770
00:36:20,911 --> 00:36:24,014
We're gonna go on another call, gentlemen.
771
00:36:24,048 --> 00:36:25,582
It's impossible to know
772
00:36:25,616 --> 00:36:27,417
if this technology could have saved
773
00:36:27,451 --> 00:36:29,720
those six firefighters in Worcester,
774
00:36:29,753 --> 00:36:32,657
but it's hard to believe it wouldn't have helped.
775
00:36:37,194 --> 00:36:38,729
Back on Cathedral Ledge,
776
00:36:38,762 --> 00:36:41,632
Jim is about to see if his new bionic leg
777
00:36:41,665 --> 00:36:44,835
will help him scale a 700-foot sheer rock face.
778
00:36:44,869 --> 00:36:47,572
I'm just gonna kinda bring everything.
779
00:36:50,608 --> 00:36:52,276
All right.
780
00:36:52,309 --> 00:36:55,746
My own personal M.I.T. pit crew.
781
00:36:55,780 --> 00:36:57,581
- Got the socket.
- The socket...
782
00:36:57,614 --> 00:36:59,816
What we're gonna do today
783
00:36:59,850 --> 00:37:03,387
is climb on Cathedral Ledge with a new robot foot
784
00:37:03,420 --> 00:37:05,623
designed specifically for climbing.
785
00:37:10,127 --> 00:37:12,697
We can set up camp here.
786
00:37:13,664 --> 00:37:14,631
All right,
787
00:37:14,665 --> 00:37:17,001
we should be ready to start calibrating.
788
00:37:18,935 --> 00:37:20,671
Counterflex.
789
00:37:22,139 --> 00:37:23,540
Rest.
790
00:37:23,574 --> 00:37:26,110
- You're driving now.
- That's me.
791
00:37:26,143 --> 00:37:27,411
How's it feel?
792
00:37:27,444 --> 00:37:29,246
Pretty accurate.
793
00:37:29,280 --> 00:37:31,949
It's going everywhere that I'm telling it to go.
794
00:37:31,982 --> 00:37:34,584
This climb is gonna be very challenging,
795
00:37:34,618 --> 00:37:38,122
because there's a variety of holds at different angles,
796
00:37:38,155 --> 00:37:39,657
different heights.
797
00:37:39,690 --> 00:37:41,792
I'd say there's a high probability
798
00:37:41,825 --> 00:37:45,263
of there being some falling action here and there.
799
00:37:47,965 --> 00:37:50,167
I was really afraid,
800
00:37:50,200 --> 00:37:51,868
very... worried
801
00:37:51,902 --> 00:37:55,006
whether or not I could make it all the way up a climb.
802
00:38:05,483 --> 00:38:07,051
We good there, Joe?
803
00:38:13,224 --> 00:38:14,959
Harness is on.
804
00:38:16,426 --> 00:38:18,362
I got plenty of gear.
805
00:38:22,466 --> 00:38:24,635
All right, we're climbing.
806
00:38:30,273 --> 00:38:32,476
I think I'm at a crux section here.
807
00:38:36,814 --> 00:38:38,983
Well, first fall.
808
00:38:39,016 --> 00:38:40,985
I'm not sticking very well.
809
00:38:47,190 --> 00:38:50,061
It's hard. Hard business.
810
00:38:56,967 --> 00:38:57,935
Slack!
811
00:39:04,975 --> 00:39:06,010
Oh!
812
00:39:08,479 --> 00:39:10,114
We have failure.
813
00:39:10,981 --> 00:39:13,183
The whole mechanism broke.
814
00:39:13,216 --> 00:39:15,419
I remember looking down at it,
815
00:39:15,452 --> 00:39:18,622
seeing the foot at a strange angle,
816
00:39:18,655 --> 00:39:20,757
and, "Holy crap, that is gonna hurt.
817
00:39:20,791 --> 00:39:23,727
"That--" Like, I was bracing for pain.
818
00:39:23,761 --> 00:39:26,363
I mean, how much more of a part of you
819
00:39:26,397 --> 00:39:27,732
does it need to be?
820
00:39:35,005 --> 00:39:36,406
Oh, my God.
821
00:39:36,440 --> 00:39:38,509
I would call that a catastrophic failure.
822
00:39:38,542 --> 00:39:39,943
Pretty catastrophic.
823
00:39:39,977 --> 00:39:42,546
But it was... it was a strange sensation, though,
824
00:39:42,579 --> 00:39:45,749
because all of a sudden, my ankle was broken,
825
00:39:45,782 --> 00:39:49,286
and you feel like you're losing your limb
826
00:39:49,319 --> 00:39:51,121
all over again.
827
00:39:51,154 --> 00:39:54,558
How're you feeling? You feel like you wanna go down again?
828
00:39:54,591 --> 00:39:56,127
We... we did bring a spare.
829
00:39:56,893 --> 00:39:58,629
Uh, sure.
830
00:39:58,662 --> 00:40:00,698
Okay, we'll swap it over to this one.
831
00:40:02,599 --> 00:40:04,101
In engineering,
832
00:40:04,134 --> 00:40:07,004
we're kinda used to things not exactly going right
833
00:40:07,037 --> 00:40:07,904
the first time,
834
00:40:07,938 --> 00:40:10,006
so that's why we have contingency plans.
835
00:40:10,040 --> 00:40:12,810
So, this is the last climbing robot leg
836
00:40:12,843 --> 00:40:13,977
in the world, Jim.
837
00:40:16,079 --> 00:40:17,648
We're good to go again.
838
00:40:23,720 --> 00:40:27,158
I'm a little nervous about trusting this foot now.
839
00:40:28,392 --> 00:40:29,526
Watch me here.
840
00:40:29,559 --> 00:40:31,796
If the left-- if the robot breaks...
841
00:40:33,397 --> 00:40:35,733
I'm going for a ride.
842
00:40:43,173 --> 00:40:44,809
Actually, it did that move.
843
00:40:44,842 --> 00:40:45,876
Well done.
844
00:40:46,843 --> 00:40:49,513
We're rock climbing, dude.
845
00:40:49,546 --> 00:40:54,285
With the robotic leg, I found that I could move more naturally.
846
00:40:54,318 --> 00:40:55,586
Life on the edge, man.
847
00:40:55,619 --> 00:40:58,555
I was pain free, and it was, I don't know,
848
00:40:58,588 --> 00:41:00,925
it was just kind of fun and satisfying.
849
00:41:00,958 --> 00:41:03,560
We have always hypothesized
850
00:41:03,594 --> 00:41:07,231
that if we can link the nerves of a human being
851
00:41:07,264 --> 00:41:08,865
to a bionic limb,
852
00:41:08,898 --> 00:41:11,402
the limb would become part of the person,
853
00:41:11,435 --> 00:41:13,504
part of identity.
854
00:41:14,271 --> 00:41:16,173
Toppin' out.
855
00:41:16,206 --> 00:41:18,275
Remarkably, it's happened.
856
00:41:18,309 --> 00:41:20,844
Cyborg power!
857
00:41:23,713 --> 00:41:25,716
It's a tall peak,
858
00:41:25,749 --> 00:41:27,550
but pales in comparison
859
00:41:27,584 --> 00:41:30,521
to the one Hugh is ultimately trying to climb.
860
00:41:30,554 --> 00:41:31,988
We also have the goal
861
00:41:32,022 --> 00:41:35,492
of extending human capability beyond physiological function,
862
00:41:35,525 --> 00:41:38,294
jump higher, or run faster...
863
00:41:38,328 --> 00:41:42,432
So bionics not only seeks to achieve normative function in humans,
864
00:41:42,466 --> 00:41:45,269
but also to extend human expression
865
00:41:45,302 --> 00:41:47,771
beyond what people were born with.
866
00:41:48,805 --> 00:41:51,508
Human enhancement and augmentation
867
00:41:51,541 --> 00:41:53,310
have been around through human history
868
00:41:53,344 --> 00:41:54,777
and mythology,
869
00:41:54,811 --> 00:41:57,314
from Prometheus stealing fire
870
00:41:57,348 --> 00:41:59,049
to the Civil War.
871
00:42:00,417 --> 00:42:02,852
Using tools to improve our abilities
872
00:42:02,886 --> 00:42:04,721
is a fundamental human development,
873
00:42:04,754 --> 00:42:08,625
whether it's stone spears to protect our families
874
00:42:08,658 --> 00:42:10,895
or airplanes to transport us farther.
875
00:42:10,928 --> 00:42:14,498
...and that's really what we're seeing, the transformation of society,
876
00:42:14,531 --> 00:42:16,366
and not just racing, not just sports,
877
00:42:16,399 --> 00:42:19,803
is really using these A.I. tools, and they'll become commonplace,
878
00:42:19,836 --> 00:42:21,371
won't even be thought about otherwise.
879
00:42:21,404 --> 00:42:23,440
A.I. and machine learning...
880
00:42:23,473 --> 00:42:25,509
they're just tools,
881
00:42:25,543 --> 00:42:30,213
ones that makes us stronger, smarter, faster.
882
00:42:30,246 --> 00:42:33,249
A.I. will play an increasingly dominant role
883
00:42:33,283 --> 00:42:36,786
across all the many dimensions of what it means to be human.
884
00:42:36,820 --> 00:42:38,455
There's a good chance
885
00:42:38,488 --> 00:42:42,893
A.I. will continue to enhance us in ways both known and unknown,
886
00:42:42,926 --> 00:42:46,930
size:84% position:42% eventually becoming as invisible as the air we breathe.
887
00:42:46,963 --> 00:42:49,600
That narrative will play out
888
00:42:49,633 --> 00:42:51,568
across all types of human conditions.
889
00:42:51,602 --> 00:42:54,671
That will enhance human capability,
890
00:42:54,705 --> 00:42:58,042
fundamentally change who we are as a human race.
891
00:42:59,609 --> 00:43:01,978
The question then becomes,
892
00:43:02,012 --> 00:43:03,280
if it does,
893
00:43:04,648 --> 00:43:08,786
what do we do with our newfound superpowers?
894
00:43:16,660 --> 00:43:19,029
This one's meant to be
895
00:43:19,062 --> 00:43:21,698
kind of an all-around athletic foot.
896
00:43:21,732 --> 00:43:25,502
I can run with it, hiking, biking, whatever I want.
897
00:43:25,536 --> 00:43:26,870
I even use it for surfing,
898
00:43:26,904 --> 00:43:30,307
'cause it has a good bit of flex to surf with.
899
00:43:30,340 --> 00:43:32,575
I actually liked the fit so much
900
00:43:32,609 --> 00:43:34,978
that it's the only one I use now.
901
00:43:35,011 --> 00:43:37,114
As good as this fit is, it still...
902
00:43:37,147 --> 00:43:40,183
like I said, high activity days, I get some pressure sores.
903
00:43:40,217 --> 00:43:42,652
Every night, you have to look over the skin,
904
00:43:42,686 --> 00:43:45,522
make sure you've got nothing nasty going on,
905
00:43:45,556 --> 00:43:48,192
nothing growing where it shouldn't be.
906
00:43:48,225 --> 00:43:51,561
This guy is a gel liner.
907
00:43:51,595 --> 00:43:52,730
It doesn't breathe,
908
00:43:52,763 --> 00:43:55,866
but this is what keeps the leg on.
909
00:43:55,899 --> 00:43:57,133
A lot of, um, amputees
910
00:43:57,167 --> 00:43:59,703
talk about forgetting that they don't have a leg
911
00:43:59,736 --> 00:44:01,071
in the middle of the night,
912
00:44:01,104 --> 00:44:03,440
and they get up in the middle of the night
913
00:44:03,473 --> 00:44:04,707
to go to the bathroom,
914
00:44:04,741 --> 00:44:06,943
and then instantly fall on their faces.
915
00:44:06,976 --> 00:44:09,112
That's only happened to me once,
916
00:44:09,146 --> 00:44:12,582
but I managed to catch myself before I hit the ground.