The Rise of Digital People: Mustafa Suleyman and Ian Bremmer on the Future of AI

♪ [music] ♪

- [Ian] In the past year,

artificial intelligence
has captured our imagination

like never before
and like nothing else.

It's transformed everything
from how we work,

to how we perceive
the world around us,

to increasingly
who we are as people.

And with this technological
renaissance,

there are very few names

that stand out as prominently
as Mustafa Suleyman.

He has a career
that's deeply rooted in AI.

He's been at the forefront
of this field

for well over a decade,

co-founding DeepMind

and playing a critical role

in its mission
to solve intelligence

and to use it
to make the world a better place.

And today,
we're sitting down with Mustafa

to discuss
both the exciting potential of AI,

as well as concerns
about the future of AI.

Mustafa, thank you for joining us.

Mustafa, you've been working
on artificial intelligence

for almost 20 years now, right?

So what's happened
that has made this

such an omnipresent thing
in all of our lives?

- [Mustafa] Well,
for much of that time,

like you said,

I've been among a small group
of very fringe AI researchers

who mostly have been considered
to be a little bit crazy.

Back in 2010,
when I first started DeepMind,

most people who heard
that I was working on AI

thought that I was really

nothing to do
with mainstream culture,

a weirdo futurist,

and working on something
that was very speculative.

So for as long as I can remember,

we've been an outsider.

And in the last few years,
I would say,

it's grown
in the popular imagination.

I think a lot of that started
several years back with AlphaGo,

which was an AI that we designed

to play the ancient Chinese
board game of Go,

which is played
on a 19 x 19 square grid.

- It's a lot more complicated
than chess, absolutely.

- Yeah. It was
the next big frontier for AI

after IBM beat the game of chess
with Deep Blue back in 1997,

and obviously, as everyone knows,
that's on an 8 x 8 grid,

and the pieces can only move
in very fixed ways,

whereas on a 19 x 19 grid,
not only is it that much larger,

but all the pieces are equal.

There's just black
and white stones,

and you can place them
anywhere on the grid.

So the number of possible moves
that can happen in the board game

are just exponentially larger,

something like 10^170 possible
configurations of the board.

That is more atoms than there are
in the known universe.

So it's a 10 with 170 zeros
next to it, after it.

It's just an incredibly insane
possible number of configurations.

So all the traditional methods
of rule-based search,

where you would say, "If there are
some pieces in this area,

then don't place them there,
but place them adjacent" --

which is traditionally how people
trained AIs to play chess --

those methods didn't work

because the size of the space
was just so huge.

So we had to invent
these learning methods,

and over the years,

we've applied
those learning methods

to other domains.

We started off with games,

then moved into image recognition,

then moved
into audio transcription,

so you can write down
the words that I say phonetically.

And then,
in the last couple of years,

coming back to your question

of why things have gone crazy
in the last year or so,

is because we've actually
been able to apply

a similar suite of methods --

deep learning --

to generate new text
that is unique, right?

And that's the incredible thing
that has happened here.

We've gone from classification --

so understanding
the content of images,

understanding that two languages
translate in the following way,

understanding the content

of a piece of text
or a paragraph --

to then being able to generate
a new example of that paragraph,

or a new image,
or some new speech or music

at human-level performance.

A lot of these AIs now

are pretty much as good
as most humans

at being creative
or answering questions.

- So nowadays,

we can have a conversation
with an AI bot,

and we can't really tell
the difference

between it and another human being.

That's a fair point, right?

- Yeah. That is one
of the surreal moments

that we live in.

When you talk to one of these AIs,

like ChatGPT,

or my own company,
Inflection, for example,

makes an AI called Pi,

p-i, which stands for
"personal intelligence."

When you talk to Pi,

it's just like chatting
to a regular human.

You can actually phone Pi,

and it will speak to you

in a very fluent, smooth,
conversational voice.

It has empathy.
It has emotional intelligence.

And yeah, in many respects,
it's just like talking to a human.

And so many people would say

that we have nearly passed
the famous Turing test,

which was proposed
back in the 1950s

by this great computer scientist
and mathematician

called Alan Turing,

who said that if you could design
a computer

to speak to another human,

and it would be impossible to tell

whether it was actually a computer
or a human speaking,

then we could say that that AI
or that computer system

was intelligent
and it had passed the Turing test.

And now, we're pretty much
at that moment, I think.

It's still possible to tell,
if you really pay attention

or if the conversation
goes on long enough.

But if it's only just two or three
or five turns of conversation --

it's very difficult to tell.

And that's an amazing moment.

- It's not that the AI
is intelligent.

It's just that we perceive it
to be intelligent.

In reality,

it's just taking all of this data
and pattern recognition,

and it's predicting things

on the basis of the prompt
that we write in, right?

It's a program
at the end of the day.

It's not thinking.

- Well, it's interesting,

because with every new technology,

we're forced to reconsider
our basic assumptions.

So we thought we had a good grasp
of what intelligence is

because it was compared
to passing the Turing test.

Whereas now we've passed
the Turing test,

people are like,

"Well, maybe it's not
that intelligent after all," right?

So maybe it wasn't
a very good test.

And that is the process of science,
is that we posit a hypothesis,

we develop some experiments
to test that hypothesis,

and then, we review the evidence,
and we generate a new test.

And so today,

we have to generate a new test,

because it's pretty clear

that these things
are not really intelligent,

even though they're very capable.

So I've proposed a new test,
a modern Turing test,

which actually evaluates
what an AI can do,

not just what it can say.

So I think
a better measure of an AI

is an artificial
capable intelligence,

an ACI.

So what can this thing do
in the labor market?

What jobs can it do?

Can it write emails?

Can it negotiate contracts?
Can it invent new products?

Can it market and promote a product
and persuade people to buy it?

And if it can do
all of those things

and do it in a way

that enables it to make
a profit on a product --

that's like a mini entrepreneur.

It's like a little startup person.

And I think that over the next
three or four years,

there will be AIs that can do
all those things I've described

and actually turn a real profit
from a new product.

And that'll be a watershed moment,

because that's many of the skills

that a lot of people use
in their day-to-day jobs.

- So I want to get to the future,

but let's start with the present
just for a moment.

For young people today

that are hearing
about artificial intelligence,

as so many of us are,
for the first time,

if they're curious,

what should they be doing
with artificial intelligence?

How should a person start engaging

in a way that will be constructive
and useful for their future?

- Well, the first thing to say
is that these AIs

have all of the knowledge

that has been put
on the open internet.

So they're actually
extremely smart.

Not only have they been trained
on Wikipedia many, many times over,

but they've read
millions of blog posts,

millions of news articles,

many books, in many cases,
that are available on the open web,

and so, they're very knowledgeable.

So the first thing to do
is just talk to one.

Pick a topic
that you are really interested in,

maybe one that you know
something about,

and try to test the limits
of the AI's knowledge

by probing it, and questioning it,
and going back and forth.

I'm sure many people
have already tried them out.

And I think, you know,
give Pi a go.

It's really
an incredible experience.

And once you play around with it
a little bit,

you realize A) the magic
and how impressive it is,

but B) where it trips up.

Sometimes it goes in circles.

Sometimes it doesn't remember
things correctly.

And that gives you an intuition

for where the cutting edge
is today,

where it's weak
and where it's strong.

And then, I would say,
if you're really interested,

is try and prompt one of these AIs.

Try and give it a stylistic guide.

Try and get it to talk
in the style of President Obama,

or one of your favorite
celebrities, or Shakespeare.

And invent something with it,

treat it as an aide,
a creative brainstorming partner,

and then you can see, again,
what the shape of it is.

And if you want to go even further,

many of these models

are available
in open-source software,

and you can have a go
at trying to program one of them.

- And what's something
that you could program

an AI today to do
that would be interesting?

You could program it, for example,

to be an expert
in Formula One racing

and talk to you
in the style of Shakespeare,

if you like.

It can embody
the character and persona

of anything that you can imagine,

and take advantage
of the depth of knowledge

that it has been trained on.

So it might be an expert
in talking about cactuses.

It might know everything
about Harley Davidsons.

It might know
all about the dinosaurs.

Anything that you can think of,

it is going to be able to imitate
that knowledge

in a particular style.

And so, the world is your oyster.

You can create game characters.

You can create little aides,

or assistants, or friends

to play with or to talk to,

fiction that you can
co-write together --

you write part of the story,

the AI writes
the other part of the story.

So, it's really limitless
what can be done with these things.

- You and I are great optimists

on how much AI is doing
and where this technology is going.

But we're also very well aware

that with all
of these opportunities

to program AI
to do incredible things,

if you want to program it
to do bad things,

you probably can.

And you can use it
to learn and promote information,

but you can use it
to promote disinformation,

and to fake people,

and to get people
to believe things that aren't true.

What do we need to do
to limit the potential

for AI to be used in ways
that are dangerous for our society?

- Well, one of the things is that,
at the moment,

many of the AI services
are available

through large,
established providers,

and those providers
have all committed

to responsible
and ethical principles.

And it's incumbent on all of us
to hold those providers to account

based on what they've said --

so about not spreading
misinformation,

not spreading
factually untrue content,

and crucially, I think,

not imitating
a known public figure.

Because what we don't want
is to have a bunch of AIs

where in the future

it will be impossible
to tell whether a celebrity,

or a politician,
or a business person

had in fact given a message,
or said something,

made a recording,
issued a statement,

but in fact,
it was actually a deep fake,

it was a made-up piece of content.

And at the moment,

most of the big providers
of these AIs are very responsible

and take lots of efforts
to prevent those kinds of things.

I think, in the future,
these models are going to be

increasingly available
in open source,

and so that's going to get harder

to contain and to moderate.

- And when you say in open source,

you mean
you're not going to just get it

from Meta, or Google,
or Microsoft, or Inflection,

that you could get it
just like on the open web.

And then the rules are

whatever the rules happen to be
for that provider.

- Yeah. And I think
that's going to become

more and more a challenge
of the open Internet,

is, where are
the boundaries of that,

and how does it get restricted?

Because you're going to be able
to take this software

and run your own AI

independent of any big provider
in five years' time.

Today, you can run them
in the open source,

and they're pretty good,

but in the future,

they're going to be
really, really good.

So we just have to think
about the right way to make sure

that we have stable
and peaceful outcomes

and that the transition
to this new AI future

doesn't happen too quickly
and isn't too chaotic,

because, as we've seen in the past,

sometimes there can be
unintended consequences.

- Now, we aren't yet at the point

that we would use an AI
to be our lawyer.

We might use it to help our lawyer,

but we wouldn't use it to be it,

or to be our teacher,
or to be our nurse.

But it sounds like,
in very short periods of time,

you believe that AI
is going to be able to replace

a lot of these functions.

Take us a little bit
along that path.

Not all the way yet,
but just like the next year or two.

Because we keep seeing
all of these new announcements,

and you said,
"Well, AI can generate text,

and it can sound
like a human being."

Yeah. You can have
a conversation with it.

Well, now we see AI
can generate images,

and there are images
that are more impressive

than almost any artist
or graphic design.

Now we see, just recently,
AI can generate video

and can generate a movie

that you would see
in Hollywood, maybe,

or really close.

What's coming next?

What do we see
in the next year or two

that's going to blow our minds

that we're going to start
using everywhere?

- I think that the reality is

that over the next two
to three years,

we are going to be surrounded
by a new species of digital people.

We have wrestled
with different metaphors

for describing
this new technology era

for many decades,

and none of them seem to be
sufficient or up to the task.

Some people have compared AI

to another general-purpose
technology like electricity.

General purpose
because it's like a raw commodity

that enables
many, many other technologies,

and products, and services
to be built on top of it.

Who could think of living
in a modern world today

without electricity?

Going back even further,

the printing press was an earlier
general-purpose technology

because it enabled anybody

to broadcast their ideas,
and organize, and plan, and so on.

So many meta capabilities arose
because of that platform.

- Or the Internet. Same thing.

- Or indeed the Internet.

That's another good example
of a general-purpose technology.

But today, it's hard to say that AI
is a general-purpose technology.

It certainly is,
but that's not all it is.

It is interactive.

It produces new
and dynamically emergent content

in a personalized way,
uniquely to you at every moment.

That is very different

to the very predictable nature
of electricity, for example.

The actual infrastructure
of the Internet

was very stable and predictable.

We know exactly
how many packets can be sent

across a certain wire
of a certain speed

at a certain time.

Whereas here,

this is like a completely
new design material.

No two answers
to the same question

will be the same.

Every interaction
is very different.

And now that these interactions

are becoming completely dynamic --

you say something,

the AI says something,
you say something --

it's actually
much more like talking

to a full digital person.

So in two or three years' time,

there will be an avatar
that will be a human-like

or other kind
of character representation

that is very animated

and just like you or I
speaking to one another now.

As you say, it will be able
to generate video in real time

completely seamlessly
on your phone,

on your desktop,
on your tablet, in your car.

And so, rather than browsing
a web page,

today, when you go
to look for information,

you type a query into Google,

and you get a static web page

that was probably made
two years ago,

or maybe even some cases,
five years ago.

And that's like a billboard.

It's a static representation
that doesn't change.

And it certainly doesn't change
to adapt to you or me.

It's just we both see
the same thing.

You type in a website.

We both see
exactly the same thing,

regardless of the time of day,
or the location,

or back history,
or what we're interested in.

In the future,

content is going to be served
to every individual

in a completely personalized
and interactive way.

So your web page
of images, and text, and video

is going to unfurl itself
on the fly,

completely novel,

adapted to your interests

and what you've talked about
previously with your AI.

And that's just
a completely different paradigm

that I think people
are not quite yet grasping.

- Mustafa, when you say
"digital people,"

what do you mean by that?

- Well, if you think about it,

what makes a person a person

is my ability
to speak to you right now,

my ability to see what you see,
and my ability to take actions,

so buy things, book things,

plan, arrange, coordinate,
write emails, make phone calls.

At some point
in the next few years,

an AI is going to be able to do
all of those things

pretty much as well as a human.

But of course,
it won't be a human.

It will be a digital person.

And I think that's probably
the best metaphor

to help us understand what's coming
over the next few years.

- What does it mean

as we move into an environment

where so many of the interactions
that we will have

will be with digital people,

as opposed to people people?

How do you think
that changes the economy?

How does it change society?

How does it change government?

What are some of your thoughts
about that?

- Well, one of the amazing things
about these digital people

is that they can actually be made
to be very controlled.

You can actually design
very precise behaviors.

And so, for example,
in the AI that we've made, Pi,

it is very kind and empathetic.

It's very supportive.
It's very encouraging.

It's infinitely patient.

It doesn't judge you.

And so many of the downsides
of human interaction,

where you might feel
socially anxious,

you might feel a bit paranoid

about what the other person's
thinking,

you might feel pushed around.

Or you might feel

that that other person
didn't hear you out.

You were telling a story
about your ski trip,

and suddenly, they're talking
about their ski trip

that they had last year,

and you're like, "Wait,
but I haven't finished my thought."

Your AI doesn't do that to you.

Your AI is infinitely patient
and supportive.

And so, there's a huge amount
of upside there,

but it's also
a big transition, right?

Because I think, increasingly,

people will choose
to spend time with their AIs,

perhaps more than they spend
time with other humans.

And so one
of the design considerations

that we have to factor in,

and we think about a lot
at Inflection,

is to really pay attention

to the values of the AI

and how we condition
and shape the AI,

for example, to encourage you

to spend more time
with your loved ones,

to encourage you to be brave,

and overcome your social anxiety
and go to the party,

to provide you with a safe space

to practice for your interview
or your exam,

but still have you focused

on being out in the real world,
connected,

having experiences
with other humans.

And so, every single discipline,
every single area of society

is going to have to grapple
with this new reality

that there will in fact
be digital people

that are as significant
and as important

as every other relationship
in our lives.

It would be impossible
to consider today

not having a smartphone
in your life or a laptop.

That's just become second nature
in less than a decade.

- Yes.

- Six billion people have
a smartphone now or more.

And so, that's probably
the trajectory we're on

for these personal intelligences.

This is the natural evolution
of technology

from personal computing
to personal intelligence.

- And I think people
will be relieved to hear you say

that you find it important
that the values of these AI people

that you are developing,
that you're inventing,

needs to be humane

and needs to keep people
engaging with other people.

Of course,

you and I can both imagine

that there are going to be
lots of corporations

that want to maximize profitability

and therefore ensure that people
are engaging with their AI

as much as humanly possible,

just as some companies do

with their smartphones
or their applications,

just as sometimes you want to sell

as much food
as you possibly can to a person,

so even if it means
that they're obese.

All of these things.

And I wonder, do you think that --

How do we guard
against the excesses

that comes from a technology
that is changing so much faster

than our ability to understand it,
train with it, prepare for it?

We're going to be mostly
the people we are right now,

and these things are suddenly
just going to be "poof!" around us.

It's not like we can train people

to, like, "Okay, you've got to grow
into becoming an adult,

and here's what it is."

It's just going to be there.

- Well, look, I think that
we are making incredible progress

as a civilization,

including on the corporation front.

Society is changing much faster
than I think people fully realize.

If you roll back to the 50s,

and the kinds of companies
that we had,

and the kind of way

that they would externalize
their downsides --

whether it was dumping chemicals
into the river

or really mistreating their staff
in horrible ways --

I think that it's inconceivable
that we would have companies

that are really pushing smoking

in the way that they did
back in the day,

or really pushing
obesity and fatty foods.

We're really, I think,
making a march forward,

and look,
it's not a solved problem.

Fundamentally,

companies are constantly
in a battle

to try to be more responsible

and to be more considerate
and respectful of their people.

All I can say is that,
for my part, at Inflection,

we've tried to structure
the company

in a way that we proactively think
about those consequences,

and we're actually
registered legally

as a public benefit corporation.

- Oh, okay.

But now, let's go maybe
five years in the future --

not ten, not twenty, just five --

where we are already
starting to see

AI that is able to do
a lot of the jobs

that people have today,

that young people are thinking
about having in the future.

How should someone
considering their career

adapt to a future of AI

that is so explosive,
so transformative,

and so near-term,

so much uncertainty
about what society will look like,

more than ever
in the history of human beings?

How do young people
think differently

about preparing themselves
for the future?

- Well, the cool thing about it
is that more so than ever before,

these technologies are accessible
and programmable

by people who don't have
technical skills.

So you don't have to be
a professional engineer

to be able to play
with an AI model.

You can prompt it
using your own language

with your own ideas.

That means that you can bring
all your creativity

and all of your inventiveness

to an off-the-shelf AI model
very, very easily.

So that's the first thing
I would say is, "Don't be afraid

because people think

that there's a big
technical barrier to get over" --

that's not true anymore.

Secondly,

there's clearly a huge benefit
to people who come from outside

of software engineering
and technical fields

actually playing with these things.

So we need
more wide-ranging voices,

people with different backgrounds

grabbing hold of this stuff
and making new things with it,

because all perspectives
are needed at this time.

This is a massive transition,

just like the arrival
of the Internet

or the arrival of light.

Can you imagine what it was like

for the people who first
saw a light bulb turn on

and saw the power of electricity?

That sparked a revolution
in people inventing things,

in microelectronics,

and in myriad ways to make
our lives more comfortable.

So think of this as a creative
and exciting moment

to be an inventor
and to use these tools

to basically make
our dreams come true.

This is a great moment.

- Well, let me ask,

all of these transformations
also create disruptions.

Some of those disruptions
are super opportunities.

Some of them are more challenging
if you're not ready.

Give us a couple of jobs

that you think people
shouldn't actually want to get into

in the next five or ten years --

because they're not going
to be around with AI --

and why.

- Well, any new technology
disrupts certain jobs, right?

So, it's pretty clear

that call center operators,
for example,

who manage
customer service requests

or even who do sales,
for example --

there are going to be AIs

that try to do that
more efficiently,

and we're already seeing
those kinds of things,

so I'd be worried
about those kinds of jobs.

I think that the most valuable
skill sets

are going to be those
that straddle creative skills,

as well as problem-solving
and technical.

So breadth is
more important than ever.

That's one of the things that AIs
don't do so well

is integrate a wide range
of different skills

into a single source.

So I would say be bold

and do multidisciplinary
educational courses

that teach you the best of both.

- So Mustafa, this is a technology

that will improve
human capabilities

to anyone that has them.

It'll lead to faster inventions,

reduced waste,
more efficiency in every field.

As you say, it's more than just
a transformative technology.

It also changes how we think
about the entire world and society.

I'm wondering,

for young people today,

who might think about having 50,
60, 70 more years on the planet,

how do you think
about their future?

Will they even be recognizable
as human beings

when they're adults?

Do you think they'll have
limitless lifespan?

When I think about applying
artificial intelligence to medicine

and to biotechnology, and genetics,

it really does seem staggering

how much the world could change

from what we can
even imagine today.

- I think that's true.

Technology and the scientific
process of invention

is there to reduce our suffering.

It's there to make our lives
more peaceful and more enjoyable.

Just a few hundred years ago,

the average life expectancy
was closer to 50 years old.

So science has massively advanced
our well-being and health

because we've invented drugs,

and we've found ways
to get more crops, for example,

out of the same square hectare.

I think that is
an incredible achievement

of creativity and invention.

And we're about to take
that engine of creativity,

which was our human intelligence,

and turn that into a commodity.

We're going to make it
widely available

to millions of people

to be able to be creative
and inventive.

So I think you're totally right.

By 2050, I wouldn't be surprised

if there were people
who were being born

that might live
for 200 to 300 years,

as we are going to make
very fundamental breakthroughs

in medical science

that tackle aging,
that cure disease.

And so that raises the question,

"Well, what do we do
with our lives?

How do we live?

If a large chunk of people

are not working for their income
for most of the day,

then how do we find
meaning and purpose?"

And people often ask me
this question,

and I actually think,

"Well, remember
when you were young,

and you had hobbies, and passions,
and ambitions, and desires,

and you were obsessed
with things? --

that creativity and playfulness

is increasingly going to be
available to adults

and people of all ages."

Because, I think,
in the long-term future,

the real challenge for us
is figuring out

how we support people
when they don't work

and maybe give them
a universal basic income,

so that for at least
a portion of their week,

they're freed up
from the everyday grind of work

to take care of their families,
to look after the elderly,

to be more involved
in bringing up children,

to support community service,

to be creative.

I think that's
an important thing to remember.

We didn't create society

in order to artificially invent
jobs and work for the sake of it.

We work so that we can play, right?

And so the goal is to reduce

the amount of work
that we are forced to do

and increase the amount of time
that we have for play

and for chosen work.

You may choose
to still work super hard --

that could be a choice.

It could be very productive.

It could be very creative
and inventive.

But I think
that's a much better society

that we want to live in,

where most people,
most of the time,

are choosing how to spend
most of their week.

- Yeah. And it feels
like the pandemic

has in some ways helped us think
about that transition, right?

You already have people who say,
"You know what?

I don't want to be in the office
from nine to five every day

with a one-hour commute both ways.

I actually want to spend
more time with my family,

spend more time
with my friends, with my pets,

engage more,

live in a place
that I'm happy living."

It turns out that the pandemic

and the technology that came
from distance engagement,

Zoom calls, all the rest,

technology facilitated
more independence of choice

for people all over the world
to live the way they want to live

and balance their lives
with what they do for a living.

And artificial intelligence
just turbocharges that.

Is that the way
you're thinking about this?

- Yeah. I think that we got
a good taster of prioritization

during the pandemic,

and in fact, that's actually
a really interesting point

because it shows you

how many unquestioned assumptions
are actually buried

underneath the structure
of society.

Who would have thought

that we could actually work
entirely remote

and be pretty productive?

It's not that it was
without consequences,

but the world carried on,

and there are actually
huge benefits now

to working part-time

and not being five days
in the office.

Huge, huge benefits.

So I think that --

Who would have thought
that actually that was something

that we could re-engineer
in terms of society?

So I'm very interested
in this question.

What are the other things
that we take for granted today?

What are the silly rules?

What are the silly social habits,

and customs, and practices,
and structures

that just are because they were

that we could re-engineer,
and turn upside down,

and make them more favorable

to every single one of us
as individuals?

- And it turned out
we could re-engineer it immediately

because we had to,
because the pandemic forced it.

It didn't take years.
It took weeks.

It took weeks, and people showed

they could actually work
completely differently.

I can think of a few things
that could be re-engineered

by a society that didn't
have to work for wealth.

Gender bias, right?

Racism. Nationalism.

A world where people

can actually live
the way they want to --

and not just a few people,

but anyone that has access
to these technologies --

will be a world
where people choose

not to be discriminated against
to a much greater degree,

because they don't have to play
those power games

that society forces upon them

because that's the way
it's been done

for generations and generations.

- Yeah. And I agree.

I think one of the other things
that is coming down the line

that is going to change that

is the ability to generate
power, electricity,

in a decentralized way.

If we really do get
a battery breakthrough

in the next 20 years

such that renewables --

I guess primarily solar,
but also wind --

can be generated
very far away from cities

and stored,
and can carry you through --

that's going to completely change
the way that cities operate,

and it's going to change

how much emphasis we put
on living close to one another

or in existing cities
and urban areas.

So there's lots of those
technological breakthroughs

which I think are cooking away
in the background

which will completely change
the social structure

that we are currently so used to.

- Mustafa Suleyman,
thanks so much for joining.

- [Narrator] And for our viewers,

if you're intrigued
by the possibilities of AI

and you want to dive deeper
into this fascinating world,

we invite you to check out
our other videos in this series.

And thank you for watching.

♪ [music] ♪