Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
Every New Year, it's tradition to set
0:02
new goals. well. imagine setting a goal
0:04
now that last the rest of your
0:07
life for a limited time, Rosetta Stone
0:09
is offering a lifetime membership for just
0:11
fifty percent off. That's not an annual
0:13
payment, that's a one time payment. You
0:16
get full access to twenty five languages
0:18
and with it access to the world
0:20
one language at a time. Visit Rosetta
0:23
stone.com now and save fifty percent. That's
0:25
fifty percent off a lifetime membership. Welcome
0:29
to your twenties. When he three work recap
0:31
this year, you've been to one hundred and
0:33
twenty seven think meetings. You spent fifty six
0:36
minutes or team for files and almost missed
0:38
a deadline. He a. Twenty
0:40
Twenty Four can and should sound different
0:43
with monday.com You can work together easily,
0:45
collaborate and share data files and up
0:47
Dave. So all work happens in one
0:50
place and everyone's on the same paid.
0:52
Go to monday.com or tap the banner
0:54
to learn more. Imam.
1:05
I'm older and this is
1:07
clear in vivid conversations about
1:10
connecting in communicating. Since
1:15
I was a little girl
1:18
like eleven twelve year old,
1:20
I just loved Physics. A
1:22
was my first love, and
1:25
in hindsight, what happened is
1:27
I think I love that
1:29
audacious quest to the unknown
1:31
mystery of the universe. Then.
1:34
Enlighten. Than a manner
1:36
that I realized my oh
1:38
audacious question that I love
1:41
the most is what is
1:43
intelligence What makes intelligence? How
1:45
do we build intelligent machines
1:47
And that shift in the
1:49
middle of end of my
1:51
college year. Was
1:54
how I discovered a
1:56
I. That's. Faye Faye
1:58
Lee She's off. Called somewhat
2:00
to her embarrassment, the godmother of
2:03
Ai. My conversation with
2:05
Or is the first of three
2:07
special episodes of Clear In Vivid,
2:09
exploring the dramatic impact that artificial
2:11
intelligence his head in the last
2:13
twelve months since chat bots such
2:15
as Said She P T burst
2:17
on the scene. Doctor.
2:19
Lee is written a wonderful book
2:21
recounting her work laying the foundation
2:23
for those chat but some fifteen
2:25
years ago. but the book called
2:27
the World's I See. Is
2:30
also a vivid portrayal of our own
2:32
personal journey from her childhood in China
2:34
through a bone and risky gamble with
2:36
her career. To. Her and
2:38
concern. Today. I should
2:40
help humanity. Not. Destroys.
2:44
Thank you so much for doing this with
2:46
me. I'm on! Was guilty talking to a
2:48
win for these few minutes because I feel
2:50
like I'm stealing you away from making a
2:52
future for all of us. Law.
2:55
Think you're not on I I I
2:57
think part of the theater south of
2:59
communicating with the public. Year
3:02
when you certainly do a great job
3:04
in your book. Such a vivid
3:06
picture of the people in your life who
3:08
have given you. The. Foundation for
3:10
so many things as you do
3:12
that affect all of our lives
3:14
near like the book very much
3:16
Vermont Congratulations the I was so
3:18
interested in huge stories about your
3:20
life in China where you were
3:22
born, And I don't know
3:24
if it was as important a moment to
3:26
you, but it would have been to me
3:28
that moment where. You. Were doing so
3:31
well in school. And. Then
3:33
one day the tissue said at the end is
3:35
a day. The. Boys should stay in,
3:37
the girls are free to go now and you've
3:39
lingered outside the door to listen to what was
3:41
going on. What? What did the teachers say? Oh
3:44
yeah, I have to contextualize though.
3:46
it is still parts of the
3:48
world that top. The. The
3:51
role of tender and that
3:53
deep history is still, you
3:55
know, not the way we
3:57
thought to ask here and
3:59
the teacher. I'm pretty
4:01
sure she was. Coming
4:03
from me. Not
4:06
intentionally trying to hurt anybody, but
4:08
she wanted the boys to be
4:11
better and she wanted to encourage
4:13
the boys And the way see
4:15
encourage them is the. Say.
4:18
Was you probably be least in
4:20
which is that the boys will
4:23
have more potential. The have buffets
4:25
was likely to be smarter or
4:27
happen to us and the. You.
4:30
Know as a little girl who was
4:32
a little feisty. At
4:35
a certain areas and. Didn't rob
4:37
me well. It was a great for
4:39
for for me to. See for myself
4:42
being the person outside the door
4:44
listening to them and a woman
4:46
to advance telling the boys you're
4:48
smarter than girls biologically, I expect
4:51
more from you. But don't worry,
4:53
the few years during adolescence still
4:55
get stupid or. With
4:58
I mean it's crushing and which wonderful
5:00
was you didn't get crushed He due
5:02
to it's almost took it as a
5:05
challenge. Maybe. For. That's because
5:07
of the people around me.
5:09
And dumb. You. Know first
5:12
of all I'm not trying to
5:14
paint and the entire world with
5:16
that message right? This is one
5:18
teacher at it is It is
5:21
rooted in certain aspect of the
5:23
culture of the in the meantime
5:25
my local culture. My family I
5:28
was the the firstborn granted the
5:30
and I was a girl that
5:32
other so supported by by parents
5:35
and by my grandparents that time
5:37
I didn't feel that crossing on.
5:40
Get. Out of. Lack.
5:42
Of confidence in me by
5:44
them and does so as
5:46
a kid I needed just
5:48
a few people who supported
5:50
me. Interested: A and I
5:53
think that's probably. To.
5:55
To a lot of us, we
5:57
just need a few people who.
6:00
Love does so unconditional.
6:02
Issue to move to the
6:04
United States when you were
6:06
fifteen, right? Yeah, That must
6:08
have been a really difficult
6:10
experience to live through because
6:12
you hadn't few words in
6:14
English as I understand. That.
6:17
Plans very, very little if
6:19
any, A was challenging
6:21
but a was also character building
6:24
that someone us ask me this
6:26
question let's get some this actually
6:29
if I did the leave the
6:31
home country. What
6:33
would happen? That's a lie. I
6:36
think I have to say. That
6:38
experience helped me to
6:40
become so much more
6:42
resilient and also. Be.
6:45
Aware of. What?
6:48
One. Had to do
6:50
inside yourself. And
6:52
get all the talent? Is that that
6:55
that really made me who I am?
6:57
Intensity? And you had your
6:59
math teacher in New Jersey yes
7:01
who is so supportive and continued
7:03
throughout your life to be supportive.
7:06
Yeah. I mean that's is why
7:08
the book is the also a
7:10
tribute to those people like him.
7:12
Pops Abella he is but all
7:14
measures. Such. A. Home
7:17
and. Public high school
7:19
teacher in our country. She
7:22
is not. Unique.
7:24
Yet he's so special, his
7:26
so. He embodies the
7:29
value of that this new country.
7:32
Are instilled in me? that
7:34
kind of. I'm.
7:37
Compassion. A kind of respect to
7:39
the kind of polar and the kind of.
7:42
Generosity. That forever
7:44
you know just left at me
7:46
and I'm so grateful for that.
7:48
And I think of that moment. The.
7:51
Little girl standing outside the door. Hearing
7:54
so be recruiter be discouraging and
7:56
only a few years later, you
7:58
get a full scale. To
8:00
Princeton University. That's. A
8:03
testament to with your ability to
8:05
knock yourself out learning. I
8:07
did work hard but it's hard for
8:09
me to just give credit to myself
8:11
that the book is a love letter
8:13
to the to the people who have
8:16
supported me. Once
8:22
you were launched in this
8:24
pursuit of science, what turned
8:26
you on the Ai. Yeah.
8:28
So hello that? that's a good question.
8:30
So. It really started
8:32
with says I was stuff
8:35
I don't know like soon
8:37
as I was a little
8:39
girl like eleven twelve year
8:41
old, I just loved physics
8:43
A was my first love
8:45
and in hindsight, What? Happened
8:47
is I think I love
8:50
that or thesis quest to
8:52
the unknown mystery of the
8:54
universe like physics allows you
8:56
to ask the craziest clusters
8:59
like beginning of space time,
9:01
boundary of the universe, the
9:03
smallest particle of a matter
9:05
and. Then. In
9:09
the middle prince of physics
9:11
I discovered even the physicists
9:13
themselves like Albert Einstein and
9:15
or when shorting North. They
9:18
turned their attention. Some.
9:20
Was truly equally or basis
9:22
quests and of that sort
9:25
of physics. Question is about
9:27
life and I became so.
9:30
Enlightened than a manner that I
9:32
realized my own audacious class so
9:34
that I love the most is
9:37
what is intelligent. What?
9:39
Makes intelligence. How do we
9:41
build intelligent? missing With and
9:44
that shift. In. The middle
9:46
of end of my college here. Was
9:49
how I discovered a I and
9:51
I didn't know we were in
9:53
a I winter and all I
9:55
I I had no idea. I
9:57
didn't care I it doesn't matter
9:59
that. Indoor spraying. It was
10:01
summer for me cause it's
10:03
a soul fascinating. We are to
10:05
explain for those on haven't come across
10:07
that term said You mean. That.
10:10
He I has had times in our
10:12
history when it's bloomed and when people
10:14
have got discouraged and dropped an interest
10:16
in it is least as far as
10:19
the public is concerned we are huge
10:21
You were talking about are in a
10:23
talk I saw you give about how
10:25
you were living in a town in
10:28
New Jersey. While. Thirty
10:30
miles away. Was sell
10:33
outs. Yeah, where where I live
10:35
in the cones that who was
10:37
when you get out of developing
10:39
neural networks. Yeah he had
10:41
a hand many scientists. I had
10:43
no idea that there will. I
10:46
landed in America a was them
10:48
moment that it. A
10:51
I was little. Is
10:53
it still not downturn?
10:55
yet? Computer scientists like
10:57
silicone. We're. Working
10:59
on their you're Not for It's
11:01
just thirty miles away from my
11:03
whole life American. The Homeless Services.
11:06
So. How did you get from
11:08
that? To. Concentrating on
11:10
images. At naturally
11:12
a visual person because neither
11:15
my early childhood my dad
11:17
takes me to this natural
11:19
excursions and we looked at
11:22
butterflies withdraw the pictures of
11:24
mountains and lowest assassination of
11:27
saying I find that. understanding.
11:30
Visual Intelligence to be the
11:32
most fascinating aspect of intelligence
11:34
with alleged a sculling to
11:37
visit. I've heard you
11:39
say that vision is more than
11:41
just as sense, that it's an
11:43
experience. Vision is
11:46
intelligence. Vision is experience
11:48
this and then our
11:50
understanding of vision is
11:52
is planning business decision
11:54
making. This in the
11:56
that he says this
11:58
is a very. During
12:00
Cornerstone Teeth of Intelligence.
12:03
Itself At one point you
12:05
fascination with images. Cause.
12:08
You. To start to work
12:10
with Image net which had like to
12:12
have you help me understand was better.
12:14
It's important understand because I think that's
12:16
see. That's the
12:18
project. That. Made people call
12:21
you the Godmother of Artificial Intelligence As
12:23
we know it, today seems a way
12:25
to be. I know you know an
12:27
excuse for graduate yourself too much, but
12:29
I've heard that said about you and
12:31
I'm trying to figure out in what
12:33
way. Was it a milestone? But.
12:36
I think it's best explained with
12:39
absolute today's breakthrough in say Ted
12:41
Zip T Base Y is. Is.
12:44
Ugly! Seen that a i
12:46
break fool because we see
12:49
powerful algorithms trained on. A.
12:51
Vast amount of data, other
12:53
data of the internet and
12:56
that gives us such powerful
12:58
breakthroughs. But this concept. Of
13:01
new are not for at
13:03
st on large data was
13:05
nonexistent. Think for. Twenty.
13:08
Twelve basically because I was
13:10
still going through a phase
13:12
that we don't know what's
13:14
the best pass to make
13:17
powerful ai a. Two things
13:19
converged. One is there is
13:21
a group of people like
13:23
Professor Death Sentence who are
13:25
continue to. Continuing to
13:28
explore your network. But.
13:30
There is another ingredient in
13:33
another part of the puzzle
13:35
is. Big. Data is
13:37
at that time nobody saw
13:39
a big data with power
13:42
A I and that's where
13:44
Image that came to plays
13:46
a pivotal role is that
13:48
my students and I recognize
13:50
the power of data. We
13:53
hypothesized am. The I
13:55
guess before most people that. A
13:59
I wo. Have a paradigm
14:01
shift if we power it
14:03
with enormous scale, tie and
14:05
amount of data is isn't
14:08
say it has centric date
14:10
of first an approach and
14:12
because of that we were
14:14
working on vision so we
14:17
wanna make the fittest visual
14:19
dataset. And in
14:21
order to make the biggest
14:23
visual data said we had
14:25
this crazy idea of downloading
14:27
almost all the pacers we
14:29
can get on the internet
14:31
back in two thousand and
14:33
seven and. Organize.
14:36
A curated catalog and
14:38
inmates complete this in
14:40
terms of visual objects
14:42
And that's when we
14:44
made after. Three years.
14:47
Between. Two Thousand and seven To Two Thousand Nine.
14:50
We. Made a dataset of
14:52
fifteen million images across twenty
14:54
two thousand categories. And that's
14:56
out of cleaning up a
14:59
billion images and of. That's.
15:01
What emits and that was
15:04
image that brought the date
15:06
of first approach to a
15:09
I. It also implicitly showed
15:11
the Ai world. A
15:13
important Northstar problem for work
15:15
on which is object recognition
15:17
and also this commitment to big
15:20
data that was a. A
15:22
big intellectual job because. Putting.
15:26
Together, that kind of scale of
15:28
big data was. Was
15:30
part of a was a
15:32
gamble A was. A
15:35
career risking so before one
15:37
does, it is hard to
15:39
know. If. We should do is.
15:42
That's what we have to do. And I
15:45
think you said what inspired you
15:47
a great deal with your progress
15:49
was would you call Pyramids number
15:51
Now. Was. The importance
15:53
of beauty number. That. You
15:55
importance of Biederman number
15:57
is. A keeper.
16:00
A number on the size
16:02
of the visual Intel As
16:04
a flood. How? What is
16:06
a number to define visual
16:08
intelligence? And that is? Really?
16:10
A hard question is almost like
16:13
ask you how many stars are
16:15
out there and done and. For
16:18
someone like me who was try
16:20
know I'm. Feeding crack
16:22
the problem of visual intolerances.
16:25
I was looking for a
16:27
clue to tell me. What?
16:30
Is the scale and scope of the
16:32
problem. idol is if we. If
16:34
we the solve. For.
16:36
Objects. Is that enough? If we
16:39
saw twenty all deaths, Is it
16:41
enough? If we solved a hundred,
16:43
Objects is a enough nobody knows
16:46
thousand. But when I discover the
16:48
Biederman number. I feel like
16:50
I discovered the A. A.
16:53
Really important. I'm.
16:57
Conjecture that no one was
16:59
noticing. Witches. Tens
17:01
of thousands of categories
17:03
and. Be them and put
17:06
it as Thirty Thousand Calgary A. Once
17:08
I got that number I feel like.
17:10
I. Have a clue about the
17:13
size? Of this problem
17:15
same would you mean by
17:17
categories. Of
17:19
the categories are the
17:21
natural way the human
17:23
with conceptual as objects
17:25
may tend to conceptualize
17:27
them as German Shepherds
17:29
microwave, yellow hum of
17:31
a sports car you
17:33
know of course sometimes
17:35
with think about my
17:37
German Shepherd your microwave
17:39
but in general that
17:42
classification of visual concept
17:44
is a fundamental. Visual
17:47
Intelligence. A problem
17:49
that humans have worth down
17:51
and the solved and it's very
17:53
foundation of to our visual
17:55
intelligence. Source you collect. A
17:59
great number. Eight years under the
18:01
category of dog and a great
18:03
number pictures under the category of.
18:06
Cat. So machine
18:08
is able to sort through dead and put
18:10
a name on it. When. It sees
18:13
a picture. Was an online
18:15
Joel There are hundreds of thought
18:17
species selling him as neither father
18:20
died when we actually had hundreds
18:22
of different.terrier German Shepherd Core the
18:24
we have the the different kinds
18:27
of course season so it's a
18:29
lot more than those dog vs.
18:31
it's. Right right? Shows
18:33
you've got subcategories is? well,
18:35
yeah. Totally. I mean a
18:37
lot most of the mets and
18:40
that is the sub categories clay
18:42
right? Like as that hundreds of
18:44
dog has ah hundreds and hundreds
18:47
of birds and have since you
18:49
know many different kind of cars
18:52
will do the buildings trees flower
18:54
of the you know it's a
18:56
It's a very very vast catalogue
18:59
of the visual world. Where
19:01
you got inspired by Biederman number you
19:04
have been working up until then. On.
19:07
A on a hundred and one
19:09
categories which was hard work and
19:11
took a long time. What
19:14
made you think that you. Could.
19:16
Possibly get. Tens. Of
19:18
thousands. Of categories
19:20
in catalogue them. What gave you the
19:22
idea that you could do it? I.
19:26
Would like to say they lose. A
19:30
was like that kid who went
19:32
on a treasure. Sound and sound.
19:35
A clue to the. Biggest.
19:38
Quests, And mystery and allows was
19:40
finding that clue. I was so
19:42
excited and then I got so.
19:45
Soft. By myself, it's like
19:47
it's so much bigger and I have
19:49
no idea how to do that. Accepted
19:52
the in the back of my mind
19:54
because he was. I was getting a
19:56
test with that, the scale and I
19:59
was like what to do and then
20:01
serendipitously. A couple of years later when
20:03
I was at Princeton Temples and heard
20:06
linguists. More. Cataloguing words. They
20:08
were not thinking about the visual world.
20:10
They were not thinking about Ai. But
20:12
they were just. Making.
20:14
Knowledge. Taxonomy of words.
20:18
And. The when when I maple.
20:21
Connection between Beethoven's number
20:23
which is visual concepts
20:25
and knowledge. Taxonomy. I
20:28
think it was a moment where I fell.
20:32
So. Sure,
20:35
That this is important enough I've
20:37
gotta do with. I just
20:39
kind of forgot about. How
20:42
hard it would be. I kind
20:44
of set off. I just
20:46
needed to do it. I don't know
20:48
how open. At it. If
20:51
I don't do it it I
20:53
cannot sleep. I give our so
20:55
so. So. That was. Kind
20:58
of momentum. Armed.
21:00
Robber. And. That's that's how
21:02
I started the Image That project.
21:05
And it took a few years of
21:07
struggle. and good is also a few
21:09
years realizing. I
21:11
was so delusional li. I'm
21:15
fearless. How
21:18
do you make a decision like that?
21:21
What goes into it? Would you know
21:23
you bring your own careers to discover
21:25
how do you weigh the factors? Risk
21:27
again to change the to work. See.
21:32
I assume that Allah, you don't you
21:34
can't because if you do, you wouldn't
21:36
do as. If.
21:38
You go with your gut. You.
21:41
Go with that scientific.
21:45
Converts, And. It it
21:47
it is.at the at the end of the
21:49
day has to go to the.level but it
21:51
wasn't a eve rational. It's not like I
21:54
came up random. Better grades, The
21:56
scientific conviction as and I'm
21:59
seeing that. And I
22:01
have. To make a half and and
22:03
I can be wrong. Honestly
22:05
I did have some too much time thinking
22:07
about what have them fiverr I guess I
22:09
go is to another dry cleaner shop. For.
22:13
Which which is a reference to part of
22:15
your life that was so interesting was you
22:17
were was a while you were getting your
22:19
phd you were we were you working. On
22:22
my entire undergrad at Princeton
22:24
and the first three years
22:26
of my Phd is a
22:28
three years altogether I was
22:30
running the family dry cleaner
22:32
shop business. Which uses up
22:34
and on and on weekends. Yeah,
22:37
every weekend you did it. Work
22:39
with starch. No starch. Yeah,
22:41
actually, that was more more than
22:43
that. I'm an expert. the. Same
22:46
faces. When
22:56
we come back from our break safe
22:58
a lead tells me how the question
23:01
from her mother who was ill in
23:03
a hospital bed at the time helps
23:05
face a lead determine who present passion
23:07
which is to ensure the day I
23:09
developed should watch the clothes or human
23:11
centered way. Just
23:16
a reminder to be clear and visit his
23:18
nonprofit. With. Everything after expenses
23:21
going to the Center for Communicating
23:23
Science at Stony Brook University. Both
23:25
the show in the center dedicated
23:27
to improving the way we connect
23:30
with each other and all the
23:32
ways it influence our lives. You
23:34
can help by becoming a patron
23:36
of clear and visited patriotic.com. At
23:39
the highest tier, you can join a
23:41
monthly sad with me and other patrons,
23:44
and I'll even record voice mail message
23:46
for you either a polite, dignified message
23:48
from me explaining your inability to come
23:50
to the phone, or a slightly snarky
23:52
one where he explained you have no
23:55
interest in talking with anyone at the
23:57
moment. And I'm happy
23:59
to. Ported A Snarky One is by
24:01
far more popular. If
24:04
you'd like to help keep
24:06
the conversation going, about, connecting
24:08
and communicating, join us at
24:10
patriotic.com/clear and Vivid! P.
24:12
A T R E
24:14
O in.com/clear and vivid.
24:17
And thank you. For.
24:19
Delicious meals you can whip up
24:22
with fresh ingredients in chef crafted
24:24
recipes at a price I bet
24:26
you're like say hello to hello
24:29
Fresh. I lane and
24:31
I gave it a try and I'm
24:33
glad we did. The ingredients came to
24:35
our door in a box, the captain
24:37
cold and fresh and any exact right
24:39
portions. Are lean loves to cook
24:41
with. This seemed like something that would be fun to
24:43
do together. And. In a little more
24:45
than twenty minutes, we had a delicious meal on
24:47
the table. And. It was the would
24:49
have been quicker of I wasn't in her ways.
24:52
Are lame. Was impressed as the
24:55
freshness of the ingredients and how
24:57
easy it was and it made
25:00
our taste buds happy. And to
25:02
top it off if you subscribe
25:04
to leave and include a free
25:07
breakfast in every bucks so join
25:09
us go to hellofresh.com/all the free
25:12
and use code. All the free
25:14
for Free Breakfast for life. One
25:16
breakfast item per box. Well subscription
25:19
is acted. That's Free Breakfast for
25:21
Life and hellofresh.com. Slash all
25:23
the free with code. All
25:25
the Free! Hello Fresh America's
25:27
number One meal kit. Every
25:30
year people com with a list of
25:32
resolutions and an attempt to improve themselves,
25:35
but without a curse. many many of
25:37
resolutions fail. But Rosetta Stone offers
25:39
more than encouragement. They offer a financial
25:41
incentive for a limited time get fifty
25:44
percent off Rosetta Stone lifetime membership. Just
25:46
a one time payment gives you full
25:48
access to all of it's twenty five
25:51
different languages. Sounds a lot easier than
25:53
losing ten pounds rent. Visit Rosetta stone.com
25:55
now and save fifty percent. It
26:01
is clear and vivid. And now back
26:03
to my conversation with Safe Elite. She
26:05
tells a story in her book about
26:08
how her mother, who was ill in
26:10
a hospital at the time has the
26:12
question about how a I could help
26:15
ordinary people. And that sparked Safe a
26:17
lead to help create the Institute for
26:19
Human Center Day I'd at Stanford University.
26:24
Of course you know when I wrote
26:26
wrote that story in a punk. I
26:29
guess. I'll highlight at that
26:31
moment, but. In. Reality. I
26:33
think that moment combined with
26:35
the moment in history including
26:37
my career, what I was
26:40
seeing in Silicone Valley's my
26:42
experience in an industry on
26:44
my sabbatical as well as
26:46
my mother taking care of
26:48
my mother and sharing her
26:50
reflection of all this combined
26:53
and com told me who.
26:55
We examine. That.
26:57
Technology I was making re
27:00
examine my role in this
27:02
technology And that's where I
27:04
realized. Ai has.
27:07
Come. Of age. It.
27:09
In a more profound way
27:11
than. What? I.
27:14
Knew when I was entering
27:16
a I. We now have
27:19
a technology that to actually
27:21
makes real human impact. good,
27:23
bad and ugly. And
27:25
it's so important. We
27:27
create of him as a human
27:29
centered framework so that we can
27:32
continue to develop this and deploy
27:34
this and governors in a human
27:36
centered way. What we've
27:38
seen in the Technology movement
27:40
said whispers you to think
27:43
like this So. What? First
27:45
of all like I have had. Decades.
27:48
Of experience taken care of. My.
27:51
Parents. Ailing mom and
27:53
I think it's a very
27:55
grounding experience. I I know
27:57
that I don't have the
28:00
profile of a the leader.
28:02
I'm not a silicone valley
28:04
bro, I'm a woman, I'm
28:07
an immigrant, I'm a caretaker.
28:09
The background mean Humanity A
28:11
grounded me with humility and
28:14
understanding of human dignity, human
28:16
self respect, human compassion, and
28:18
I wanted to see a
28:20
technology helping humans. In the
28:23
meantime, I was. At google.
28:25
And I was very fortunate
28:28
to be leading Google cloud
28:30
units a I am machine
28:33
learning and their I Learned
28:35
this technology is Phyllis is
28:37
impacting or industries We were
28:40
little li working with Japanese
28:42
cucumber farmers and the way
28:44
to fortune five companies and
28:47
every sector of and health
28:49
care to finance to energy
28:52
to entertainment to he commerce
28:54
and seen that sweeping. Impact.
28:57
Made. Me realize. While. The
28:59
is just the beginning We really
29:01
need to. We've. Really
29:03
need to recognize. The.
29:06
Responsibility of this technology:
29:08
rubber hitting the road.
29:10
And we need to think deeply
29:12
about humane packed at that point.
29:14
So what are some of the things
29:17
that Institute for Human centers? A Yard
29:19
Guns were would? What are you working
29:21
on? So. We'll work com
29:23
three things. Research Human
29:25
Center Research. Human. Centered.
29:28
A I Education. And.
29:31
The Human Center A I've policy. So
29:34
in research we are
29:36
the leading institute. A
29:39
cell for as well as in
29:41
academia world that. Gives
29:44
grand and and build
29:46
interdisciplinary ah Ai research
29:49
for example in Ha
29:51
I we have digital
29:53
economy lab studies a
29:56
eyes economic impact we
29:58
have center. On both
30:00
a model stuff, study the A
30:03
March language models and a I
30:05
O N and it's impact. The
30:07
we have neuroscience projects that look
30:10
at how we can combine brain
30:12
five years. As A As A,
30:14
we have a lot of of
30:17
interdisciplinary research. In. Education: We
30:19
put a lot of
30:21
focus on baking ethical
30:23
framework into computer science
30:25
curriculum. We put a
30:28
lot of emphasis on
30:30
educating our civil society
30:32
like congressional staffers, like
30:34
journalists, like business leaders,
30:37
And. Crazy ecosystem of knowledge and
30:39
lane policy. We put a lot
30:41
of the i'm for some. Thought.
30:44
Leadership and Public Discourse
30:46
forum of discussing a
30:48
eyes policy implications and
30:50
also advocating for important
30:52
policy changes. For example:
30:54
Public Sector Investment of
30:57
a I. I get the
30:59
impression that is a big effort on your
31:01
part to make sure that. The
31:03
motivation for working on a
31:06
I developing a I further
31:08
is that it's benefit humanity,
31:10
death, Because is a Tennessee
31:12
I guess. Prove hey I to be
31:14
to be considered something he competes with
31:16
humans. Read it in assisting humans. Yeah,
31:19
this really bothers me because I
31:22
think we need to be very
31:24
clear what our relationship with schools
31:26
are. A eyes a piece of
31:29
tool, it's a very powerful piece
31:31
of to was a humanity has
31:33
had it's struggle with. That
31:36
the relationship between us and
31:38
the tools. But it's important
31:40
to recognize. That we
31:42
should have. The. Narrative We
31:44
should have the agency. In
31:47
responsibly creating the using and
31:49
governing that tool. So this
31:51
thing about. Lead to
31:53
a I compete with us more let
31:56
a I take care of us or
31:58
let a I. The whole
32:00
of is so. Knob. How
32:02
how I see the technology
32:04
is. It's wrong to give
32:06
agency to a I. It's
32:08
important we actually take that
32:10
agency. So people like me
32:12
I'm a technologist. I should
32:14
feel responsible. For. What I
32:17
build. And. Or in
32:19
the meantime, I I hope
32:21
that business leaders also feel
32:23
responsible. I feel I hope
32:25
civil society feels responsible week.
32:27
We have to recognize that
32:30
agency a responsibility. You
32:32
seem to put a lot of emphasis on. Enabling
32:35
the people who create a. Bigger.
32:38
And better he eyes to do
32:40
would resume or not. More.
32:42
Than a not put a
32:45
recognition of importance of serving
32:47
people he said. Catching.
32:49
Home in time. To
32:52
make it before it accelerates out
32:54
of our control. You
32:56
tell me Allah lay I'm definitely hear
32:59
a my own voices more than any
33:01
but. I.
33:05
I do feel. I'm
33:08
on this uphill battle. Of
33:10
trying to communicate a
33:12
I in a responsible way.
33:15
I feel sometimes are
33:17
airwaves is I'm filled
33:19
with the son. Of.
33:22
A. I is gonna dominate as a I
33:25
as well as do things to us. We
33:27
are. Screwed. Because a
33:29
I have club. Blah. Blah blah
33:31
and I really wanna just. Stand.
33:34
On top of the mountain and and
33:36
and and sell that. Let's.
33:39
Recognize a I as a
33:41
tool we collectively made and
33:44
will need to collectively. Apply
33:47
and govern. So. It
33:50
would truly is our own
33:52
responsibility to do this right.
33:54
Not a eyes. Before
34:01
we end our conversation, I wanted
34:03
to ask you that I saw
34:05
in the book repeatedly the importance
34:08
to this whole field of people
34:10
who were born outside the
34:12
United States, the value of
34:14
immigrants to our country in general,
34:16
but certainly to this field. And
34:19
I think your nonprofit organization
34:22
of AI for all is
34:25
working on that. Well,
34:27
a repeated theme in my book
34:30
is the recognition of people of all
34:32
walks of life. That
34:35
includes people of different gender, it
34:38
includes people of all kinds of
34:40
countries' origins. You know, I think
34:43
most of people who have helped
34:45
me come
34:47
from another country. And
34:49
I also believe that if
34:51
we want to make this technology
34:53
human centered, we have to recognize
34:56
the human diversity of
34:58
the makers of this technology, as
35:00
well as who's going to use
35:02
and deploy it. So
35:04
with that mindset, I
35:06
did start co-founded this
35:09
nonprofit called AI for
35:11
All that focuses on
35:13
K-12 education of
35:16
AI and to
35:18
try to lift tomorrow's leaders
35:20
from all walks of life,
35:22
from all backgrounds, whether you're
35:25
in rural America, or if
35:27
you're in inner city America,
35:30
or if you are a young
35:32
woman, or if you're people of
35:34
different color
35:36
and different backgrounds. So
35:38
that's the goal. But
35:41
the bottom line is, if
35:43
AI is going to change all of us
35:45
tomorrow, because it's going to impact
35:47
our world, I want to make sure
35:49
the people who are changing AI represent
35:51
all of us. That's
35:54
well said. And in the process, you're
35:57
creating an unknown number of faith- they
36:00
leave, that'll get
36:02
us there. I
36:04
wish I had more time to talk with you. I
36:08
have hours of things to ask you about
36:10
and learn from. But before we
36:12
end every show, we
36:14
have seven quick questions. Sure.
36:17
Okay. Of all the things there
36:19
are to understand, what do
36:21
you wish you really understood? Of
36:24
all the things in the world? Of all the
36:26
things, not necessarily related to your work, could
36:29
be, whatever it is. My
36:32
children. Ha ha ha ha
36:34
ha ha. Ha ha ha ha. Ha
36:36
ha ha ha. Well,
36:39
remind me of my good friend, Steve
36:41
Strogatz, the mathematician who said he wished
36:43
he understood his dog Murray. But
36:46
yours is even more useful. Well,
36:49
I don't have a puppy. Ha ha ha
36:51
ha ha. So,
36:54
how do you tell someone
36:56
they have their facts wrong? That's
36:59
a great question. I
37:02
wanna say it depends on who they are. But
37:05
I guess I'll start by
37:08
I respectfully disagree. Okay,
37:11
let's go to the next one. What's the
37:13
strangest question anyone has ever asked you? The
37:17
strangest question. Oh,
37:20
I got one that's good. What
37:22
is the favorite category in
37:24
ImageNet? Oh, that's interesting.
37:27
What is the favorite? Is there
37:29
one? It is so hard to answer. I
37:33
would say I constantly take a lot
37:35
of joy in
37:39
browsing the pictures of the
37:41
category of wombats. Ha
37:43
ha ha ha ha. Wait
37:46
a second. Why is
37:49
that? Well, yeah, I
37:51
don't know. I love wombats.
37:54
Ha ha ha ha ha. Ha ha ha ha ha.
37:58
Well, that's not only the strangest question. It's
38:00
the strangest answer I ever got. Okay, I'm glad
38:02
you think that way. How
38:06
do you handle a compulsive talker? I
38:11
stay silent. And
38:13
you stay put. It depends. If
38:16
I'm rushing somewhere else, I need
38:18
to extract myself.
38:23
Okay, let's say you're sitting at a dinner
38:25
table next to someone you don't know, never
38:27
met before. How do you
38:30
begin a really genuine conversation? I
38:34
actually love those opportunities because it's such
38:36
a curious way of learning. I guess
38:38
I would ask what do you feel
38:40
so excited about these days? Hmm,
38:43
yeah, great. Give a right to the emotion. What
38:47
gives you confidence? Humanity.
38:51
This is why as
38:53
an AI scientist, I
38:56
find it strangely, I use
38:58
the word human so much these
39:00
days because humanity
39:03
is flawed. I'm
39:05
going to admit it. But
39:08
we're also incredibly resilient.
39:11
Our arc of history with
39:13
all its ups and downs
39:15
is bent towards goodness. At
39:17
least we want
39:20
to. I believe in Dr. King.
39:24
The arc of history has been towards justice.
39:27
It's not just justice. For
39:29
me, there's
39:31
goodness in humans. So I do
39:34
believe in humanity. I have confidence
39:36
in humanity. Doesn't mean I
39:38
have confidence in every single individual. That
39:43
opens up a whole other conversation. We'll get to
39:45
it sometime when we meet. Last
39:48
question. What book changed
39:50
your life? Great
39:52
question. So
39:55
many, right? The
39:58
nerdy one, I would say. Home and let
40:01
service. Man. That's
40:04
great when I I do wish
40:06
we could talk longer. You're you're
40:08
You're an extraordinary person and mosquito
40:10
things you could. We have you
40:13
on the planet. Of the
40:15
whole were like I said
40:17
I think this moment communicating
40:19
ai part of my worth
40:21
because I think humanity. So.
40:24
Need. On us
40:26
to eat and authenticity in.
40:29
Parking. About a I
40:31
worry. That. The if we
40:33
don't. You for not
40:36
honest about talking about ai we're we're
40:38
not doing justice to to the suicide
40:40
is he took to the. To.
40:42
Our community. And
40:45
a being honest involves
40:47
talking about. The. Great
40:49
benefits it'll probably come
40:51
and also. Find
40:53
that a danger to do harm than
40:56
good com and you don't want to
40:58
emphasize one is a disadvantage of the
41:00
either. Way. And you
41:02
don't wanna be a hyperbolic,
41:04
eliminating nuanced than thoughtful. Which
41:06
means. You're no fun for
41:08
news. Has rotted,
41:10
was a disaster. gets the his
41:13
mind. Was
41:15
made you so much pleasure I
41:17
think. This
41:27
has been cleared vivid and we say hopes
41:29
of. My. Thanks to
41:31
the sponsors this podcast and to
41:33
one of you who support our
41:35
show and patria you keep clear
41:37
and vivid up and running and
41:39
after we pay expenses whatever is
41:41
left over goes to the All
41:43
The Center for Communicating sign and
41:45
said Stony Brook University. So your
41:47
support is contributing to the better
41:49
communication of science. We're very grateful.
41:53
Say. Leaves the inaugural Sequoia
41:55
Professor in the computer Science
41:57
department at Stanford University. In
42:00
Code Director of Stanford's Human Center
42:02
Day I Institute. During
42:05
or sabbatical from Stanford from January
42:07
two thousand and seventeen to September
42:09
two thousand and eighteen, she was
42:11
vice president at Google. And
42:13
served as chief scientist if Google
42:15
Cloud. For development
42:17
of Image Net ushered in
42:19
the age of big data
42:22
in a laying the foundation
42:24
for sad but slight said
42:26
Gp teased it's successors a
42:28
wonderful book is the world's
42:30
I see curiosity, exploration and
42:32
discovery. At the dawn of a
42:34
i. This. Episode
42:36
was edited in produced by
42:38
our executive producer Graham Said
42:41
with help from our associate
42:43
producer James who may or
42:45
publicist is Sarah Hill a
42:47
researcher is Elizabeth. though Haney
42:49
and the sound engineers Eric
42:51
have won. The music is
42:53
courtesy of the Stefan Thirty
42:55
trio. Next
43:06
in our series of conversations I
43:08
talk with Eric Schmidt being the
43:10
Ceo of Google from two thousand
43:12
and One to two Thousand Eleven
43:14
followed by four years is Googles
43:16
executive chairman has given him a
43:18
birds eye view of the development
43:20
of Ai. He has many concerns
43:22
about the harm the day I
43:24
said cause he's especially worried about
43:26
a I generated deep face but
43:28
he also sees benefits, benefits the
43:30
today we can scarcely imagine. For
43:33
instance, while chat bots work by
43:36
predicting the next words, they are
43:38
many, many examples. Were predicting
43:40
the next word is also a technique
43:42
the to the news to protect the
43:44
next. Zoom. The next
43:47
protein and it uses the same
43:49
principles. So. What does this
43:51
mean? How bout. Better batteries.
43:54
How about more efficient
43:56
energy distribution? How about
43:58
better carbon? Management. Climate.
44:01
Change alone one of the greatest,
44:03
a dangerous to humanity in the
44:05
long run will be materially improved
44:07
by this plastic so paints a
44:10
pollutant pollutants of one kind or
44:12
another. We're going to look back
44:14
on this period and say we
44:16
were so ignorant because we were
44:18
using said simple. Materials.
44:21
Components are so forth and are
44:23
built existence and this is how
44:25
progress goes on. It's great. And
44:28
all of these are happening at
44:30
a speed that is incomprehensibly fast
44:33
compared to what it was twenty
44:35
years ago, thirty years. Eric
44:38
Schmidt on the good as well as
44:40
the bad and ugly of ai. Next
44:43
time on we're in Vivid. For
44:46
more details about Clear and Vivid
44:48
into sign up for my newsletter
44:51
please visit alanalda.com and you can
44:53
also find us on Facebook and
44:55
Instagram at Clearing Vivid. Thanks.
44:58
For listening five miles. Welcomes.
45:13
Year Twenty Twenty Three Work recap This
45:15
year you've been to one hundred and
45:18
Twenty Seven think meetings use density thickness,
45:20
searching for files and almost missed a
45:22
deadline. He added. Twenty
45:25
Twenty Four Ten. And so it
45:27
sounds. difference with money.com you can
45:29
work together easily, collaborate and share
45:31
data files. And I say though,
45:33
all work happens in one place
45:35
on the same page monday.com or
45:38
Tax Advantage to learn. More. jokes
45:44
next round as bad start and already yeah
45:47
yeah just shutting for a car car of
45:49
honor for real yeah come on a makes
45:51
it super conveniences shop whenever wherever for real
45:53
that's a ton of car options yep and
45:56
israel with my price range or really real
45:58
you can afford that with carbon And boom!
46:00
Just like that, I'm getting it delivered in
46:02
a couple days. For really, really real? You
46:05
just bought a car! For
46:07
real, and you just lost, my turn.
46:09
Visit carbona.com to shop for thousands of vehicles
46:11
under $20,000.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More