A. Richard Newton Distinguished Innovator Lecture Series – Vivienne Ming


(introspective music) – Good evening everyone, today’s
speaker is Vivienne Ming. She’s a theoretical neuroscientist, technologist, entrepreneur and author, so we’ll be hearing from a lot. She cofounded Socos where machine learning and cognitive neuroscience are combined to maximize students’ life outcomes. She’s also developed a
predictive model for diabetes to better manage glucose levels and systems to predict manic
episodes in bipolar sufferers. This was actually one of the reasons that I wanted to introduce her because I’ve always been interested in seeing how technology can
do more aspects of social good, so I thought it was really cool that she was applying it to medicine. Previously, Vivienne
was a visiting scholar at UC Berkeley’s Redwood Center for Theoretical Neuroscience, pursuing her research in
cognitive neuroprosthetics, and she also spoke at
TEDxBerkeley in 2017, so you might already be familiar with her. So, named as one of 10 women
to watch in tech in 2013 by Ink Magazine, please
welcome Vivienne Ming. (audience applauds) – All right. This’ll be fun, if anyone, just pure curiosity, has anyone ever actually seen me talk before? My apologies, we won’t be
talking about the same things, but it will be the same stupid jokes because that’s how I roll. But I give a lot of talks, they used to be academic talks because I used to be at the Redwood Center here at UC Berkeley, and before that I was at Stanford and
CMU getting my bona fides done in publishing papers
and doing presentations at NIPS, and I love being an academic. I think that if everything
else in my life blew up, I could come right back here,
get my lab set up again, and I’d be thrilled. The thing is it turns out I
don’t really need to choose, I get to do whatever the hell I want to, which is pretty much the
coolest job in the world. I’m a professional mad scientist, I just get to come up with crazy ideas and try ’em out and see if they work. Who pays for ’em? Well, variety of people,
but if no one else will, I get to pay for ’em, ’cause
my life’s been good to me. Do we commercialize them? Sometimes, but usually
we just give ’em away, because I don’t need the money, and the world’s better as a result. In the end, when I engage in a project, I don’t think, is this
a research experiment or is this a piece of public policy, am I doing something philanthropic, am I starting a company? All I think is, will this
make the world a better place? And it turns out in a great many cases it really is just me by
myself doing my own thing, and a lot of other people look at it, the work that I do, and think
that it’s weird and bizarre, but the coolest thing in the world happens as a result of me
being weird and bizarre, which is people are alive. I get to measure the impact of my days in how many people are alive at the end. Does make my days kinda
hard to keep track of, I just flew back from Cape Town, I was giving a talk down in South Africa, it’s my tenth visit there
in the last like five years. It’s about a 60 hour round trip, 22 hours flight to London
or Schiphol in Amsterdam and then another 11 hours
and then 11 hours down to Johannesburg or Cape
Town, then 11 hours back and 11, plus you hang out, I know where all the best plugs in
Heathrow Airport are. And it’s not exactly a hard life, right? I get to fly to one of
the most beautiful places in the world from one of
the most beautiful places in the world and give a talk
to a whole bunch of rich people that for whatever reason are interested in what I have to say. I fly on a bed that folds out on a plane and they give me fancy meals. I really shouldn’t complain about my life, turns out it’s one of the
few things I’m really good at is complaining, so I
complain about it anyways, but my life is really pretty amazing. And everyone wants to know
though when I fly there, so this most recent trip I flew down and I was there for 18
hours and then I flew back. So a 60 hour trip, 18 hours on the ground. And the reason’s really simple for me, I don’t have any, I gave my talk, I met with the Mandela Foundation and I met with Napster’s foundation, there’s a newspaper in South
Africa that invested in Alibaba when it was a pipe dream and so now they’re a multibillion dollar business, and we’ve schemed up some
crazy ideas of things to try, and then I was done. And I don’t really care
about going to a museum or going to the top of Table
Mountain or going on safari. I mean they’re very great, if you ever have a chance, go do it. I got back on the plane
so I could get back here and either be with my family
or get back to my work because that’s what I love to do. So I could take this talk
in any number of ways, I have some prepared slides, but as the people of Cape Town discovered, actually it was YPO, so the people from all around the world,
sort of obscenely wealthy, it’s interesting YPO standing for Youth Presidents Organization, I couldn’t have told
you that five years ago. Nobody there is young. Turns out CEOs don’t tend to be young. Want to know, this is actual research, want to know what the, if
I took photos of people and I had you all rate
them for competence. Id just take out random photos, you get 500 milliseconds, is
this person competent or not? Bam, bam, bam. You’re actually pretty good at it from the perspective that you agree about what makes a competent face. And if I took those
ratings of those photos and unbeknownst to you some of those people were
CEOs and some of them weren’t, your ratings of competence
are very predictive of who is a CEO and who wasn’t. And amongst the people that are CEOs, some are CEOs of large public companies, and some are CEOs of small private ones, and your ratings of their competence on a half of a second judgment is predictive of who’s the
CEO of a big public company and who’s a CEO of a small private one. Interestingly enough, it
is not remotely predictive of the performance of those businesses. So if you didn’t know, the
executive recruiting industry is a multi, multibillion dollar industry and we just explained 70% of its variance in a 500 millisecond decision by just random people on the street. If you want to know, by the way, if you run this experiment
in the United States, what that face looks like
that gets the highest ratings, it’s an older white male. It’s maybe not so surprising. Again, it probably doesn’t correlate with anything we might call competence, but it is highly predictive
where people go in their lives. So I flew down to South
Africa and then I gave a talk, I don’t even remember what it was about ’cause I just tossed out the slides and I made up a talk on the spot. Turns out this is a lot more fun than giving a talk, some prepared talk. It’s interesting that you
mentioned TEDxBerkeley. I gave an actual TED Talk once, the only prepared talk
that I’ve ever given. I had 10 minutes and like you
have to give them a script and stick to it and you do the performance and you gesture, like you
practice in front of a camera, just ridiculous. So I went off script and they
were not happy about that. But it was a lot of fun, it’s a lot of fun to get up and explain to people something about the world that
they don’t understand. And I get to do it all the time. What’s interesting is I don’t have a social skill in my body, I have no instinct for networking, I am decidedly curmudgeonly, you may not want to ask me
my opinion about blockchain given who the speaker is next week, but I will say it is the
one technology in world that I know of that is
most desperately in need of an actual problem to solve. But it’s fascinating
to be in this position to make a difference in people’s lives, to have that be the deciding factor in how you go through life. What am I gonna do today? What can I do so that the most number of people will be alive at the other end? It’s pretty fascinating. Don’t mean to not sound like it’s gonna be a talk about entrepreneurship, but it turns out along the way
I’ve founded five companies. Actually this guy at
YPO didn’t see my talk, but took me aside to have lunch afterwards and was just begging me to give
capitalism a second chance, ’cause I think he’d heard
that I gave this talk about doing good in the world and therefore assumed that I was some sort of dyed in the wool socialist that wanted to tear
down the banking system. A rather bizarre
assumption, but it turns out people are full of
assumptions all the time. So I’ve founded five companies, I’ve been the founding
executive at several others. One of them didn’t go anywhere, but the others we either sold
or are still alive today. Turns out I don’t really
like running companies, I like solving problems. So I take them to the point
where a problem is solved and then I hire actual
adults to come run the thing so that I can go do
more crazy mad science, ’cause that’s what I love to do. And, in fact, about two years ago I started thinking hard that I have built these companies
and I’m proud of what they do, but it takes time and I
feel like I’m not maximizing my impact by building a
company, so what should I do? And a friend of mine said, well, you do all of these
fascinating little projects, what if you founded like a
think tank or an incubator where you just take your
mad science projects and you train other people how to do ’em and you just do these things? And then to paraphrase her and put it in a South African context, charge them like a wounded buffalo. So I decided that I
would take the first part of the advice and not the second, we founded this thing called Socos Labs, and that’s what we do. We get calls from, my child
has a life threatening allergy, please save their life. My child has 500 seizures a
day, please save their life. Or the secretary general of
the UN calls up and says, our donors never give the
full amount that they promise, what could you do to get them
to make their commitments? Or Ericson calls me up one day and says, we’re sponsoring this UN project
to reunite orphan refugees with their extended family members, could you do something with that? And we do, in every case
we take those projects, regardless of whether there’s any budget or whatever a VC has said, or
even if there’s a good reason to think we could possibly
make a difference, and we run our mad science experiments. And if we knew it would
work it wouldn’t be science, and if anyone else was
doing it it wouldn’t be mad, so we’re always way out there on the edge, and yet literally millions
of people are alive. And in almost every case
I pay for everything and we just give it away. I’m an intellectual venture capitalist, I’m investing in a future
that I want to be a part of, that I want my kids to grow up in. Both, I have an 11-year-old,
who you’re gonna hear a little bit about and a seven-year-old, and they’re amazing and they’re awesome. In some ways it’s a little controversial to talk about in other parts of the world, but one of the cool is they’re mixed race and here in Berkeley that’s
the single largest ethnic group and it’s cool, I love that. Whereas in other places I
would be decidedly out of place even though I get to travel
around the world and be me, but I’m kind of a weirdo,
whereas it isn’t always the case for people that are
different that they feel like they can make an impact on the world, that they can actually
offer something back. So I’ve rambled enough as an introduction, I’m just trying to do
a little level setting and give you a sense of what I do. If you want to know one
other thing about me that would be informative,
it would be that, again, I get to give
lots and lots of talks. The talk which was most fun for me ever was to actually give a talk at
a science fiction conference. I love science fiction. Amongst the various books
that I’m writing right now, there are now two screenplays and a piece of science fiction. Actually someone made a short film about my neuroscience work, and I never read anything about myself. I never read news articles,
I never watch my talks or interviews with me,
nothing, nothing, nothing. It’s just there’s nothing but
madness in that direction, you end up like Elon Musk. So I instead, so I’d never seen this, but this film festival calls me up and said, would you be willing
to come to the festival and talk about this film? So I said sure, I’m never gonna watch it, but I’ll come talk, I mean
it’s my work, I know what I do. So I showed up and in two sessions answered people’s questions
about a film I’ve never seen. And after one of them, this
guy walks up to me afterwards and says, so, Dr. Ming, have
you ever consulted on movies? And I’m a native Californian
so I know what he means. As far as I can tell,
in fact, the only reason to be faculty member at UCLA
or USC is to consult on movies, so I said I’ve never
done it, but I’d be happy to help if there was something I could do, and he said, oh, well, my
name is Darren Aronofsky and I’d really like it
if you could help me with some movies I’m working on. Not exactly the direction
I thought my life would go, but we showed up at his production office a couple months later in Brooklyn and he just says, so, I’ve got a, forget the movies, I’ve got an HBO deal, what do you want to do? So life gets weird. And the funny thing about
that is it was completely out of the blue. And you might think random
chance, this is how life works, and it does. One of my favorite papers of all time goes by the title of An
Experimental Analysis of Inequality and Unpredictability in Artificial Online
Markets, something like that, it’s good enough that you could Google it. So this was a science paper
from several years ago, and what they did is they paired with an actual new music site, and unbeknownst to the
people visiting the site, they were signed up for
one of nine virtual worlds. In the first world you
showed up at the site and you saw a random array
of songs to choose from with no ratings from
anyone, you picked a song and you gave it your own rating, and then you picked
another song, moved on. And this was new music, I
mean like you’ve never heard of the bands, much less the songs. Then in the other eight virtual worlds you saw the songs ordered by
and with the average rating of the people in your virtual world. Again, you didn’t know that
that’s what was going on, you thought it was everyone’s rating. So they found three things. First, there really is good and bad music. So the people in the independent world where they didn’t see anyone’s rating, if three people thought a song was good they gave it a seven, which
would be a very high score in that world. The fourth person probably
gives it a high score as well. Similarly, low scores, threes and fours. It’s not a huge effect,
but there’s definitely this nice linear good/bad music trend. And people agree independently. Soon as you see other people’s scores, finding two, bam, big nonlinear effects, the best songs become tens, which means, of course, all
the variance disappears, everyone’s giving it a 10. The worst songs become zeros. And the stuff in the middle,
people stop listening to. So you get this big flat nonlinear effect. Eh, whatever, so people are weird and we’re subject to social, there was just a paper recently showing that about 12 to 13 years old, literally our perceptions
of the world change based on what people around us say. You know the, well, is this dress green or yellow or whatever,
literal perception of color changes as a result of the
opinions of people around you. So we’re not so shocked that people are changing their ratings, but I do fancy-schmancy
machine learning stuff, so it’s a nonlinear
system, surely it preserves the cardinality of the ratings, we could just squash that back to its original scores, right? But here’s the actual
finding of the paper, which is that across
those eight virtual worlds we had all those different
people interacting, no song that got the
nonlinear boost in one world got it in any other. In other words, if you reran reality, then, well, I guess if you reran reality it might be Michael
Jackson wasn’t a pedophile, but the other thing is
he also was an also-ran who had some good songs in the 70s and then we never heard from him again. Famous movie star, the economic
outcomes of large countries, the performance of musicians
or writers or even scientists, I guarantee you citations for scientists should follow this same effect, you have to be good to be great, after that it’s a little
bit of a coin toss. So maybe that was all
that was going on with me, some famous Hollywood director
just happened to be there and I just happened to be there, and he liked my rambling nonsense. He is, in fact, well versed
at the artistic expression of rambling nonsense, and so, wow, we’re totally in synergy and he comes in and asks if I want to work on something. But it turns out it’s not chance. Part of what I’m gonna argue here today is I do my work for a purpose, that purpose has nothing to do with whether a VC will fund my companies, it has nothing to do with whether
it will make me any money, it has nothing to do with
what other people think of me. Not that I don’t care, but
I make a supreme effort never to even find out. It has to do with whether I’m
doing what I think is right. And one of the great paradox of purpose is that the people that
follow a strong purpose actually end up earning more
money, they live longer, they walk faster when they’re 65, they have more friends,
doesn’t apply here, but they literally live longer. So it’s kind of funny because I was able to analyze the behavioral
measures of purpose. So I use artificial intelligence
to observe people doing whatever they’re doing, we call
it naturalistic assessment. I don’t like running surveys
or doing experiments, it’s fine, but don’t
really have to nowadays. One of those NIPS papers I mentioned, we built what we called listen bots, and we listened to university students talk to each other online, pardon me, that was gross. And, (lip smacking) gyro. We listened to them
online and it turns out we knew at week one what
grade they would get just from them talking
about the course material. By week three we were
outperforming the final exam in understanding how they’d
perform in the class. So it was cool, and, again, it
was a nice paper to publish, but in doing this sort of work, we discovered this
really interesting thing. You can measure stuff
that feels unmeasurable. Does someone have a
strong sense of purpose, is someone highly resilient? Do they have good
perspective taking skills or analogical reasoning? Turns out we could build models that gave us hard numbers
around that sort of thing just by observing people
living their lives. You want to know the single
best behavioral measure of purpose, is sacrifice. If I see evidence of
you making a sacrifice for something that is bigger than you. A sacrifice which you will
never directly recoup from, that is something that’s gonna take longer than a lifetime to play
out, you make a sacrifice. So think about the paradox
from that perspective, purpose is defined by
people taking a step back, and yet over the course of their life they end up farther forward. I find myself in front of
a whole lot of audiences that are obsessed with how
much money they’re making. They’re obsessed with out
competing their classmates, the person sitting next to them, the person across the street,
the startup in their space. Or my opinion is your
startup’s better than mine, how can I help? All I care about is solving the problem. I can’t say that my VCs
have always been thrilled with that attitude, but it turns out I’ve been a lot more successful
than most people have, and I think it’s fundamentally because I’m not doing it for me. So anyway, at the end
of this I’ll tell you sort of how I stumbled
into this perspective apart from the sort of
sciency talk around it, and instead maybe I’ll launch into some of the slides I was talking
about to justify my existence. All right, you already
know a little bit about me, probably more than you ever wanted to, and I apologize for that, I have to say the thing
that I’m most interested in is neuroprosthetics,
kinda literally jam things in your brain and make you smarter, and the answer’s unambiguously yes. So if you want to volunteer, go for it, I think it would be awesome. The survival rates
aren’t super duper high, but for those of you that do survive I will have a data point. By the way, the CRISPR
thing, it is unbelievable, did anyone know that the
gene that they edited has to do with HIV and that’s the story, turns out it’s also well known that it is related to
cognitive development. I mean really well known that that gene’s related
to cognitive development. No one was editing for HIV. Want to know why people are interested in genetic engineering is the same reason I’m interested in neuroprosthetics, I want to make smarter people, ultimately. Why do you think the two biggest funders in space are DARPA and
the Chinese military? But I love this sort of stuff. I love it not because
I’m completely devoid of any care about how it gets applied, I love it because I want
to live in the world where it gets used correctly. So here’s just a glance
at a couple of projects we’re working on,
instead of really talking in any detail about them, I’m gonna talk about a
couple of new projects I’m working on right now, and you can think about how
they might get commercialized. I admit I don’t particularly care, but that’s because I am now
at a certain point in life that I get to be very indulgent,
but when I got my start, I had to care about all
this stuff back then. But just to cover a
little bit, for example, that heat map of Manhattan,
that was the bipolar project, was just to give you a sense of where each of these came
from, total serendipity. One evening I was meeting with someone near Madison Square Garden,
she took me out to dinner, she was pitching me a startup idea. I don’t even really remember what it was, I get pitched lots of ideas. Like people think I’m gonna
invest in their companies. I donate all of my money every year, I don’t have anything to
invest in anyone’s company. Besides, it always fails. You have to hit such a phenomenal home run for an early small scale investor to actually make any
meaningful money back. But we had this fun conversation, and at the end she was
walking out the door and she says back over her shoulder, you know what, I would forget it all if you could figure
out how to fix bipolar. I don’t know why, I
don’t know the context, is it her, is it a parent, a loved one? I have no idea. All I know is I was hooked. So I didn’t go to sleep that night, I spent the whole night reading papers about manic depression, bipolar disorder. And I, you know, I’m a neuroscientist, but I’m not that kind of neuroscientist. The whole reason I became a
theoretical neuroscientist, if you don’t know what that is, we build machine learning
to study the brain and we come up with better algorithms by studying the brain. The real reason we do it
is because brains are messy and bloody and icky and
it’s just really cooler to build a fake brain on a computer. So we think of all of the
actual neuroscientists as being stupid and they
think we’re all lazy, and we’re both kind of right. So I didn’t know that much detail, I certainly don’t know a
lot about neuropharmacology, so I start studying and I came
across this paper that said, hey, you know if you take people with the genetic predisposition
for manic depression but they’ve never had an episode, they are not symptomatic at all, and you bring them into
a clinic every week and run a big battery of tests on them, things like a tilt table test, they’re meant to trigger responses. You can see behavioral evidence of their first ever manic episode six to eight weeks before it happens. And I thought, wow, that was
really interesting, right? Remember, I make these
naturalistic assessment systems, could I build something into a phone just randomly for all of you and if you were headed
towards a manic episode it could say something? Well, then serendipity stepped in. The next day I was being
pitched by another startup, in this case funded and everything, and they were asking
me to join their board. And they were building a
system to use mobile phones to passively track people’s
mood states in real time. And you can imagine this
was, I don’t know what, seven years ago and back
then maybe we could sell ads against it, maybe we could
play music against it, also two things that I
really don’t care about, but the one thing I can
say is I have zero interest in selling ads to anybody. But I was intrigued. I talked them into focusing
on market research, at least it seems like
an interesting field of doing some study, you know, how does this person feel
before they got on a plane to Cape Town, how did
they feel when they get off the plane to Cape Town,
and how much of a difference did our experience make? And so I did this and then I said, you know what, I don’t want
a piece of the company, ’cause, again, it’s never gonna pay off, it’s nothing against them
and they were super nice and motivated guys, but
startups just don’t pay off. Please don’t build a startup
’cause you want to get rich, it’s like the lottery, that’s not a particularly
rational reason to do a startup. But I was intrigued, so
I said here is my ask, it’s not about money, you’re
gonna give me all of your data and we’re gonna build a system
to predict manic episodes. And, in fact, I was right, the
company went out of business, they could never quite figure out how to monetize what they’d built, but we did build that system. It could passively,
particularly based on the GPS, gyroscope, accelerometer movement data, could predict manic episodes
by three to four weeks. 25% of severe sufferers will
go on to kill themselves, and three or four weeks
beforehand we could let them know, we could let their doctor
know, we could let their wife or their sister or a child know, hey, this person is heading
towards a manic episode. What could you do about that? It’s mad science, nobody knows ’cause no one’s ever done it before. We’re still figuring that out. And we gave it away, I
think Johnson & Johnson is building an advanced
version of that for a system. All I know is if nothing
else, you could get a note three weeks ahead of time saying your brother has a high
probability of a manic episode in the next three weeks,
maybe he should take some time off from work or
check in with his doctor, instead of a note taped to the front door that says, please call
911 and don’t come inside, I don’t want you to see me this way, which is the way most siblings find out. Wouldn’t that be better? So we built this thing
and it still exists, the company doesn’t, but that does. It did way more good than
a company doing mood-based market research ever would have, and I never made a dime off of it other than occasionally giving talks and then I give all
that money away anyways, but the world’s better for that work, and that’s actually all that matters. So that’s one, you can see
the woman smiling there, it’s going in real
time, every one of those has a good story behind
it, so I’m gonna indulge. That was my undergraduate,
it was my first ever machine learning project. I was an undergrad at UC San Diego, I’m actually going to tell
that story in a little bit, but I thought I was, back then I thought was gonna be a wet neuroscientist, you know, stick wires into cat brains. And took my first ever and
only programming course and the professor recommended me to be a lab assistant at this place called the Machine Perception Lab that had a contract with CIA
to do real-time lie detection off of raw video. So it’s a little morally
gray, but it was cool as hell. We got to build these things, and they learned on
their own, 20 years ago, way before deep neural networks. And actually Jeff Heaton got
his start at UC San Diego. So we did this thing and
it was pretty difficult 20 years ago, but since then I’ve got to spin off a bunch of other projects. So, for example, the one right next to it, this faces, what you’re seeing there actually is a early prototype version of a deep neural network. It’s learning not just
about face categories, but about how people perceive faces. We built this sleazy little
game called Sexy Face and people would play
it, and our promise was for free we’ll find everyone on Facebook that thinks you’re sexy, and for $5 we’ll find everyone who
thinks you’re sexy, whatever, you got what I was saying there. It was all a scam, which
is to say a Trojan, people would play the game
and then it would confess, no, we’re not actually truly awful people, it’s just a mind reading
game, keep playing. And they played they
trained up the model for us. We tapped into this
atrocious human behavior, and they then trained our model, and we were able to use it
to reunite orphan refugees with extended family members by analyzing the UN’s database of a
million faces of little kids lost in refugee camps around the world, based on terrible, sleazy human behavior and funding from the CIA, and then that lab actually got bought. So they spun off as a startup and could never quite find their stride. This is one of the things
I’ve certainly learned about startups, don’t
build a platform company. Nobody wants to opt-in
to your startup platform. A platform is valuable
because it’s a platform. A platform is valuable ’cause
a billion people are using it or you have major corporate customers. Your startup idea as a platform
is never gonna go anywhere. So they were trying to kick around, what is emotion recognition
as a platform, as a product, and they couldn’t quite get it. Deep neural networks come along, their performance jumps way up, and then they have this clever idea. They did a real-time
facial expression analysis of one of the like 2015
Republican presidential debates when there’s like 157 people up on stage and everyone’s wondering whether Ted Cruz actually has human emotions. And so they ran this in real time, simulcast on YouTube an
expression recognition analysis of the candidates while
they’re doing their thing, literally the next day
the bidding war starts between Facebook and Apple. So if anyone has an iPhone
10, all that face recognition technology, that’s my old lab. And the Animojis, which are not up there, but, you know, you talk into the phone and it like animates a
cat on the other end. There’s an important
lesson, we were talking about doing that 20 years ago. If you think you’ve got
a new idea, you don’t, it’s about the execution of
the idea that really matters. But the funny thing is,
the lesson that tells about innovation, people
think about innovation driving the world, innovation
changing, innovation for good. The real rule you should
take away from all of that is that tens of millions
of dollars of CIA funding, 20 years worth of research,
dozens of dissertations, have ended up animating
cats on your phone. In fact, that’s what all
innovation is in the end is animating cats on the
internet or on your phone. I don’t care what you’re
doing, it could be blockchain, God forbid when the cats
get into blockchain, it will be a mess. So we get to do all these cool things, there’s some neuroprosthetics
over there on the side, the braid equalizer is a
pretty crazed idea out there that I’ll skip over. We run projects in
education, in workforce, health and mental health, in inclusion, anything involving people. We get a lot of fancy-schmancy job offers, Uber once pinged me about
being their chief scientist, took me about three
seconds to say hell no. (audience laughs) Back then Travis was still involved and the place was a cultural nightmare. And, listen, I’m pretty clear why they would have wanted to hire me as one of their senior executives. The one that was fun that you already know a little bit about me,
director of algorithms at Netflix would’ve been cool, you know, getting to design the algorithms that decide who stars in their shows and what scripts to invest
in and things like that, that would’ve been awesome. But it took me three
minutes to think hard, how am I maximizing human potential, how am I making other
people’s lives better by recommending movies? Had to pass. The one job invitation let’s call it, that I’ve ever considered, Amazon wrote me and said, Dr. Ming, in seven years we’re gonna be a one
million person company, and your job will be to
make their life better. That was it, that was their pitch. I would be the chief scientist for people, Amazon as my mad science
lab to do whatever I wanted to make their lives better. So for the first time ever
I actually went and visited. I did a lot of collaborative stuff with a lot of big tech companies, but I’ve never actually
seriously considered a job with any of them,
but I flew up to Seattle, met with everybody, and pretty quickly came to understand that
Jeff’s definition of better is different than mine. Now, I don’t think he’s a bad guy. I think his definition of better is based on a core idea
that people don’t change. My fundamental belief, driven
by some massive data sets, is not only can everyone change, everyone’s already
amazing, they just vanish, only a few of us will
ever get to live a life that allows us to actually express that. So when I look at the opportunity to take a million people’s
lives to make ’em better, I looked at it as an opportunity to grab a million people
or maybe if we split it up, think about the logistics operation, the warehousing and so
forth, about 800,000 of those people would probably be on that side of the business. So that’s 800,000 people for two years that I get a chance to grab them and pull them into the creative economy. You know, to say, hey, this
is how you are amazing, and I’m sure you didn’t expect to spend the rest of your life
working in a warehouse, and guess what, Amazon
doesn’t even want you more than two years
working in their warehouse. Literally one of my jobs would’ve to make them happily want
to leave after two years, which I’m quite confident I could’ve done, but I have zero interest in. What a waste, what a
waste of human potential. What a waste of a world
that is so fully connected that not tapping into the
most that people can give means you’re robbing from yourself. But if Jeff is right
and people can’t change, then they’re taking a rational approach. Why be stuck in a job you’re gonna hate. They’re like virulently anti-union, I’m meeting with their
director of research and algorithms there and
he’s like halfway through, we’re talking about some NIPS papers and he’s like, so you know we
don’t really like unions here? I’m like, what the, what
were we just talking about? You can have your own
feelings about unions, but that was a bit of a strange
break in the conversation. So in the end I didn’t say no, I said, listen, if I can
make a million people’s lives better, I’ll do it for free. I don’t really need the money. Call me a fellow, give
me all the same data, and I will do what I think is right only in the service of Amazon. Will publish in papers, but leave bold stuff only for Amazon, and they turned me down. It’s not really that surprising, there’s only one job
description at Amazon, it’s do what Jeff says. And clearly that’s worked
pretty well for them, he probably should’ve
taken that advice recently. I’m completely at a loss of why anyone wants to take a photo of their
junk and sent it to anybody, but has anyone ever been happy to receive such a photograph, ever? So we ended up walking away from that. And it sounds crazy,
how could you walk away? That Uber job, good God,
do you know how much money? Now they’re about to go public, I don’t know how much I would’ve
come away with from there, but it’s reasonable to
assume it would involve lots and lots of zeros. Took me three seconds, I
have zero regrets about it. The net result of someone’s life isn’t how much money they make, go visit Monaco sometime. If you’re like me, just my least
favorite city in the world, gambling and all of the horriblenesses of the shallow signs of wealth. If anything, I think Monaco
should sue Donald Trump for devaluing the brand value
of gold, it’s just obscene. So this isn’t me, I have to
do what I think is right, not because I have something
against capitalism, but because that’s actually
what makes me successful. The reason my life is so good, the reason I didn’t need the money is because I was willing
to walk away from it, because had I ever
taken one of those jobs, that would’ve been the end. I know it, I know that it’s true. I’m not being courageous
other than understanding that being me is the
single best thing I can do, even for me. So I get to do all these cool projects. We have a couple of new ones where, keep your fingers crossed, this is taking astonishingly large time
to get the board to agree, but we are hopefully working
the Make-A-Wish Foundation. We’re building a machine
learning based system that will help Make-A-Wish help
the kids make better wishes, as measured by survival rates. Turns out granting a wish to a dying child literally increases their
likelihood of survival. So there’s a couple papers out about that, and we thought if that’s true, then if we could use some
of the cognitive modeling techniques that we’ve developed, then we should be able to
target the right kind of wish. You want to go to Disneyland? Awesome, it’s the number one wish and the number one sponsor of Make-A-Wish. What if you brought your
three best friends with you? So there’s this need evidence,
it looks like for certain, not everyone, but for
certain kinds of kids, making a wish more social
increases their survival rates. You want to meet Chris
Evans, great, he’s hunky, but how about you and Captain America go fight that crime together? So as you probably, I don’t know, you were all like 12 years old, but Batkid here in San Francisco several years ago, there
was another one in Seattle, where these little kids went
out and actually fought crime, and the reporters wrote up the stories like they really happened,
it was just amazing. By the way, you can’t go to
a Make-A-Wish board meeting and come out either with a
human soul or without tears, it’s one or the other, it’s brutal, but it’s worthwhile. So we’re building this thing, we have the full unanimous support of their medical advisory
board, I’ll pay for everything. Amazingly they still haven’t
approved the project, ’cause I just don’t
think that they believe that someone will do it for free. So we’ll just, we’ll keep working on it and we’ll have it ready,
when they’re ready, we’ll give it to ’em. We’re working on another project, so right there, this guy, it’s actually was just over at Skydeck, it’s
a small startup called HUMM, originally based in Perth, talk
about the middle of nowhere, and they actually came out here and they’re building a
cognitive neuroprosthetic, a working memory neuroprosthetic. So a lot of my work involves
looking at the factors that predict the best life outcomes. So I’ve founded a couple of
educational technology companies and I mentioned one of them we were able to outperform final exams
and standardized tests, in fact, that was the
whole point of the company, it was, let’s end all standardized tests. We know they’re biased, we know they don’t actually predict much
other than who goes on to the next level, by design. So what if we could just obviate all of it and turn learning itself
into the assessment? Great, everyone loved
it, I mean we gave talks at the Department of Education. In fact, even ACT and the College Board came and invited me to give talks. The reason they were so blase about it is they knew something I didn’t, which is the coolest
technology in the world, the best idea doesn’t matter if no one actually wants
to buy your product. So it turns out no one actually wants to end high stakes testing. Everyone wants to talk about ending it, nobody wants to end it. It’s like a big, as they
would call it in Mexico completely normal standoff, where the parents and the schools and the companies and the testing company and the high school, nobody
wants to drop their gun. Nobody wants to be the
first one that says no test, so that was a hard thing to learn. But we built this amazing technology and then as I was
rethinking doing the post, the autopsy on that startup, we ended up selling it
to do something else. What we understood was
I actually don’t care whether people get good grades on tests or do well in their grades, what I actually care about is
they have a wonderful life. I care whether these kids
will grow up to live longer, be happier, and have a
bigger, more meaningful impact on the world, that’s all I care about. In fact, that’s the whole
purpose of the education system anywhere in the world is that, the tests, the universities,
everything else is just a mediator of those outcomes. So crazy mad science, right? How could you possibly predict where a four-year-old
will go in their life? Anyone here ever heard of
the marshmallow experiment? A couple, it’s a very famous experiment. I take three-year-olds and four-year-olds and I put a marshmallow in front of them and I say, if you don’t eat this, I’m gonna be back in 10 minutes, if you don’t eat the marshmallow, I’ll give you a second one. But if you do, you only get the one. And the original version
of this marshmallow was just to figure out where something called executive control came online. I get the hint I’m supposed
to shut the hell up soon, I haven’t even really started my talk. So when does executive control come on? Turns out three-year-olds eat
the marshmallow every time, four-year-olds don’t, but it turns out there’s this even more interesting effect. The four-year-olds wait a
different amount of time before they eat the marshmallow. Some of ’em you’ll like
you put the marshmallow in front of them and you say, now I’m gonna be, and then there’s just like an empty spinning
plate, like a cartoon, ’cause they’re just
(simulates whip cracking) and some of ’em would
wait like 20 minutes, that’s about as long as
that little kid will wait. You will note that means the researchers waited 20 minutes after they told the kid that they would only do 10 minutes, so the experiment
teaches these little kids two incredibly terrible
lessons about life, adults always lie and they
like to do cruel things to you. So every kid gets the second marshmallow and they wait for every kid to break. But the ones that wait the
longest before they break, this actually is a very robust, though controversial predictor, it actually predicts
university admissions, some of the variance in
university admissions. It predicts some of the
variance in lifetime income and health outcomes. Well, so does working memory span, so we checked 50 different constructs and then somehow I magically wrap up what appears to be a nine-hour talk in the next five minutes. We look at 50 different constructs, all of which are measurable using our various methodologies, are highly predictive, almost
certainly causally related to long-term life outcomes, and they can be changed about people. Turns out, you really want
to know what’s going on in the marshmallow experiment? It’s not that some kids
have executive control and some don’t, what’s really going on in the marshmallow experiment, you can do version of
this with adults as well, is do you believe your
hard work is gonna pay off? And I mean believe in
a neuroscientist sense, not like articulate it to me, but do you really believe? That is what predicts whether the kid will eat the marshmallow or not. In fact, you can run a little
manipulative experiment ahead of time, vary whether
the adult lied to them beforehand, control for their
background, trusted adults, it predicts how long they wait. What’s actually disturbing
about the marshmallow experiment isn’t that the kids that
take the shortest amount of time, like we should
send them, teach them how to dig ditches ’cause all they’ll do for the rest of their life,
that’s the total wrong takeaway. What’s truly horrific about
the marshmallow experiment is that by four years of age, so many little kids have learned that it’s not worth waiting. That’s terrible. But working memory span is another one, hugely predictive of life outcomes. Maybe not on an individual
level, but at a population level. And HUMM developed a technology, alternating currents,
transcranial stimulation of the midline theta that
literally you just flip a switch. Anyone know the Simon game, this old game with like lights and sounds and you have to do a
different length sequence, most of you, you’re above
average in this room, I would be willing to guess, so you probably won more like
six, seven, and nine lengths, maybe six, seven, eight,
I remember how to count. So flip a switch and you
are now eight, nine, tens, so it increases working
memory span by about 20%, literally with a flip of a switch. Working memory span is
predictive of how long you live, it is predictive of how much money you’ll make in your lifetime, it’s predictive of how far
you’ll go in your education. Increasing working memory span by 20% would be the equivalent
of a population shift in lifetime earnings of
about $30,000 a year, would be a huge impact on people’s lives. That’s not what they’ve invented, but still it’s been on
my website for 20 years, what if it was 20 plus or minus two? So if any of you are cognitive scientists, that’s a reference to the famous old, the very first paper
on working memory span, the magic number seven plus or minus two, what if we could just change that and we could make the average person higher than the most exceptional genius you will ever come across? What would the world be
like if that was true? I have no idea, no one
could actually know. I know that they look at the rest of us like we have Down syndrome,
but hey, that’s life. So what are we doing here,
and then I will shut up. So I’m advising them,
help get them some funding and helping them do some of the statistics and develop out the technology, and in exchange I get as much
of that technology as I want. We’re designing, running, and paying for an experiment with kids
with traumatic brain injury. One of the most common symptoms
of traumatic brain injury is loss of working memory span. They fell off their
bike, they were abused, they were in a car accident, something that had no fault of their own and their whole life
story has been taken away. Can we put the pen back in their hand to write their own life story? So I was gonna tell all
these other stories, but instead I wasted my
time with this stuff, and I didn’t even bother
telling any dirty jokes, that’s probably disappointing. – [Moderator] Well, Dr. Ming, maybe those will come out in questions, but we want to leave about
15 minutes for some Q and A if that’s all right, it’s so fascinating.
– That’s great, so I’m gonna finish up here. So I actually did give a TED
Talk and go check it out, it’s, anyone know who that is, by the way? I always like asking this question. Really, in Berkeley, right? So, Rodney Mullen, probably
the greatest skateboarder of all time, go see the TED
Talk to know what it’s about. If you’re gonna start a company, this is an incredibly important talk. It was originally titled The Fanatic, but it now goes under the title, ’cause apparently that was controversial, the title of the Paradox
of Incentive Insensitivity, and in that talk I
explain how we discovered that the best salespeople in the world don’t care about their bonuses, and that’s why they’re the best. Which is also true of the
best software developers and everyone else. This is another one, I’m not
going to explain what it is, I already talked a
little bit about purpose, so you have a sense, one
of my favorite definitions of purpose is paraphrasing
a famous old saying, the world gets better when
old men plant trees, right? When an old man plants a tree, he isn’t gonna be around to
lie down under its shade, this is for someone else, maybe
not even for his grandkids. Well, this is an actual old man, he’s planting trees single-handedly on the edge of a desert in India to hold back the desert all by himself, and he’s been doing it for decades. Talk about acting with a purpose. He will never reap the benefit
of what he’s doing, never, but he does it anyways. This is fun, so that’s me,
and anyone that doesn’t know my history, so I gave a talk, so I was at UC San Diego,
I will just take a moment and share this little bit of my story. So when I was a little kid, I was supposed to win a Nobel Prize, I don’t mean that in the sense that my parents were like dragon parents and they were, you will practice violin everyday, and it just was supposed
to happen, just period, I’m supposed to win a Nobel Prize. Somehow I was also supposed
to be a kicker in the NFL, I don’t know how those two
things were gonna work out at the same, but it was gonna happen, and it was just expected. And the more I tried to be that person, the worse things got. Everyone knew I was smart,
but the more I tried to be a smart kid, the worse I did, and the more unhappy I became. And the more unhappy I
became, the worse I performed, and then I felt like I
was letting everyone down. And by the time reached high school, if it wasn’t, ironically,
for standardized tests I would’ve washed out entirely. I would typically fail all of my classes right until the last
week and then scrape by by acing the big exam. And then I went to university in 1989, which I know is so long
ago that all of you are scoffing at the idea
that such a date existed, but fuck you. (audience laughs) And I almost immediately flunked and then ended up homeless for many years. So I was trying to figure
out, living in my car and I had a gun, ’cause it’s America, (audience laughs) And I spent a night in
1995, and, again, fuck you, trying to figure out
why I should be alive. I’d never done anything worthwhile, I’d only ever let people
down, what’s the point? Not a cry for help, trust
me, this is just euthanasia, life was awful, and I was
the main reason it was awful, so why not fix that? So it turns out the only
two people left in my life were my parents, and I
didn’t want to hurt them, so I needed to come up
with a reason to be alive. And now years later I
can look back on that and see that I found a purpose, or more accurately in
the scientific literature I developed construct strength of purpose. Turns out it doesn’t really
matter what the purpose is, as long as it’s bigger than you and you won’t complete
it in your lifetime. So the purpose I walked into backwards at the point of a gun that I was holding was live a life that other
people’s lives better and maximize human potential, as my marketing team once put it. And magically my life was fixed. Of course not, I was living in my car, it took five years to crawl
back out of that hole, but I did, ’cause now
I didn’t care anymore. The lesson I learned that night was that it wasn’t about me and it wasn’t about my happiness, it was about whether I was serving my purpose. In my case it’s a very
grounded humanist purpose, but if it’s a spiritual
one for you, that’s fine. I don’t know that Manchester
United counts as a purpose, but find something that
carries you through. Everyone has to have a purpose, if you can’t find yours,
there’s a whole list of people that want to sell you a purpose. One of ’em is this orange
gnome in the White House or it’s Boko Haram or it’s whatever, they’ve got purpose,
the world’s overflowing with purpose for you, what’s actually important
is that you build your own, that you are the kind of person that can see what it is that moves you. And the cool thing about it is, and this comes right out of the research, it isn’t something, there isn’t one thing you’re meant to do in this world, and it isn’t something
you have to go find, you get to build it tonight. You don’t have to wait at all, construct your own purpose, that’s what the research shows clearly. So I went back to school,
back to UC San Diego 10 years later where I had
flunked out previously, and I did my whole undergraduate
degree in a single year and I got perfect scores in every class. And then I went and did
my PhDs at Carnegie Mellon and then I was at Stanford
and then I started, I was here and I started
a bunch of companies. What it took for me to be successful is I had to let go of it all and not give a goddamn what anyone cared except what I believed was right. I needed to learn how to serve my purpose, and it served me
incredibly well since then. One of the steps along
the line was realizing that I didn’t want to be
that little boy anymore, and about 12 years ago I
went, actually right here at UC Berkeley, I went
through gender transition. Pure aside, first day
I ever showed up as me in my own lab was the last day anyone ever asked me a math question. Little disturbing, those of you, guys in the room that wonder if the girls have it differently, yeah, yeah, it’s very different, it’s sometimes obnoxiously different, like the time a VC patted me on the head and told me I should be
so proud of what I built. I’ve always wanted to
track that guy down since and kick him in the balls. (audience laughs) So what it really should say
is just fucking die already, just get it over with. Set an end date, you guys
want to start a company? Great, you’ve got three
years or five years, whatever it is, pick a date. As soon as you actually
get great at that, die, and start over again. And then die and start over again. I’ve been a scientist and an entrepreneur, I’ve been a writer and speaker, I’ve been a man and a woman, I’ve been all these different things. Every time I change I get better. I have a purpose, every one of my lives is a lesson in how to do it better. Every one of ’em counted,
and they count for something, but I never dwell on any of them. Saturday Morning Breakfast
Cereal, anyone a fan? Dirty jokes for physicists,
this is not my crowd. Go check it out, it’s very funny, but this is one of the ones,
and I will finish with this, that I really liked. Every now and then he gets
a little philosophical and this is one of the ones, here’s something that’s true,
one day you will be dead. Here’s something false,
you only live once. It takes about seven
years to master something, the truth is the science is a
little more complex than that, but the spirit is true, therefore starting at age
11, if you live to be 88, you have 11 opportunities to
be truly great at something. These are your lifetimes. When I first read this, it just incredibly resonated with me. I had already gone
through gender transition. I had already led
completely different lives. And he goes on to detail
people that just didn’t get it. And, boy, I understand all of them, ’cause I’ve been there. This is my favorite
panel of the whole bunch, two years ’til I die, I
wonder what I’m gonna do next. I said that once when I
was the chief scientist at one of the first companies
ever doing AI for hiring, and my CEO was sitting next to me and I caught out the corner of his eye like, oh shit, we’re
supposed to IPO in two years. But this is it, this is how
I go through everything. Soon as I get great at something, I’m out. Not because I don’t care, but because we’ve already made it happen, and I have no desire to
coast, I don’t need the money, I don’t even want it. I want to go make another difference. So spend a lifetime writing poems, spend another building things. Spend a life looking for facts, spend another looking for truth. These are your lifetimes. I know you feel very early, it feels like it’s
already been a lifetime, it hasn’t been anything. Nothing you’ve done,
I’m not saying your life to date hasn’t mattered, but I know, ’cause I can remember and
because it never goes away, it feels like the next decision you make will be the single-most
important decision forever, your whole life will be set as a result and you will never be able to
recover if you make a bad one. Boy did I make some bad decisions, and my life is as superlatively good as most people can imagine. And it isn’t because I’m special, anyone can do the work that I do. In fact, that’s what drives
me is that very thought. The kid that is gonna
come up with the cure for my child’s disease was
just born in a favela in Rio or in a shanty outside Cape Town or just down the street in Oakland, and they will never have
the chance to live that life and I will suffer as a result. How insane is it to not believe in the potential of other people? It’s incredibly insane. Go read research by Raj Chetty, he’ll win a Nobel Prize someday unless it turns out he’s been hitting on all of his female grad students (audience laughs) That was nervous laughter, (audience laughs) but he’s an amazing guy and
he runs a phenomenal research and he has this great
stuff, so go read it, especially the stuff about lost Einsteins. So you have this opportunity
to live all these lives and you’re so early on nothing you can do will take you so off track
that you can’t come back and do an amazing in the next life. Whatever you’re doing, go all in, build your own purpose, and
just fuckin’ die already. Because the greatest thing
you will ever do in your life is plant a tree, and you don’t need to wait, you don’t have to wait for
a VC to fund your tree. You don’t have to wait for your
board to approve your tree. You don’t have to wait
’til you’ve mastered arboreal arts and, God,
everyone wants people to learn how to program nowadays, what terrible career advice. If in 10 years anyone’s
programming, I will be shocked. So you just get to go do it, tonight, you get to do that. And I’m begging you,
every one of these lives is a chance to go and plant a tree, and I hope that you have the chance to use that life wisely. So thank you very much,
I’ll answer some questions. (audience applauds) – [Moderator] Dr. Ming, I told the class you’d be exceptional,
and you are exceptional. I think we are gonna students
do a feedback survey, but I’m gonna, if it’s okay with you, ask students to come up to you one on one and ask their questions. – Yeah, I did a terrible
job about leaving time for questions
– I know, I think that we can do this.
– But if anyone needs to take off, and, do the survey, and
then do what you need to do, but I’m happy to hang out
here for a little while, it’s that I, again, I
loathe social networking, but I love talking about my work. It’s the paradox of me, which is I think I’m a pretty ridiculous figure, but I am phenomenally passionate
about the work that I do. So, anyone wants to ask questions, anyone wants to pitch
me a blockchain company, please come up and do it. (audience laughs) I promise I’ll be gentle. All right, thanks. – [Moderator] Thank you. (audience applauds)

Leave a Reply

Your email address will not be published. Required fields are marked *