- Good evening, everyone.
My name is Rob Reich, I am delighted to welcome you here
to Stanford University for an evening of conversation
with Yuval Harari, Fei-Fei Li, and Nick Thompson.
I'm a professor of political science here
and the faculty director of the Stanford Center
for Ethics in Society, which is a co-sponsor
of tonight's event along with the Stanford Institute
for Human-Centered Artificial Intelligence
and the Stanford Humanities Center.
Our topic tonight is a big one.
We're going to be thinking together about the promises
and perils of artificial intelligence,
the technology quickly reshaping our economic, social,
and political worlds for better or for worse.
The questions raised by the emergence of AI
are by now familiar, at least to many people here
in Silicon Valley, but I think it's fair to say
that their importance is only growing.
What will the future of work look like
when millions of jobs can be automated?
Are we doomed or perhaps blessed to live in a world
where algorithms make decisions instead of humans?
And these are smaller questions in the big scheme of things.
What, might you ask, of the large ones?
Well, here are three.
What will become of the human species
if machine intelligence approaches
or exceeds that of an ordinary human being?
As a technology that currently relies
on massive centralized pools of data,
does AI favor authoritarian centralized governance
over more decentralized democratic governance?
And are we at the start now of an AI arms race,
and what will happen if powerful systems of AI, especially
when deployed for purposes like facial recognition,
are in the hands of authoritarian rulers?
These challenges only scratch the surface
when it comes to fully wresting with the implications of AI
as the technology continues to improve
and its use cases continue to multiply.
I wanna mention the format of the evening event.
First, given the vast areas of expertise
that Yuval and Fei-Fei have, when you ask questions
via Slido, those questions should pertain
or be limited to the topics under discussion tonight.
So this web interface that we're using, Slido,
allows people to up vote and down vote questions,
so you can see them now if you have
an Internet communication device.
If you don't have one, you can take one of these post cards,
which hopefully you got outside, and on the back,
you can fill in a question you might have
about the evening event and collect it at the end,
and the Stanford Humanities Center
will try to foster some type of conversation
on the basis of those questions.
A couple housekeeping things.
If you didn't purchase one already, Yuval's books
are available for sale outside in the lobby after the event.
A reminder to please turn your cell phone ringers off.
And we will have 90 minutes
for our moderated conversation here
and will end sharp after 90 minutes.
Now, I'm going to leave the stage in just a minute
and allow a really amazing undergraduate student here
at Stanford to introduce our guests.
Her name is Anna-Sofia Lesiv, let me just tell you
a bit about her.
She's a junior here at Stanford, majoring in economics
with a minor in computer science.
And outside the classroom, Anna-Sofia is a journalist
whose work has been featured in The Globe and Mail,
Al Jazeera, The Mercury News, The Seattle Times,
and this campus' paper of record, The Stanford Daily.
She's currently the executive editor of the Daily,
and her daily magazine article from earlier
in the year called CS + Ethics examined the history
of computer science and ethics education at Stanford,
and it won the student prize for best journalism of 2018.
She continues to publish probing examinations
of the ethical challenges faced
by technologists here and elsewhere.
So ladies and gentlemen, I invite you to remember
this name, for you'll be reading about her
or reading her articles or likely both.
Please welcome Stanford junior Anna-Sofia Lesiv.
- Thank you very much for the introduction, Rob.
Well, it's my great honor now to introduce
our three guests tonight, Yuval Noah Harari,
Fei-Fei Li, and Nicholas Thompson.
Professor Yuval Noah Harari is a historian,
futurist, philosopher, and professor at Hebrew University.
The world also knows him for authoring some
of the most ambitious and influential books of our decade.
Professor Harari's internationally best selling books,
which have sold millions of copies worldwide,
have covered a dizzying array of subject matter,
from narrativizing the entire history
of the human race and sapience
to predicting the future awaiting humanity,
and even coining a new faith called dataism in Homo Deus.
Professor Harari has become a beloved figure
in Silicon Valley whose readings are assigned
in Stanford's classrooms and whose name is whispered through
the hallways of the comparative literature
and computer science departments alike.
His most recent book is 21 Lessons for the 21st Century,
which focuses on the technological, social,
political, and ecological challenges of the present moment.
In this work, Harari cautions that,
as technological breakthroughs continue to accelerate,
we will have less and less time to reflect upon
the meaning and consequences of the changes they bring.
And this urgency is what charges
professor Fei-Fei Li's work every day
in her role as the co-director
of Stanford's Human-Centered AI Institute.
This institute is one of the first to insist that AI
is not merely the domain of technologists,
but a fundamentally interdisciplinary
and ultimately human issue.
Her fascination with the fundamental questions
of human intelligence is what piqued her interest
in neuroscience, as she eventually become one
of the world's greatest experts in the fields
of computer vision, machine learning,
and cognitive and computational neuroscience.
She's published over 100 scientific articles
in leading journals and has had research supported
by the National Science Foundation,
Microsoft, and The Sloan Foundation.
From 2013 to 2018, professor Fei-Fei Li served
as the director of Stanford's AI lab,
and between January 2017 and September 2018,
professor Fei-Fei Li served as vice president at Google
and chief scientist of AI and machine learning
at Google Cloud.
Nicholas Thompson is the editor-in-chief
of Wired Magazine, a position he's held since January 2017.
Under Mr. Thompson's leadership,
the topic of artificial intelligence has come
to hold a special place at the magazine.
Not only has Wired assigned him more feature stories
on AI than on any other subject,
but it is the only specific topic
with a full time reporter assigned to it.
It's no wonder then that professors Harari and Li
are no strangers to its pages.
Mr. Thompson has led discussions
with the world's leaders in technology and AI,
including Mark Zuckerberg on Facebook and privacy,
French president Emmanuel Macron on France's AI strategy,
and Ray Kurzweil on the ethics and limits of AI.
Mr. Thompson is a Stanford University graduate
who earned his BA double majoring in Earth systems
and political science, and impressively,
even completed a third degree in economics.
Of course, I would be remiss if I did not mention
that Mr. Thompson cut his journalistic teeth
in the opinion section of the Stanford Daily.
So Nick, that makes both of us.
Like all our guests today, I am
at once fascinated and worried by the challenges
that artificial intelligence poses for our society.
One of my goals at Stanford has been
to write about and document the challenge
of educating a generation of students whose lives
and workplaces will eventually be transformed by AI.
Most recently, I published
an article called Complacent Valley with the Stanford Daily.
In it, I critiqued our propensity
to become overly comfortable with the technological
and financial achievements that Silicon Valley
has already reached lest we become complacent
and lose our ambition and momentum
to tackle the great challenges the world has in store.
Answering the fundamental questions
of what we should spend our time on,
how we should live our lives has become much more difficult,
particularly on the doorstep of the AI revolution.
I believe that the kind of crisis of agency
that author J. D. Vance wrote of in Hillbilly Elegy,
for example, is not confined to Appalachia
or the deindustrialized Midwest,
but is emerging even at elite institutions like Stanford.
So conversations like ours this evening hosting speakers
that aim to recenter the individual at the heart of AI
will show us how to take responsibility in a moment
when most decisions can seemingly be made
for us by algorithms.
There are no narratives to guide us through
a future with AI, no ancient myths or stories
that we may rely on to tell us what to do.
At a time when humanity is facing
its greatest challenge yet, somehow,
we cannot be more at a loss for ideas or direction.
It's this momentous crossroads in human history
that pulls me towards journalism and writing in the future,
and it's why I'm so eager to hear
our three guests discuss exactly such a future tonight.
So please, give me a very, please join me
in giving them a very warm welcome this evening.
Thank you so much, Anna-Sofia, thank you, Rob.
Thank you, Stanford, for inviting us all here,
I'm having a flashback to the last time I was on a stage
at Stanford, which was playing guitar at the CoHo.
And I didn't have either Yuval or Fei-Fei with me,
so there were about six people in the audience,
one of whom had her headphones on.
But I did meet my wife.
Isn't that sweet?
All right, so a reminder, housekeeping.
Questions are gonna come in in Slido,
you can put them in, you can vote up questions,
we've already got several thousand.
So please, vote up the ones you really like.
If someone can program an AI that can get
a really devastating question in and stump Yuval,
I will get you a free subscription to Wired.
I want this conversation to kind of have three parts.
First, lay out where we are.
Then talk about some of the choices we have to make now.
And last, talk about some advice for all
of the wonderful people in the halls.
So those are the three general areas,
I'll feed in questions as we go.
We may have a specific period for questions at the end,
but let's get cracking.
So the last time we talked, you said many,
many brilliant things, but one that stuck out,
it was a line where you said, we are not just
in a technological crisis, we are in a philosophical crisis.
So explain what you meant, explain how it ties to AI,
and let's get going with a note of existential angst.
- Yeah, so I think what's happening now
is that the philosophical framework of the modern world
that has been established in the 17th and 18th century
around ideas like human agency and individual free will
are being challenged like never before.
Not by philosophical ideas, but by practical technologies.
And we see more and more questions,
which used to be, you know, the bread and butter
of the philosophy department being moved
to the engineering department.
And that's scary partly because, unlike philosophers
who are extremely patient people.
They can discuss something for thousands of years
without reaching any agreement and they are fine with that.
The engineers won't wait.
And even if the engineers are willing to wait,
the investors behind the engineers won't wait.
So it means that we don't have a lot of time.
And in order to encapsulate what the crisis is,
I know that, you know, engineers, especially
in a place like Silicon Valley, they like equations.
So maybe I can try and formulate an equation
to explain what's happening.
And the equation is B times C times D equals H.
Which means biological knowledge multiplied
by computing power multiplied by data equals
the ability to hack humans.
And the AI revolution or crisis is not just AI,
it's also biology, it's biotech.
We haven't seen anything yet because
the link is not complete.
There is a lot of hype now around AI and computers
but just, there is just half the story.
The other half is the biological knowledge
coming from brain science and biology.
And once you link that to AI, what you get
is the ability to hack humans.
And maybe I'll explain what it means,
the ability to hack humans to create an algorithm
that understands me better than I understand myself
and can therefore manipulate me, enhance me, or replace me.
And this is something that our philosophical baggage
and all our belief in, you know, human agency
and free will and the customer is always right
and the voter knows best, this just falls apart
once you have this kind of ability.
- Once you have this kind of ability and it's used
to manipulate or replace you, not if it's used
to enhance you.
- Also when it's used to enhance you, the question is,
who decides what is a good enhancement
and what is a bad enhancement?
So our immediate fallback position is to fall back
on the traditional humanist ideas
that the customer is always right.
The customers will choose the enhancement.
Or the voter is always right, the voters will vote,
there will be a political decision about enhancement.
Or if it feels good, do it.
We'll just follow our heart, we'll just listen to ourselves.
None of this works when there is a technology
to hack human on a large scale.
You can't trust your feelings or the voters
or the customers on that.
The easiest people to manipulate are the people
who believe in free will, because they think
they cannot be manipulated.
So how do you decide what to enhance if,
and this is a very deep ethical and philosophical question.
Again, that philosophers have been debating
for thousands of years, what is good,
what are the good qualities we need to enhance?
So if you can't trust the customer, if you can't trust
the voter, if you can't trust your feelings,
who do you trust, what do you go by?
- All right, Fei-Fei, you have a PhD, you have a CS degree,
you're a professor at Stanford.
Does A times B times C equal H?
Is Yuval's theory the right way
to look at where we're headed?
What a beginning, thank you, Yuval.
One of the things, I've been reading Yuval's book
for the past couple of years and talking to you.
And I'm very envious of philosophers now,
because they can propose questions in crisis,
but they don't have to answer them.
No, as an engineer and scientist, I feel like
we have to now solve the crisis.
So honestly, I think I'm very thankful, I mean,
personally, I've been reading your book for two years.
And I'm very thankful that Yuval, among other people,
but have opened up this really important question for us.
And it's also quite a, when you said the AI crisis,
and I was sitting there thinking,
this is a field I loved and felt passionate about
and researched for 20 years.
And that was just a scientific curiosity
of a young scientist entering PhD in AI.
How did, what happened that 20 years later,
it has become a crisis?
And it actually, speak of the evolution of AI
that got me where I am today and got my colleagues
at Stanford where we are today with the human-centered AI
is that this is a transformative technology,
it's a nascent technology, it's still
a budding science compared to physics, chemistry, biology.
But with the power of data, computing,
and the kind of diverse impact that AI is making,
it is, like you said, it's touching human lives
and business in broad and deep ways.
And responding to that kind of questions and crisis
that's facing humanity, I think one
of the proposed solution, or if not a solution,
at least a try, that Stanford is making an effort about
is can we reframe the education, the research,
and the dialogue of AI and technology in general
in a human-centered way?
We're not necessarily gonna find a solution today.
But can we involve the humanists, the philosophers,
the historians, the political scientists, the economists,
the ethicists, the legal scholars, the neuroscientists,
the psychologists, and many more other disciplines
into the study and development of AI
in the next chapter, in the next phase?
- Don't be so certain we're not gonna get
an answer today, I've got two of the smartest people
in the world glued to their chairs and I've got Slido
for 72 minutes, so let's give it a shot.
- He said we have thousands of years.
- But let me go a little bit further in Yuval's question.
So there are a lot, or Yuval's opening statement,
there are a lot of crises about AI that people talk about.
They talk about AI becoming conscious
and what would that mean, they talk about job displacement,
they talk about biases, and Yuval has very clearly laid out
what he thinks is the most important one,
which is the combination of biology
plus computing plus data leading to hacking.
So he's laid out a very specific concern.
Is that specific concern what people who are thinking
about AI should be focused on?
- So absolutely.
So any technology humanity has created, starting from fire,
is a double edge sword.
So it can bring improvements to life, to work,
and to society, but it can bring the perils,
and AI has the perils, you know.
I wake up every day, worry about the diversity,
inclusion issue in AI.
We worry about fairness or the lack of fairness, privacy,
the labor market, so absolutely, we need to be concerned,
and because of that, we need to expand the study,
the research, and the development of policies
and the dialogue of AI beyond just
the codes and the products into these human wrongs,
into these societal issues.
So I absolutely agree with you on that,
that this is the moment to open the dialogue,
to open the research in those issues.
- Even though I would just say that, again,
part of my fear is that the dialogue,
I don't fear AI experts talking with philosophers,
I'm fine with that, historians, good,
literary critics, wonderful.
I fear the moment you start talking with biologists.
That's my biggest fear, when you and the biologists will,
hey, we actually have a common language.
And we can do things together.
And that's when the really scary things I think will be--
- [Fei-Fei] Can you elaborate on, what is scaring you
that we talk to biologists?
- That's the moment when you can really hack human beings
not by collecting data about our search words
or our purchasing habits or where do we go about town.
But you can actually start peering inside
and collect data directly from our hearts
and from our brains.
- Okay, can I be specific?
First of all, the birth of AI is AI scientists talking
to biologists, specifically neuroscientists, right,
the birth of AI is very much inspired
by what the brain does.
Fast forward to 60 years later, today's AI
is making great improvement in healthcare,
there's a lot of data from our physiology
and pathology being collected and using machine learning
to help us.
But I feel like you're talking about something else.
- That's part of it, I mean, if there wasn't a great promise
in the technology, there would also be no danger
because nobody would go along with that path.
I mean, obviously, there are enormously beneficial things
that AI can do for us, especially
when it is linked with biology.
We are about to get the best healthcare in the world
in history, and the cheapest, and available
for billions of people via their smartphones,
which today they have almost nothing.
And this is why it is almost impossible
to resist the temptation.
And with all the issue, you know, of privacy,
if you have a big battle between privacy and health,
health is likely to win hands down.
So I fully agree with that, and you know,
my job as a historian, as a philosopher,
as a social critic, is to point out the dangers in that,
because, especially in Silicon Valley,
people are very much familiar with the advantages,
but they don't like to think so much about the dangers.
And the big danger is what happens when you can hack
the brain, and that can serve not just
your healthcare provider, that can serve so many things
from a crazy dictator to--
- Let's focus on that, what it means to hack the brain.
Like what, right now, in some ways, my brain is hacked,
right, there is an allure of this device,
it wants me to check it constantly,
like my brain has been a little bit hacked,
yours hasn't because you meditate two hours a day,
but mine has, and probably most of these people have.
But what exactly is the future brain hacking going
to be that it isn't today?
- Much more of the same, but on a much larger scale.
I mean, the point when, for example,
more and more of your personal decisions in lives
are being outsourced to an algorithm that
is just so much better than you.
So you know, we have two distinct dystopias
that kind of mesh together.
We have the dystopia of surveillance capitalism,
in which there is no like Big Brother dictator,
but more and more of your decisions
are being made by an algorithm.
And it's not just decisions about what to eat
or what to shop, but decisions like where to work
and where to study and whom to date
and whom to marry and whom to vote for.
It's the same logic.
And I would be curious to hear if you think
that there is anything in humans which
is by definition unhackable.
That we can't reach a point when the algorithm
can make that decision better than me.
So that's one line of dystopia which is
a bit more familiar in this part of the world.
And then you have the full fledged dystopia
of a totalitarian regime based
on a total surveillance system.
Something like the totalitarian regimes
that we have seen in the 20th century,
but augmented with biometric censors
and the ability to basically track each
and every individual 24 hours a day.
And you know, which in the days of, I don't know,
Stalin or Hitler was absolutely impossible
because they didn't have the technology,
but maybe might be possible in 20 years to 30 years.
So we can choose which dystopia to discuss,
but they are very close in the--
- Let's choose the liberal democracy dystopia.
Fei-Fei, do you wanna answer Yuval's specific question,
which is is there something in dystopia A,
liberal democracy dystopia, is there something
endemic to humans that cannot be hacked?
- So when you asked me that question just two minutes ago,
the first word that came to my mind is love.
Is love hackable?
- Ask the Internet, I don't know.
- That's a defense--
- Dating is not the entirely of love, I hope.
- But the question is, which kind of love
are you referring to?
If you're referring to this, you know, I don't know,
Greek philosophical love or the loving kindness of Buddhism,
that's one question, which I think
it's much more complicated.
If you are referring to
the biological mammalian courtship rituals and,
then I think yes, I mean, why not?
Why is it different from anything else
that is happening in the body?
- But humans are humans because we are,
there is some part of us that are beyond
the mammalian courtship, right?
So is that part hackable?
- That's the question, I mean, you know,
in most science fiction books and movies,
they give your answer.
When the extraterrestrial evil robots are about
to conquer planet Earth and nothing can resist them,
resistance is futile, at the very last moment,
humans win because the robots don't understand love.
- Last moment, there's one heroic white dude that saves us.
- Why we do this?
- No, no, it was a joke, don't worry.
But okay, so the two dystopia, I do not have answers
to the two dystopias.
But I wanna keep saying is, this is precisely why
this is the moment that we need to seek for solutions.
This is precisely why this is the moment
that we believe the new chapter of AI needs to be written
by cross-pollinating efforts from humanists,
social scientists to business leaders to civil society
to governments to come at the same table,
to have that multilateral and cooperative conversation,
and I think you really bring out the urgency
and the importance and the scale of this potential crisis.
But I think in the face of that, we need to act.
- Yeah, and I agree that we need cooperation,
that we need much closer cooperation between engineers
and philosophers or engineers and historians.
And also, from a philosophical perspective,
I think there is something wonderful
about engineers philosophically.
- Thank you.
- That they really cut the bullshit.
I mean, philosophers can talk and talk, you know,
in cloudy and flowery metaphors.
And then the engineers can really focus the question.
Like I just had a discussion the other day
with an engineer from Google about this.
And that he said okay, I know how to maximize people's time
on the website.
If somebody comes to me and tells me, look,
your job is to maximize time on this application,
I know how to do it because I know how to measure it.
But if somebody comes along and tells me, well,
you need to maximize human flourishing or you need
to maximize universal love, I don't know what it means.
So that's what the engineers go back to the philosophers
and ask them, what do you actually mean?
Which, you know, a lot of philosophical theories collapse
around that, because they can't really explain what,
and we need this kind of collaboration
in order to move forward. - We need a equation for that.
- But then, Yuval, is Fei-Fei right, if we can't explain
and we can't code love, can artificial intelligence
ever recreate it, or it is something intrinsic
to humans that the machines will never emulate?
- I don't think that machines will feel love,
but you don't necessarily need to feel it
in order to be able to hack it, to monitor it,
to predict it, to manipulate it.
I mean, machines don't like to play Candy Crush,
but they can still--
- This device, in some future where
it's infinitely more powerful than it's right now,
could make me fall in love with somebody in the audience?
- That's, that goes to the question
of consciousness and mind.
- We should go there.
- I don't think that we have the understanding
of what consciousness is to answer the question
whether a non-organic consciousness is possible
or is not possible, I think we just don't know.
But again, the bar for hacking humans is much lower,
the machines don't need to have consciousness of their own
in order to predict our choices and manipulate our choices.
They just need to, all right, if you accept
that something like love is, in the end,
a biological process in the body, if you think that AI
can provide us with wonderful healthcare,
by being able to monitor and predict something like
the flu or something like cancer,
what's the essential difference between flu and love
in the sense of, is this biological
and this is something else which is so separated
from the biological reality of the body that,
even if we have a machine that's capable of monitoring
and predicting flu, it still lacks something essential
in order to do the same thing with love?
- [Nick] Fei-Fei.
- So I wanna make two comments, and this is where
my engineering, you know, personality speaking.
We're making two very important assumptions
in this part of the conversation, one is that AI
is so omnipotent, that it's achieved to a state
that it's beyond predicting anything physical,
it's got into the consciousness level, it got into even
the ultimate, the love level of capability.
And I do wanna make sure that we recognize
that we're very very very far from that,
this technology is still very nascent,
part of the concern I have about today's AI
is that super hyping of its capability.
So I'm not saying that that's not a valid question,
but I think that part of this conversation
is built upon that assumption that this technology
has become that powerful, and there is,
I don't know how many decades we are from that.
Second related assumption I feel we are,
our conversation is being based on is that
we're talking about the world or a state of the world
that only that powerful AI exists
or that small group of people who have produced
the powerful AI and is intended to hack human are existing.
But in fact, our human society is so complex,
there's so many of us, right.
I mean, humanity in its history have faced
so many technology, if we left it in the hands
of a bad player alone without any regulation,
multinational collaboration, rules, laws, moral codes,
that technology could have maybe not hack human,
but destroy human or hurt human in massive ways.
It has happened, but by and large, our society,
in a historical view, is moving to a more civilized
and controlled state.
So I think it's important to look at that greater society
and bringing other players and people into this dialogue
so we don't talk like there is only this omnipotent AI,
you know, deciding it's gonna hack everything to the end.
And that brings to your topic that, in addition
of hacking human at that level that you're talking about,
there are some very immediate concerns already.
Diversity, privacy, labor, legal changes, you know,
And I think it's critical to tackle those now.
- Well, let's, I love talking to AI researchers,
because five years ago, all the AI researchers were like,
it's much more powerful than you think,
and now they're all like, it's not as powerful as you think.
All right, so I'll just, let me ask--
- It's because five years go, you have no idea what AI is,
now you're extrapolating too much.
- I didn't say it was wrong, I just said it was the thing.
Let's, I wanna go into what you just said,
but before I do that, I wanna take one question here
from the audience, because once we move into
the second section, we won't be able to answer it.
So the question is, it's for you, Yuval.
How do we, this is from Marin Nasini,
how can we avoid the formation
of AI powered digital dictatorships?
So how do we avoid dystopia number two,
let's enter that, and then let's go, Fei-Fei,
into what we can do right now,
not what we can do in the future.
- The key issue is how to regulate the ownership of data.
Because we won't stop research in biology
and we won't stop research in computer science and AI,
so from the three components of biological knowledge,
computing power, and data, I think data is the easiest,
and it's also very difficult, but still the easiest kind of
to regulate or to protect.
Place some protections there.
And there are efforts now being made.
And they are not just political efforts,
but you know, also philosophical efforts
to really conceptualize what does it mean
to own data or to regulate the ownership of data?
Because we have a fairly good understanding
what it means to own land.
We had thousands of years of experience with that.
We have a very poor understanding
of what it actually means to own data
and how to regulate it, but this is a very important front
that we need to focus on in order to prevent
the worst dystopian outcomes.
And I agree that AI is not nearly as powerful
as some people imagine, but this is why,
again, I think we think to place the bar low.
To reach a critical threshold, we don't need the AI
to know us perfectly, which will never happen.
We just need the AI to know us better
than we known ourselves, which is not so difficult
because most people don't know themselves very well
and often make huge mistakes in critical decisions.
So whether it's finance or career or love life,
to have this shift in authority from humans to algorithm,
they can still be terrible.
But as long as they are a bit less terrible than us,
the authority will shift to them.
- You, in your book, you tell a very illuminating story
about your own self and you come to terms
with who you are and how you could be manipulated.
Will you tell that story here,
about coming to terms with your sexuality
and the story you told about Coca-Cola in your book?
'Cause I think that will make it clear
what you mean here very well.
- Yeah, so I said that I only realized
that I was gay when I was 21.
And I look back at the time, and I was,
I don't know, 15, 17.
And it should've been so obvious.
How, and it's not like a stranger,
like I'm with myself 24 hours a day
and I just don't notice any of the screaming signs
that say yeah, you were gay.
And I don't know how, but the fact is, I missed it.
Now, an AI, even a very stupid AI today, will not miss it.
- I'm not so sure.
- So imagine, this is not like, you know,
like a science fiction scenario for century from now.
This can happen today, that you can write
all kinds of algorithms that, you know,
they are not perfect, but they are still better say
than the average teenager, and what does it mean
to live in a world in which you learn about,
something so important about yourself from an algorithm,
what does it mean, what happens if the algorithm
doesn't share the information with you,
but it shares the information with advertisers
or with governments?
So if you want to, and I think we should go down
from the cloudy heights of, you know, the extreme scenarios,
to the practicalities of day to day life,
this is a good example.
Because this is already happening.
- Yeah, all right, well, let's take
the elevator down to the more conceptual level
at this particular shopping mall
that we're shopping in today.
And Fei-Fei, let's talk about what we can do today
as we think about the risks of AI, the benefits of AI,
and tell us your punch list of what you think
the most important things we should be thinking
about with AI are.
- Oh boy, there are so many things we could do today.
And I cannot agree more with Yuval
that this is such an important topic.
Again, I'm gonna try to speak about all the efforts
that's been made at Stanford, because I think
this is a good representation of what we believed
are so many efforts we can do.
So in human-centered AI in which,
this is the overall theme we believe
that the next chapter of AI should be is human-centered,
we believe in three major principles.
One principle is to invest in the next generation
of AI technology that is more, that reflects more
of the kind of human intelligence we would like.
I was just thinking about your comment about AI's dependence
on data and how the policy and governance of data
should emerge in order to regulate and govern the AI impact,
while technology is, we should be developing technology
that can explain AI.
In technical field, we call it explainable AI
or AI interpretability studies.
We should be focusing on technology that have
the more nuanced understanding of human intelligence.
We should be investing in the development
of less data dependent AI technology
that would take into considerations of intuition,
knowledge, creativity and other forms of human intelligence.
So that kind of human intelligence inspired AI
is one of our principles.
The second principle is to, again,
welcome in the kind of multidisciplinary study of AI,
cross-pollinating with economics, with ethics, with law,
with philosophy, with history, cognitive science and so on,
because there is so much more we need to understand
in terms of AI's social, human, anthropological,
And we cannot possibly do this alone as technologists,
some of us shouldn't even be doing this,
it's the ethicists, philosophers should participate
and work with us on these issues.
So that's the second principle.
And the third principle, and within this,
we work with policymakers.
We convene the kind of dialogues
of multilateral stakeholders.
Then the third, last but not the least.
I think Nick, you said that at the very beginning
of this conversation, that we need to promote
the human enhancing and collaborative
and augmentative aspect of this technology.
You have a point, even there, it can become manipulative.
But we need to start with that sense of alertness,
understanding, but still promote that kind
of benevolent applications and design of this technology,
at least these are the three principles
that Stanford's Human-Centered AI Institute is based on.
And I just feel very proud within the short few months
of the birth of this institute, there are more
than 200 faculty involved on this campus in this kind
of research dialogue, you know, study education.
And their number's still growing.
Let's, of those three principles,
let's start digging into them.
So let's go to number one, explainability,
'cause this is a really interesting debate
in artificial intelligence.
So there are some practitioners who say
you should have algorithms that can explain
what they did and the choices they made.
Sounds eminently sensible.
But how do you do that?
I make all kinds of decisions that
I can't entirely explain, like why did I hire this person,
not that person, and I can tell a story
about why I did it, but I don't know for sure.
Like we don't know ourselves well enough
to always be able to truthfully and fully explain what
we did, how can we expect a computer using AI to do that?
And if we demand that here in the West,
then there are other parts of the world
that don't demand that who may be able to move faster.
So why don't we start, why don't I ask you the first part
of that question, Yuval the second part of that question.
So the first part is, can we actually get explainability
if it's super hard even within ourselves?
- Well, it's pretty hard for me to multiply two digits,
but you know, computers can do that.
So the fact that something is hard for humans
doesn't mean we shouldn't try to get the machines to do,
especially, you know, after all these algorithms
are based on very simple mathematical logic.
Granted, we're dealing with neural networks these days
of millions of nodes and billions of connections.
So explainability is actually tough, it's ongoing research.
But I think this is such a fertile ground
and it so critical when it comes to healthcare decisions,
financial decisions, legal decisions.
There is so many scenarios where this technology
can be potentially positively useful,
but with that kind of explainable capability.
So we've gotta try, and I'm pretty confident,
with a lot of smart minds out there,
this is a crackable thing.
And on top--
- [Nick] Got 200 professors on it.
- Right, not all of them doing AI algorithms.
On top of that, I think you have a point that,
if we have technology that can explain
the decision making process of algorithms,
it makes it harder for it to manipulate and cheat, right.
It's a technical solution, not the entirety
of the solution, that will contribute
to the clarification of what this technology is doing.
- But because the, presumably, the AI makes decision
in a radically different way than humans,
then even if the AI explains its logic, the fear is,
it will make absolutely no sense to most humans.
Most humans, when they are asked to explain a decision,
they tell a story in a narrative form,
which may or may not reflect what
is actually happening within them.
In many cases, it doesn't reflect.
It's just a made up rationalization and not the real thing.
Now, an AI could be much better than a human
in telling me like, I applied for a bank,
to the bank for a loan, and the bank says no,
and I ask why not?
And the bank says okay, we'll ask our AI.
And the AI gives this extremely long statistical analysis
based not on one or two salient feature of my life,
but on 2517 different data points,
which it took into account and gave different weights.
And why did you give this, this weight,
and why did you give, oh, there is another book about that.
And most of the data points would seem
to a human completely irrelevant.
You applied for a loan on Monday and not on Wednesday.
And the AI discovered that, for whatever reason,
it's after the weekend, whatever,
people who apply for loans on a Monday
are 0.075% less likely to repay the loan.
So it goes into the equation.
And I get this book of the real explanation, finally,
I get a real explanation.
It's not like sitting with a human banker
that just bullshits me.
- So are you rooting for AI?
Are you saying AI is good in this case?
- In many cases, yes, I mean, I think in many case,
it's two sides of the coin, I think that,
in may ways, the AI in this scenario
will be an improvement over the human banker.
Because for example, you can really know
what the decision is based on, presumably.
But it's based on something that I as a human being just
cannot grasp, I just don't, I know how to deal
with simple narrative stories.
I didn't give you a loan because you're gay,
that's not good.
Or because you didn't repay any of your previous loans,
okay, I can understand that.
But I don't, my mind doesn't know what to do
with the real explanation that the AI will give,
which is just this crazy statistical thing
which says nothing to me.
- Okay, so there are two layers to your comment.
One is how do you trust and be able to comprehend
the AI's explanation?
Second is, actually, can AI be used
to make humans more trustable or be more trustable humans?
On the first point, I agree with you,
if AI gives you 2000 dimensions of potential features
with probability, it's not human understandable.
But the entire history of science in human civilization
is to be able to communicate the result of science
in better and better ways, right.
Like I just had my annual physical and
a whole bunch of numbers came to my cell phone.
And well, first of all, my doctors can,
the expert can help me to explain these numbers.
Now even Wikipedia can help me
to explain some of these numbers.
But the technological improvements
of explaining these will improve.
It's our failure as AI technologists if
we just throw 2000 dimensions of probability numbers at you.
- But this is, I mean, this is the explanation,
and I think that the point you raise is very important.
But I see it differently, I think science
is getting worse and worse in explaining its theories
and findings to general public, which is the reason
for things like doubting climate change and so forth,
and it's not really even the fault of the scientists.
Because the science is just getting more
and more complicated, and reality is extremely complicated,
and the human mind wasn't adapted to understanding
the dynamics of climate change or the real reasons
for refusing to give somebody a loan.
That's the point when you have an, again,
let's put aside the whole question of manipulation
and how can I trust, let's assume the AI is benign
and let's assume it makes, that there are no hidden biases,
everything is okay.
But still, I can't understand the decisions of the AI.
- People like Nick, the storytellers has to explain.
What I'm saying, you're right, it's very complex.
But there are people like--
- I'm gonna lose my job to a computer like next week,
but I'm happy to have your confidence with me.
- But that's the job of the society collectively,
to explain the complex science.
I'm not saying we're doing a great job at all.
But I'm saying there is hope if we try.
- But my fear is that we just really can't do it
because the human mind is not built for dealing
with these kinds of explanations and technologies.
And it's true for, I mean, it's true
for the individual customer who goes to the bank
and the bank refuses to give them a loan.
And it can even be on the level, I mean,
how many people today on Earth understand
the financial system?
How many presidents and prime ministers understand
the financial system?
- In this country, it's zero.
- What does it mean to live in a society where the people
who are supposed to be running the business,
and again, it's not the fault of a particular politician.
It's just the financial system has become so complicated,
I don't think that economists are trying on purpose
to hide something from general public.
It's just extremely complicated.
You had some of the wisest people in the world going
to the finance industry and creating
these enormously complex models and tools which objectively,
you just can't explain it to most people unless,
first of all, they study economics and mathematics
for 10 years or whatever.
So I think this is a real crisis.
And this is, again, this is part
of the philosophical crisis we started with.
And the undermining of human agency is,
that's part of what's happening,
that we have these extremely intelligent tools
that are able to make perhaps better decisions
about our healthcare, about our financial system.
But we can't understand what they are doing
and why they are doing it.
And this undermines our autonomy and our authority.
And we don't know as a society how to deal with that.
- Well, ideally, Fei-Fei's institute will help that.
Before we leave this topic though,
I wanna move to a very closely related question,
which I think is one of the most interesting,
which is the question of bias in algorithms,
which is something you've spoken eloquently about,
and let's take the financial system.
So you can imagine a loan used by a bank
to determine whether somebody should get a loan.
And you can imagine training it on historical data,
and historical data is racist, and we don't want that.
So let's figure out how to make sure the data isn't racist
and that it gives loans to people regardless of race.
I think we probably all, everybody in this room agrees
that that is a good outcome.
But let's say that analyzing the historical data suggests
that women are more likely to repay their loans than men.
Do we strip that out or do we allow that to stay in?
If you allow it to stay in, you get
a slightly more efficient financial system.
If you strip it out, you have a little more equality
between men and women.
How do you make decisions about what biases you wanna strip
and which ones are okay to keep?
- Yeah, that's an excellent question, I mean,
I'm not gonna have the answers personally,
but I think you touch on a really important question,
this, first of all, machine learning system bias
is a real thing, you know, like you said.
It starts with data, it probably starts
with the very moment where collecting data
and the type of data we're collecting,
all the way through the whole pipeline
and then all the way to the application.
But biases come in very complex ways.
At Stanford, we have machine learning scientists studying
the technical solutions of bias, like you know,
debiasing data and normalizing certain decision making.
But we also have humanists debating about
what is bias, what is fairness, when is bias good,
when is bias bad?
So I think you just opened up a perfect topic
for research and debate and conversation in this topic.
And I also wanna point out that Yuval, you already used
a very closely related example.
Machine learning algorithm has a potential
to actually expose bias, right.
It, you know, like one of my favorite study
was a paper a couple of years ago analyzing Hollywood movies
and using machine learning face recognition algorithm,
which is a very controversial technology these days,
to recognize Hollywood systematically gives more screen time
to male actors than female actors.
That's, no human being can sit there
and count all the frames of faces and gender bias.
And this is a perfect example of using machine learning
to expose bias.
So in general, there is a rich set
of issues we should study, and again,
bring the humanists, bring the ethicists,
bring the legal scholars, bring the gender study experts.
- Agreed, though standing up for humans, I knew Hollywood
was sexist even before that paper, but yes, agreed.
- You're a smart human.
- Yuval, on that question of the loans.
Do you strip out the racist data,
do you strip out the gender data,
what biases do you get rid of, what biases do you not?
- I don't think there is a one size fits all, I mean,
it's a question we, again, we need this day
to day collaboration between engineers and ethicists
and psychologists and political scientists.
- [Nick] But not biologists, right, but not biologists?
- And increasingly also biologists.
And you know, and it goes back to the question,
what should we do?
So we should teach ethics to coders
as part of their curriculum.
The people today in the world that most need
a background in ethics is the people
in the computer science departments.
So it should be an integral part of the curriculum.
And it's also in the big corporations
which are designing these tools,
there should be embedded within the teams people
with background in things like ethics, like politics,
that they always think in terms of what biases
might we inadvertently be building into our system,
what could be the cultural or political implications
of what we are building.
It shouldn't be a kind of afterthought
that you create this neat technical gadget,
it goes into the world, something bad happens.
And then you start thinking oh,
we didn't see this one coming, what do we do now?
From the very beginning, it should be clear
that this is part of the process.
- I do wanna give a shout out to Rob Reich,
who just introduced this whole event.
He and my colleagues, Mehran Sahami
and a few other Stanford professors
have opened this course called ethics, computation,
and sorry, Rob, I'm abusing the title of your course.
But this is exactly the kind of classes, it's,
I think this quarter, the offering has more
than 300 students signed up to that.
The course, I wish the course had existed
when I was a student here.
Let me ask an excellent question from the audience
that ties into this, this is from Eugene Lee.
How do you reconcile the inherent trade offs
between explainability and efficacy
and accuracy of algorithms?
- Quick question, this question seems to be assuming,
if you can explain it, you're less good or less accurate.
- Well, you can imagine that if you require explainability,
you lose some level of efficiency, you're adding
a little bit of complexity to the algorithm.
- So okay, first of all, I don't necessarily believe
in that, there's no mathematical logic to this assumption.
Second, let's assume there is a possibility
that an explainable algorithm suffers efficiency.
I think this is a societal decision we have to make,
you know, when we put the seatbelt in our car, driving,
that's a little bit of an efficiency loss
'cause I have to do that seatbelt movement
instead of just hopping and drive.
But as a society, we decided we can afford
that loss of efficiency because we care more
about human safety.
So I think AI is the same kind of technology,
as we make these kind of decisions going forward
in our solutions, in our products,
we have to balance human well being
and societal well being with efficiency.
- So let me, Yuval, let me ask you
the global consequences, this is something
that a number of people have asked about
in different ways and we've touched on,
but we haven't hit head on.
There are two countries, imagine you have country A
and you have country B.
Country A says all of you AI engineers,
you have to make it explainable,
you have to take ethics classes,
you have to really think about the consequences
of what you're doing, you gonna have dinner with biologists,
you have to think about love and you have to like read,
you know, John Locke.
That's group A.
Group B country says, just go build some stuff, right.
These two countries at some point
are gonna come in conflict.
And I'm gonna guess that country B's technology
might be ahead of country A's.
Is that a concern?
- Yeah, that's always the concern with arms races,
which become a race to the bottom in the name
of efficiency and domination.
And we are in a, I mean, what is extremely problematic
or dangerous about the situation now with AI
is that more and more countries are waking up
to the realization that this could be
the technology of domination in the 21st century.
So you're not talking about just any economic competition
between the different textile industries
or even between different oil industries.
Like one country decides to, we don't care
about the environment at all, we'll just full gas ahead,
and the other country's is much more environmentally aware.
The situation with AI is potentially much worse,
because it could be really the technology of domination
in the 21st century, and those left behind
could be dominated, exploited, conquered
by those who forge ahead.
So nobody wants to stay behind.
And I think the only way to prevent this kind
of catastrophic arms race to the bottom
is greater global cooperation around AI.
Now, this sounds utopian because we are now moving
in exactly the opposite direction
of more and more rivalry and competition.
But this is part of, I think, of our job,
like with the nuclear arms race, to make people
in different countries realize that
this is an arms race that, whoever wins, humanity loses.
And it's the same with AI, if AI becomes an arms race,
then this is extremely bad news for all the humans.
And it's easy for say people in the US
to say we are the good guys in this race,
you should be cheering for us.
But this is becoming more and more difficult
in a situation when the motto of the day is America first.
I mean, how can we trust the USA to be the leader
in AI technology if ultimately,
it will serve only American interests
and American economic and political domination?
So it's really, I think most people,
when they think arms race in AI,
they think USA versus China.
But there are almost 200 other countries in the world.
And most of them are far far behind.
And when they look at what is happening,
they are increasingly terrified, and for a very good reason.
- The historical example you've made
is a little unsettling, is if I heard your answer correctly,
it's that we need global cooperation, and if we don't,
we're gonna lead to an arms race.
In the actual nuclear arms race,
we tried for global cooperation from, I don't know,
roughly 1945 to 1950, and then we gave up.
And then we said we're going full throttle
in the United States, and then why did
the Cold War end the way it did?
Who knows, but one argument would be
that the United States, you know, build up
and its relentless buildup of nuclear weapons helped
to keep the peace until the Soviet Union collapsed.
So if that is the parallel, then what might happen here
is we'll try for global cooperation in 2019, 2020, 2021,
and then we'll be off in an arms race.
A, is that likely, and B, if it is,
would you say, well then the US needs
to really move full throttle in AI
because it'd be better for the liberal democracies
to have artificial intelligence than totalitarian states?
- Well, I'm afraid it is very likely
that cooperation will break down and we will find ourselves
in an extreme version of an arms race.
And in a way, it's worse than the nuclear arms race,
because with nukes, at least until today,
countries developed them but never used them.
AI will be used all the time.
It's not something you have on the shelf
for some doomsday war, it will be used all the time
to create potentially total surveillance regimes
and extreme totalitarian systems in one way or the other.
And so from this perspective, I think the danger
is far greater.
You could say that the nuclear arms race
actually saved democracy and the free market and, you know,
rock and roll and Woodstock and then the hippies,
and they all owe a huge debt to nuclear weapons.
Because if nuclear weapons weren't invented, you needed,
there would've been a conventional arms race
and conventional military buildup
between the Soviet bloc and the American bloc.
And that would've meant total mobilization of society,
if the Soviets are having total mobilization,
the only way the Americans can compete is to do the same.
Now, what actually happened was that you had
an extreme totalitarian mobilized society
in the communist bloc, but thanks to nuclear weapons,
you didn't have to do it in the United States
or in Western Germany or in France,
because you relied on nukes, you don't need millions
of conscripts in the army.
And with AI, it's going to be just the opposite.
That the technology will not only be developed,
it will be used all the time.
And that's a very scary scenario.
- Wait, can I just add one thing?
I don't know history like you do,
but you said AI is different from nuclear technology.
I do wanna point out, it is very different because,
at the same as you're talking about
these more scarier situations, this technology
has a wide international scientific collaboration basis
that is being used to make transportation better,
used to improve healthcare, to improve education.
And so it's a very interesting new time
that we haven't seen before, because while
we have this kind of competition,
we also have massive international
scientific community collaboration on these benevolent users
and democratization of this technology.
I just think it's important to see both side of this.
- You're absolutely right, yeah.
There are some, as I said, there are also enormous benefits
to this technology.
- [Fei-Fei] And in a global collaborative way,
especially between, among the scientists.
- The global aspect is more complicated
because the question is, what happens
if there is a huge gap in abilities
between some countries and most of the world?
Would we have a rerun
of the 19th century industrial revolution,
when the few industrial powers conquer
and dominate and exploit the entire world,
both economically and politically?
What's to prevent that from repeating?
So even in terms of, you know,
without this scary war scenario, we might still find ourself
with a global exploitation regime in which the benefits,
most of the benefits go to a small number of countries
at the expense of everybody else.
- Have you heard of archive.org?
- So students in the audience might laugh at this,
but we are in a very different scientific research climate,
is that the kind of globalization of technology
and technique happens in the way that the 19th century
and even 20th century never saw before.
Any paper that is a basic science research paper
in AI today that is, or technique that is produced,
let's say this week at Stanford,
it's easily get globally distributed through
this thing called Archive or GitHub repository or this--
- The information is out there, yeah.
- The globalization of this scientific technology travels
in a very different way from the 19th and 20th century.
I mean, I don't doubt there are, you know,
confined development of this technology maybe by regimes.
But we do have to recognize that this global,
the difference is pretty sharp now,
and we might need to take that into consideration,
that the scenario you are describing is harder.
I'm not saying impossible, but harder to happen.
- I'll just say that it's not just the scientific papers.
Yes, the scientific papers are there.
But if I live in Yemen or in Nicaragua
or in the Indonesia or in Gaza, yes, I can connect
to the Internet and download the paper,
what will I do with that?
I don't have the data, I don't have the infrastructure.
I mean, you look at where the big corporations
are coming from that hold all the data of the world,
they are basically coming from just two places.
I mean, even Europe is not really in the competition.
There is no European Google or a European Amazon
or European Baidu or European Tencent.
And if you look beyond Europe,
you think about Central America,
you think about most of Africa, the Middle East,
much of Southeast Asia, it's, yes,
the basic scientific knowledge is out there,
but this is just one of the components that go
to creating something that can compete
with Amazon or with Tencent or with the abilities
of governments like the US government
or like the Chinese government.
So I agree that the dissemination of information
and basic scientific knowledge, we're at
a completely different place than in the 19th century.
- Let me ask you about that, 'cause it's something three
or four people have asked in the questions, which is,
it seems like there could be a centralizing force
of artificial intelligence, that it will make whoever
has the data and the best compute more powerful,
and that it could then accentuate income inequality,
both within countries and within the world, right,
you can imagine the countries you've just mentioned,
the United States, China, Europe lagging behind,
Canada somewhere behind, way ahead of Central America.
It could accentuate global income inequality.
A, do you think that's likely, and B,
how much does it worry you?
We've got four people who've asked a variation on that.
- Well, as I said, it's very very likely,
it's already happening.
And it's extremely dangerous, because the economic
and political consequences could be catastrophic.
We are talking about the potential collapse
of entire economies and countries.
Countries that depend say on cheap manual labor,
and they just don't have the educational capital
to compete in the world of AI.
So what are these countries going to do?
I mean, if, say, you shift back most production from,
say, Honduras or Bangladesh to the US and to Germany,
because the human salaries are no longer part
of the equation, and it's cheaper
to produce the shirt in California than in Honduras,
so what will the people there do?
And you can say okay, but there will be many more jobs
for software engineers.
But we are not teaching the kids in Honduras
to be software engineers.
So maybe a few of them could somehow immigrate to the US.
But most of them won't, and what will they do?
And we, at present, we don't have the economic answers
and the political answers to these questions.
- Fei-Fei, you wanna jump in?
- I think that's fair enough.
I think Yuval definitely has laid out
some of the critical pitfalls enough,
and that's why we need more people to be studying
and thinking about this?
One of the things we over and over noticed,
even in this process of building the community
of human-centered AI and also talking to people,
both internally and externally, is that
there are opportunities for business around the world
and governments around the world
to think about their data and AI strategy,
there are still many opportunities for, you know,
outside of the big players in terms of companies
and countries to really come to the realization
it's an important moment for their country,
for their region, for their business
to transform into this digital age.
And I think when you talk about these potential dangers,
the lack of data in parts of the world that
hasn't really caught up with this digital transformation,
the moment is now, and we hope to, you know,
raise that kind of awareness and to encourage
that kind of information.
- Yeah, I think it's very urgent, I mean,
what we are seeing at the moment is, on the one hand,
what you could call some kind of data colonization.
That the same model that we saw in the 19th century
that you have the imperial hub where they have
the advanced technology, they grow
the cotton in India or Egypt, they send the raw materials
to Britain, they produce the shirts,
the high tech industry of the 19th century, in Manchester,
and they send the shirts back to sell them in India
and out compete the local producers.
And we, in a way, might beginning to see the same thing now
with the data economy, that they harvest the data
in places also like Brazil and Indonesia,
but they don't process the data there,
the data from Brazil and Indonesia goes to California
or goes to eastern China, being processed there,
they there produce the wonderful new gadgets
and technologies and sell them back as finished products
to the provinces or to the colonies.
Now, it's not a one to one, it's not the same,
there are differences.
But I think we need to keep this analogy in mind.
And another thing that maybe we need to keep in mind
in this respect I think is the reemergence of stone walls
that I'm kind of, you know, originally, I was,
my speciality was medieval military history.
This is how I began my academic career,
with the crusades and castles and knights and so forth.
And now I'm doing all these cyborgs and AI stuff.
But suddenly, there is something that I know from back then,
the walls are coming back.
And I try to kind of, what's happening here?
I mean, we have future realities, we have 3G, AI,
and suddenly, the hottest political issue
is building a stone wall.
Like the most low tech thing you can imagine.
And what is the significance of a stone wall
in a world of interconnectivity and all that?
And it really frightens me that there
is something very sinister there, the combination of data
is flowing around everywhere so easily,
but more and more countries, and also my home country
of Israel, it's the same thing, you have the, you know,
the startup nation, and then the wall.
And what does it mean, this combination?
- Fei-Fei, you wanna answer that?
- Maybe you can look at the next question.
- You know what, let's go to the next question
which is tied to that.
And the next question is, you have the people here
at Stanford who will help building these companies,
who will either be furthering a process
of data colonization or reversing it
or who will be building, you know,
the efforts to create a virtual wall
in a world based on artificial intelligence
that are being created, funded at least,
by a Stanford graduate.
So you have all these students here in the room.
What do you want them to, how do you want them
to be thinking about artificial intelligence
and what do you want them to learn,
let's spend the last 10 minutes of this conversation
talking about what everybody here should be doing.
- So if you're a computer science or engineering student,
take Rob's class.
If you're a humanist, take my class.
And all of you, read Yuval's books.
- Are his books on your syllabus?
- Not on my, sorry.
I teach hardcore deep learning.
His book doesn't have equations.
- I don't know, B plus C plus D equals H.
- But seriously, you know, what I meant to say
is that Stanford students, you have a great opportunity,
this is, we have a proud history of bringing this technology
to life, Stanford was at the forefront of the birth of AI,
in fact, our very professor John McCarthy coined
the term artificial intelligence
and came to Stanford in 1963 and started this nation's,
one of the two oldest AI labs in this country.
And since then, Stanford's AI research
has been at the forefront of every wave of AI changes.
And this 2019, we're also at the forefront
of starting the human-centered AI revolution or
the writing of the new AI chapter.
And we did all this for the past 60 years for you guys,
for the people who come through the door
and who will graduate and become practitioners,
leaders, and part of the civil society.
And that's really what the bottom line is about.
Human-centered AI needs to be written
by the next generation of technologists
who have taken classes like Rob's class to think about
the ethical implications, the human wellbeing.
And it's also gonna be written by
those potential future policymakers
who came out of Stanford's humanities studies
and been in this school who are versed
in the details of the technology, who understand
the implications of this technology, and who has
the capability to communicate with the technologies.
That is, no matter how we agree and disagree,
that's the bottom line, is that we need
these kind of multilingual leaders and thinkers
and practitioners, and that is
what Stanford's Human-Centered AI Institute is about.
- Yuval, how do you wanna answer that question?
- On the individual level, I think it's important
for every individual, whether in Stanford,
whether an engineer or not, to get to know yourself better.
Because you're now in a competition.
You know, it's the oldest advice in the book
in philosophy is know yourself.
We're heard it from Socrates, from Confucius,
from Buddha, get to know yourself.
But there is a difference, which is that now,
you have competition.
In the day of Socrates or Buddha,
if you didn't make the effort, so okay,
so you missed on enlightenment.
But still, the king wasn't competing with you.
They didn't have the technology, now you have competition.
You're competing against these giant corporations
If they get to know you better than you know yourself,
the game is over.
So you need to buy yourself some time,
and the first way to buy yourself some time
is to get to know yourself better,
and then they have more ground to cover.
For engineers and students, I would say,
I'll focus on engineers, maybe.
The two things that I would like to see coming out
from the laboratories and the engineering departments
is first, tools that inherently work better
in a decentralized system than in a centralized system.
I don't know how to do it, but if you,
I hope there is something that engineers can work with.
I heard that blockchain is like the big promise
in that area, I don't know.
But whatever it is, part of, when you start designing
a tool, part of the specification of what this tool
should be like, I would say this tool should work better
in a decentralized system than in a centralized system.
That's the best defense of democracy.
The second thing that I would like to see coming out--
- I don't wanna cut you off 'cause I want you
to get to this second thing, how do you make
a tool work better in a democracy than--
- I'm not an engineer, I don't know.
- [Nick] Okay.
All right, we'll go to part two.
Take that, someone in this room, figure that out,
'cause it's very important--
- I can think about it and then,
I can give you historical examples of tools
that work better in this way or in that way.
But I don't know how to translate it into present day--
- Go to part two, 'cause I got
a few more questions asked from the audience.
- Okay, so the other thing though
I would like to see coming is an AI sidekick
that serves me and not some corporation or government,
so to take all, I mean, we can't stop
the progress of this kind of technology.
But I would like to see it serving me.
So yes, it can hack me, but it hacks me
in order to protect me.
Like my computer has an antivirus, but my brain hasn't.
It has a biological antivirus against the flu or whatever,
but not against hackers and trolls and so forth.
So one project to work on is to create an AI sidekick,
which I paid for maybe a lot of money and it belongs to me.
And it follows me and it monitors me
and what I do and my interactions.
But everything it learns, it learns in order to protect me
from manipulation by other AIs,
by other outside influencers.
So this is something that I think,
with the present day technology,
I would like to see more effort in the direction.
- Not to get into too technical terms,
I think you would feel comforted to know that
the budding efforts in this kind of research is happening,
you know, trustworthy AI, explainable AI,
security, you know, motivated or aware AI.
So I'm not saying we have the solution,
but a lot of technologists around the world
are thinking along that line and trying to make that happen.
- And it's not that I want an AI that belongs to Google
or to the government that I can trust,
I want an AI that I'm its master, it's serving me.
- And it's powerful, it's more powerful than my AI,
'cause otherwise, my AI could manipulate your AI.
- It will have the inherent advantage
of knowing me very well.
So it might not be able to hack you,
but because it follows me around and it has access
to everything I do and so forth, it gives it an edge
in the specific realm of just me.
So this is a kind of counterbalance
to the danger that the people with--
- But even that would have a lot of challenges
in our society, who is accountable for,
are you accountable for your action or your sidekick?
- This is going to be a more and more difficult question
that we will have to deal with.
- The sidekick defense.
Fei-Fei, let's go through a couple questions quickly.
We often talk, this is from Reagan Pollack,
we often talk about top down AI from big companies,
how should we design personal AI
to help accelerate our lives and careers?
The way I interpret that question is,
so much of AI is being done at the big companies.
If you wanna have AI at a small company or personally,
can you do that?
- Well, first of all, one solution
is what Yuval just said.
- Probably those things will be built by Facebook.
- So first of all, it's true, there is a lot of investment
and efforts putting, and resource putting big companies
in AI research and development, but it's not
that all the AI is happening there,
I wanna say the academia continue to play a huge role
in AI's research and development,
especially in the long term exploration of AI.
And what is academia?
Academia is a worldwide network of individuals,
students, and professors thinking very independently
and creatively about different ideas.
So from that point of view, it's a very grassroot kind
of effort in AI research that continues to happen.
And small businesses and independent research institutes
also have a role to play, right.
There are a lot of publicly available datasets,
we, it's a global community that is very open
about sharing and disseminating knowledge and technology.
So yes, please, by all means,
we want global participation in this.
- All right, here's my favorite question,
this is from anonymous, unfortunately.
If I am in eight grade, do I still need to study?
- As a mom, I will tell you yes.
Go back to your homework.
- All right, Fei-Fei, what do you want Yuval's next book
to be about?
- Wow, I didn't know this, I need to think about that.
- All right, well, while you think about that,
Yuval, what area of machine learning
do you want Fei-Fei to pursue next?
- The sidekick project.
- Yeah, I mean, just what I said, an AI,
can we create a kind of AI which can serve individual people
and not some kind of big network?
I mean, is that even possible or is there something about
the nature of AI which inevitably will always lead back
to some kind of networked effect
and winner takes all and so forth?
- [Nick] All right, we're gonna wrap with Fei-Fei--
- His next book is gonna be a science fiction book
between you and your sidekick.
- All right, one last question for Yuval,
'cause we've got two of the voted questions are this.
Without the belief in free will,
what gets you up in the morning?
- Without the belief in free will?
I don't think that the question of, I mean,
is very interesting or very central,
it has been central in Western civilization
because of some kind of basically theological mistake
made thousands of years ago.
But it's a really, it's a misunderstanding
of the human condition.
The real question is how do you liberate yourself
And one of the most important steps in that direction
is to get to know yourself better, and for that,
you need to just push aside this whole, I mean, for me,
the biggest problem with the belief in free will
is that it makes people incurious about themselves
and about what is really happening inside themselves.
Because they basically say, I know everything.
I know why I make decisions, this is my free will.
And they identify with whatever thought or emotion pops up
in their mind because hey, this is my free will.
And this makes them very incurious about
what is really happening inside and what is also
the deep sources of the misery in their lives.
And so this is what makes me wake up in the morning,
to try and understand myself better,
to try and understand the human condition better.
And free will is just irrelevant for that.
- And if we lose it, your sidekick
can get you up in the morning.
Fei-Fei, 75 minutes ago, you said
we weren't gonna reach any conclusions,
do you think we got somewhere?
- Well, we opened a dialogue between the humanist
and the technologist, and I wanna see more of that.
- Great, all right, thank you so much, thank you, Fei-Fei,
thank you, Yuval Noah Harari, it was wonderful to be here,
thank you to the audience.