The incredible inventions of intuitive AI

How many of you are creatives?

Designers, engineers,
entrepreneurs, artists,

or maybe you just have
a really big imagination.

Show of hands? (Cheers)

That's most of you.

I have some news for us creatives.

Over the course of the next 20 years,

more will change around
the way we do our work

than has happened in the last 2,000.

In fact, I think we're at the dawn
of a new age in human history.

Now, there have been four major historical
eras defined by the way we work.

The Hunter-Gatherer Age
lasted several million years.

And then the Agricultural Age
lasted several thousand years.

The Industrial Age lasted
a couple of centuries.

And now the Information Age
has lasted just a few decades.

And now today, we're on the cusp
of our next great era as a species.

Welcome to the Augmented Age.

In this new era, your natural human
capabilities are going to be augmented

by computational systems
that help you think,

robotic systems that help you make,

and a digital nervous system

that connects you to the world
far beyond your natural senses.

Let's start with cognitive augmentation.

How many of you are augmented cyborgs?

(Laughter)

I would actually argue
that we're already augmented.

Imagine you're at a party,

and somebody asks you a question
that you don't know the answer to.

If you have one of these,
in a few seconds, you can know the answer.

But this is just a primitive beginning.

Even Siri is just a passive tool.

In fact, for the last
three-and-a-half million years,

the tools that we've had
have been completely passive.

They do exactly what we tell them
and nothing more.

Our very first tool only cut
where we struck it.

The chisel only carves
where the artist points it.

And even our most advanced tools
do nothing without our explicit direction.

In fact, to date -- and this
is something that frustrates me --

we've always been limited

by this need to manually
push our wills into our tools --

like, manual -- like,
literally using our hands,

even with computers.

But I'm more like Scotty in "Star Trek."

(Laughter)

I want to have a conversation
with a computer.

I want to say, "Computer,
let's design a car,"

and the computer shows me a car.

And I say, "No, more fast-looking,
and less German,"

and bang, the computer shows me an option.

(Laughter)

That conversation might be
a little ways off --

probably less than many of us think --

but right now,

we're working on it.

Tools are making this leap
from being passive to being generative.

Generative design tools
use a computer and algorithms

to synthesize geometry

to come up with new designs
all by themselves.

All it needs are your goals
and your constraints.

I'll give you an example.

In the case of this aerial drone chassis,

all you would need to do
is tell it something like,

it has four propellers,

you want it to be
as lightweight as possible,

and you need it to be
aerodynamically efficient.

Then what the computer does
is it explores the entire solution space:

every single possibility that solves
and meets your criteria --

millions of them.

It takes big computers to do this.

But it comes back to us with designs

that we, by ourselves,
never could've imagined.

And the computer's coming up
with this stuff all by itself --

no one ever drew anything,

and it started completely from scratch.

And by the way, it's no accident

that the drone body looks just like
the pelvis of a flying squirrel.

(Laughter)

It's because the algorithms
are designed to work

the same way evolution does.

What's exciting is we're starting
to see this technology

out in the real world.

We've been working with Airbus
for a couple of years

on this concept plane for the future.

It's a ways out still.

But just recently we used
a generative-design AI

to come up with this.

This is a 3D-printed cabin partition
that's been designed by a computer.

It's stronger than the original
yet half the weight,

and it will be flying
in the Airbus A320 later this year.

So computers can now generate;

they can come up with their own solutions
to our well-defined problems.

But they're not intuitive.

They still have to start from scratch
every single time,

and that's because they never learn.

Unlike Maggie.

(Laughter)

Maggie's actually smarter
than our most advanced design tools.

What do I mean by that?

If her owner picks up that leash,

Maggie knows with a fair
degree of certainty

it's time to go for a walk.

And how did she learn?

Well, every time the owner picked up
the leash, they went for a walk.

And Maggie did three things:

she had to pay attention,

she had to remember what happened

and she had to retain and create
a pattern in her mind.

Interestingly, that's exactly what

computer scientists
have been trying to get AIs to do

for the last 60 or so years.

Back in 1952,

they built this computer
that could play Tic-Tac-Toe.

Big deal.

Then 45 years later, in 1997,

Deep Blue beats Kasparov at chess.

2011, Watson beats these two
humans at Jeopardy,

which is much harder for a computer
to play than chess is.

In fact, rather than working
from predefined recipes,

Watson had to use reasoning
to overcome his human opponents.

And then a couple of weeks ago,

DeepMind's AlphaGo beats
the world's best human at Go,

which is the most difficult
game that we have.

In fact, in Go, there are more
possible moves

than there are atoms in the universe.

So in order to win,

what AlphaGo had to do
was develop intuition.

And in fact, at some points,
AlphaGo's programmers didn't understand

why AlphaGo was doing what it was doing.

And things are moving really fast.

I mean, consider --
in the space of a human lifetime,

computers have gone from a child's game

to what's recognized as the pinnacle
of strategic thought.

What's basically happening

is computers are going
from being like Spock

to being a lot more like Kirk.

(Laughter)

Right? From pure logic to intuition.

Would you cross this bridge?

Most of you are saying, "Oh, hell no!"

(Laughter)

And you arrived at that decision
in a split second.

You just sort of knew
that bridge was unsafe.

And that's exactly the kind of intuition

that our deep-learning systems
are starting to develop right now.

Very soon, you'll literally be able

to show something you've made,
you've designed,

to a computer,

and it will look at it and say,

"Sorry, homey, that'll never work.
You have to try again."

Or you could ask it if people
are going to like your next song,

or your next flavor of ice cream.

Or, much more importantly,

you could work with a computer
to solve a problem

that we've never faced before.

For instance, climate change.

We're not doing a very
good job on our own,

we could certainly use
all the help we can get.

That's what I'm talking about,

technology amplifying
our cognitive abilities

so we can imagine and design things
that were simply out of our reach

as plain old un-augmented humans.

So what about making
all of this crazy new stuff

that we're going to invent and design?

I think the era of human augmentation
is as much about the physical world

as it is about the virtual,
intellectual realm.

How will technology augment us?

In the physical world, robotic systems.

OK, there's certainly a fear

that robots are going to take
jobs away from humans,

and that is true in certain sectors.

But I'm much more interested in this idea

that humans and robots working together
are going to augment each other,

and start to inhabit a new space.

This is our applied research lab
in San Francisco,

where one of our areas of focus
is advanced robotics,

specifically, human-robot collaboration.

And this is Bishop, one of our robots.

As an experiment, we set it up

to help a person working in construction
doing repetitive tasks --

tasks like cutting out holes for outlets
or light switches in drywall.

(Laughter)

So, Bishop's human partner
can tell what to do in plain English

and with simple gestures,

kind of like talking to a dog,

and then Bishop executes
on those instructions

with perfect precision.

We're using the human
for what the human is good at:

awareness, perception and decision making.

And we're using the robot
for what it's good at:

precision and repetitiveness.

Here's another cool project
that Bishop worked on.

The goal of this project,
which we called the HIVE,

was to prototype the experience
of humans, computers and robots

all working together to solve
a highly complex design problem.

The humans acted as labor.

They cruised around the construction site,
they manipulated the bamboo --

which, by the way,
because it's a non-isomorphic material,

is super hard for robots to deal with.

But then the robots
did this fiber winding,

which was almost impossible
for a human to do.

And then we had an AI
that was controlling everything.

It was telling the humans what to do,
telling the robots what to do

and keeping track of thousands
of individual components.

What's interesting is,

building this pavilion
was simply not possible

without human, robot and AI
augmenting each other.

OK, I'll share one more project.
This one's a little bit crazy.

We're working with Amsterdam-based artist
Joris Laarman and his team at MX3D

to generatively design
and robotically print

the world's first autonomously
manufactured bridge.

So, Joris and an AI are designing
this thing right now, as we speak,

in Amsterdam.

And when they're done,
we're going to hit "Go,"

and robots will start 3D printing
in stainless steel,

and then they're going to keep printing,
without human intervention,

until the bridge is finished.

So, as computers are going
to augment our ability

to imagine and design new stuff,

robotic systems are going to help us
build and make things

that we've never been able to make before.

But what about our ability
to sense and control these things?

What about a nervous system
for the things that we make?

Our nervous system,
the human nervous system,

tells us everything
that's going on around us.

But the nervous system of the things
we make is rudimentary at best.

For instance, a car doesn't tell
the city's Public Works department

that it just hit a pothole at the corner
of Broadway and Morrison.

A building doesn't tell its designers

whether or not the people inside
like being there,

and the toy manufacturer doesn't know

if a toy is actually being played with --

how and where and whether
or not it's any fun.

Look, I'm sure that the designers
imagined this lifestyle for Barbie

when they designed her --

(Laughter)

But what if it turns out that Barbie's
actually really lonely?

(Laughter)

If the designers had known

what was really happening
in the real world

with their designs -- the road,
the building, Barbie --

they could've used that knowledge
to create an experience

that was better for the user.

What's missing is a nervous system

connecting us to all of the things
that we design, make and use.

What if all of you had that kind
of information flowing to you

from the things you create
in the real world?

With all of the stuff we make,

we spend a tremendous amount
of money and energy --

in fact, last year,
about two trillion dollars --

convincing people to buy
the things we've made.

But if you had this connection
to the things that you design and create

after they're out in the real world,

after they've been sold
or launched or whatever,

we could actually change that,

and go from making people want our stuff,

to just making stuff that people
want in the first place.

The good news is, we're working
on digital nervous systems

that connect us to the things we design.

We're working on one project

with a couple of guys down in Los Angeles
called the Bandito Brothers

and their team,

and one of the things these guys do
is build insane cars

that do absolutely insane things.

These guys are crazy --

(Laughter)

in the best way.

And what we're doing with them

is taking a traditional race-car chassis

and giving it a nervous system.

So, we instrumented it
with dozens of sensors,

put a world-class driver behind the wheel,

took it out to the desert
and drove the hell out of it for a week.

And the car's nervous system
captured everything

that was happening to the car.

We captured four billion data points;

all of the forces
that it was subjected to.

And then we did something crazy.

We took all of that data,

and plugged it into a generative-design AI
we call "Dreamcatcher."

So what do get when you give
a design tool a nervous system,

and you ask it to build you
the ultimate car chassis?

You get this.

This is something that a human
could never have designed.

Except a human did design this,

but it was a human that was augmented
by a generative-design AI,

a digital nervous system

and robots that can actually
fabricate something like this.

So if this is the future,
the Augmented Age,

and we're going to be augmented
cognitively, physically and perceptually,

what will that look like?

What is this wonderland going to be like?

I think we're going to see a world

where we're moving
from things that are fabricated

to things that are farmed.

Where we're moving from things
that are constructed

to that which is grown.

We're going to move from being isolated

to being connected.

And we'll move away from extraction

to embrace aggregation.

I also think we'll shift
from craving obedience from our things

to valuing autonomy.

Thanks to our augmented capabilities,

our world is going to change dramatically.

We're going to have a world
with more variety, more connectedness,

more dynamism, more complexity,

more adaptability and, of course,

more beauty.

The shape of things to come

will be unlike anything
we've ever seen before.

Why?

Because what will be shaping those things
is this new partnership

between technology, nature and humanity.

That, to me, is a future
well worth looking forward to.

Thank you all so much.

(Applause)