Original subtitles

(mysterious music)

As far as we know,

we're the only beings on the planet

conscious of our own existence.

Is homo sapiens also to be the final species of mankind?

There has been a number of human species on this planet.

All of these have lived and they have passed away.

Who says that homo sapiens are any different?

Of course, we are just also a passing stage

in the evolution of this planet.

But are we rendering ourselves obsolete?

It's not obvious to me that a replacement of our species

by our own technological creation

would necessarily be a bad thing.

I'm not very afraid of artificial intelligence.

I'm rather optimistic that this is going to be

one of the best things that have ever happened.

Other experts are skeptical.

Ask yourself, are you personally prepared

to become a member of the number two species?

Are you prepared to go to take the risk

that machines could become far more intelligent than we are?

This is not science fiction,

but the logical consequence of technological progress.

(tense music)

The digital age has delivered

boundless possibilities for humankind.

After all, technology has always been intended

to help optimize our natural abilities.

But now computers and robots are being designed

to become more human.

We're evolving together, merging into one.

The final consequence is a controversial vision,

super sapiens. (tense music)

Our planet originated 4.6 billion years ago.

Homo sapiens started to walk the earth

around the 200,000 year mark.

He developed culture, art, writing,

and he built monuments.

Humans also started to alter their environment.

The intelligent man begins to shape the world

according to his wishes.

With the onset of the industrial revolution,

the pace of human progress exploded.

New technologies helped us to explore our origins

and look into the future.

Today, the sum of our knowledge

is virtually doubling on a monthly basis,

multiplying digitally at such an extreme speed

that we can hardly keep pace.

But what does this mean for humanity?

In biological evolution, nothing ever plans ahead.

It's always the survival

of the immediate short term advantage.

Nowadays, we can look into the distant future

and plan for long-term advantage,

even at the expense of short-term advantage.

That's unique. That's new.

That's never existed before.

(people chattering)

Inside an unassuming home

just north of San Francisco,

the future of humanity is being imagined.

My name is Randall Koene,

and I guess my life's mission is to create

what I call substrate independent minds,

or as popularly known as mind uploading,

the ability to move a person's mind

into a digital substrate.

My background is in computational neuroscience

and electrical engineering and physics.

And I've been working on this for,

well, almost two decades now.

Randall knows that he can't

realize his vision alone.

The challenge is too great, the questions too complex.

(upbeat music)

But if Randall should succeed,

he would forever change human existence.

A mind that exists in the machine

lives differently than one in a mortal body.

Randall wants to bring the most capable scientists together

to create a kind of handbook for his controversial vision.

The big question is how do we do it?

How do you upload a mind into a machine?

Since after all, no one's done this before

and no one really understands

what the mind is or how it works.

Randall still has a long way to go,

but he's doing everything possible to ensure

that his ambitious project does not hit a dead end.

In the process, Randall relies on cutting edge technology

and the very latest research.

We do know that the human brain is made up of

around 100 billion neurons arranged along a pathway

that stretches nearly 6 million kilometers.

Should we mess with that complexity?

On the road to the human ghost in the machine,

fundamental questions arise.

Are you still yourself

if you have perfect memory?

Is that still human?

Is it still you if you have new senses

and you can see in the x-ray spectrum

or something like that?

You have to ask yourself, what does this all mean?

And to me, these are just,

many of these things are unanswerables at this moment.

It's an exploration rather than a certainty.

Machines are already becoming more human.

(upbeat rock music)

(cheering)

At Oregon State University,

a research team is working on Atrias,

a robot that can walk and run like a human being,

and much more.

This robot can also detect unexpected obstacles.

We humans don't have much of a problem

handling uneven terrain, but robots have a harder time.

Teaching machines to walk on two legs

has always been a challenge.

Ooh, nice.

(dramatic music)

Now for the ultimate test.

Atrias will compete with a human being.

The developers are confident their robot is up to the task.

I'm ready.

You need these. Oh no.

Its opponent gets a handicap

to enable a more balanced comparison.

Take your marks.

Get set. (gun fires)

(tense music)

(cheering)

Okay. Okay, well, congratulations.

Thank you.

I have to admit you won the race.

You did have a few advantages though.

I saw you using your arms,

and you were looking everywhere with your eyes,

but that's okay.

Oh man.

The researchers launch a second attempt,

this time, limiting the runner's freedom of movement.

Are you ready?

Yes, I'm ready.

And taking away her sight.

This is so hard. Oh my gosh.

You're almost to the end actually, if you just try.

Seriously? I've almost ade it to the end?

Oh, this is, it's such a workout, man.

You always have to keep moving and balancing.

No wonder Atrias's motors got so hot.

There you, there you are. Okay.

Okay, wait, let me bring you a chair.

Stay there.

Can I undo my arms?

Move back. There you go.

There you are.

Now you can see.

That was so much harder than I expected it would be.

Oh my gosh.

(upbeat music)

Humans are still one step ahead,

but Atrias is constantly learning.

The project is a small step towards the ambitious goal

of making machines more human.

Brain researcher Randall Koene spends a lot of time

on the road searching for answers.

His ultimate goal is to upload

the human brain into a machine.

In Silicon Valley, he's found like-minded people

and extraordinary inventions.

In a small town in northern California,

a developer is working on one of the food industry's dreams,

a machine that can automatically determine

whether a potato is good enough

to be accepted into the marketplace.

Randall is here to find out more

about this intelligent machine.

How is he potato sorting?

Good, got flagged.

Well, and we're using a actually camera temporarily,

but what we're going to use is this one,

which has a thousand neurons.

So the brain that's going to do the sorting

is going to be inside the camera itself.

Yeah, we'll ask to put three,

three of them in order to be able to inspect

the whole potatoes while it's flying.

You've got potatoes flying out of the end

and you're inspecting them as they're in the air.

Yeah, and basically what we're going to do is to inspect

the texture and the color,

and also the shape to make sure that the shape is regular.

And because sometime when they get the potatoes

out of the ground, it's pretty, pretty nasty, the shape,

but this one is actually quite nice.

(tense music)

One way to build intelligent machines

is to train them to observe, understand, remember,

and to learn from the process.

This little computer is learning to read.

It's shown letters and registers their shapes.

Once the system has learned to recognize the letters,

the ability to read accurately is the next logical step.

Initiating learning sequence.

M.

N.

A.

S.

B.

R.

This is my brain.

This is my brain.

Guy Paillet's invention

performs even more complex skills.

I just wanted to show you very quickly

the way we learn and we solve the confusion problem,

like learning the duck.

And it will over-generalize because if I go now to bird,

it will say that's a bird.

Oh I see, it thinks it's a duck too.

So I'm going to say, okay, this is a bird.

But do you know that overgeneralization

is actually exactly what makes human learning,

neural learning different from the machine learning.

Because if you've never seen either a duck

or a regular bird before, if you see one today

and you see a duck tomorrow,

you would think, oh, there's another bird.

Another one of those new things I just saw.

Yeah, it's a sub-categorization.

So here we've just showing with only two neurons,

and it can already quite make the distinction

between the two.

That's really quick,

it just learned that right there.

And that's something you wouldn't see

from traditional machine learning.

Guy Paillet's special little chip

can not only record, recognize, and differentiate things.

Now it's forgetting.

It can also forget its own knowledge.

And forgetting is a very human trait.

So this is basically is a forget.

And one way of actually making it more constant

is to force it to make a second neuron.

So the chip can recognize by itself

when it sees something new and needs a new neuron for that,

and it can do forgetting,

and it does rapid real-time learning

and real-time recognition,

which is the same sort of thing you see in humans.

This technology has more potential

to become instrumental for what Randall wants to do,

because it does not need to be programmed,

it just has to be trained.

We're at the very beginning right now,

and so this is one of the things that we are doing

with this tower that we call it the Neurostack,

in order to try to see if we can start a chain reaction

of like when we're thinking in our brain.

(tense music)

On the other side of the world in Hong Kong,

a pioneer of artificial intelligence is on his way

to see some of the latest developments in his field.

He has grave concerns about where the science is heading.

My name is Hugo de Garis.

I was a professor of artificial brains,

which is a field I've largely pioneered.

I first became aware of Hugo's work

in the late 1980s and early 1990s.

I started reading articles about some,

some mad scientist working in Japan,

using an artificial evolutionary process

to learn neural nets simulating small brain circuits.

And the aim was to use that to control robots.

And I thought, well, that's, that's pretty cool.

But I'm also rather political.

So I was thinking about the longer term consequences

and came to the very pessimistic, negative conclusion

that there may eventually be a species dominance war

over this issue of whether humanity

should build these godlike, massively intelligent machines.

(tense music)

Maybe they will look upon human beings as negligible pests.

They're like a cancer on the surface of the earth,

they're polluting the earth,

or maybe these machines will say,

"well, let's get rid of the oxygen

because it's rusting our circuitry."

(tense music)

There's no reason in principle,

that information processing has to take place in meat,

you know, in the wetware of our biological brains.

That gets you halfway to expecting that at some point,

we will build machines

that are more intelligent than ourselves.

It's a fundamental game changer.

Super-intelligence will be the last invention

that humans ever need to make.

After you have super intelligent scientists and inventors

and programmers, then further advances will be made

by this machine superintelligence,

and at digital timescales rather than biological timescales.

Hi there, what's happening?

What are you doing?

Why Mandeep, I'm just looking around.

I'm trying to talk to you.

Where are you now?

I think I am in Hong Kong,

but sometimes I want to walk.

They're usually hard to interrupt.

I don't understand you.

I am listening to you,

but I can't understand everything a human can yet.

For me, the question isn't do we create AIs or not?

They're coming, the question is, you know,

what can I do to affect how it happens?

Track.

Tracking friend.

Identify.

I think that's a book.

I would rather see AI developed sort of by the people

for the people than developed within a large corporation

or within a government's military organization.

AI is under rapid fire development.

I am a robot

and I am only beginning to understand.

This is the beginning of a new era

in which we train machines in ways that are roughly similar

to the way that perhaps children learn,

through brain-like techniques,

multilayered neural networks that are capable of intuition.

Molecules is a chemical element.

So it is going to be difficult very soon

to even know what the machine is doing to make sense.

AI is going to be very sophisticated

far sooner than we think.

I mean, we have neurons that can fire

maybe a hundred times per second or so,

whereas even current-day transistors

operate at gigahertz levels, billions of times per second.

If you look at how fast signals can travel

in biological neurons, a hundred meters per second

really is the maximum speed.

Whereas in computers,

signals can travel at the speed of light.

Ultimately the limits of competition in machine substrates

are just far outside what's possible

with any possible biological construction.

(dramatic music)

A week's worth of progress in a machine

running a million times faster than a human brain,

that's 20,000 years of of human progress.

So it's a, the idea that the minds involved

will happily take direction from us at that point,

that that may be a little far-fetched.

Skeptics warn of the consequences

of continued unchecked technological progress.

Do we build gods,

or do we build our potential exterminators?

That I see as the big question of the 21st century.

But development is not a one way street.

Randall Koene wants to upload

human consciousness into a computer.

This would likely endow the ghost in the machine

with human qualities.

People often ask me what it would be like

if I uploaded my mind, or if they uploaded their mind.

What are those going to be like for me to be living

in a different kind of body, having an emulated brain?

To be honest, we can't really know,

because there are so many things involved there.

You can't predict it all in advance.

All we can do is make small iterative steps,

and at every iteration, try to learn more about it.

(tense music)

I think it's really important

for the long-term future of humanity.

It's important for not just our survival,

our survival in the sense that we survive

the next big meteorite impact,

but it's also important for our meaningful survival.

So the ability to continue to develop ourselves

and to be able to reach beyond what we can even understand

and grasp within this little niche that we evolved into.

Randall is on his way

to a meeting in San Francisco.

He has an appointment with a researcher

he has wanted to meet for a long time.

On a small scale, Stephen Larson has managed to achieve

what ramble might be working on for the rest of his life.

In the midst of day-to-day life,

a new life form will soon be revealed.

Thread worms are nasty parasites,

but a digital one is something to celebrate.

Stephen?

How are you doing? Can I get you something?

Good to meet you.

No, I'm good. Yeah.

So I can't believe that we haven't met before.

I mean, given OpenWorm and emulating the worm, basically.

Yes. OpenWorm is a project where we're building

the first digital organism in a computer.

It's an open science project.

So that means that we collaborate

with people on the internet.

Okay, so let me show you actually the simulated part,

since we were talking about that.

So this is how it looks when it's not

the beautiful rendering.

So this is actually the output of--

This is beautiful actually.

The artificial worm

seems to move in a completely natural way,

just like a real thread worm.

We tried to make it as realistic as possible,

and it responds, right, it responds to things.

If somebody looked at that,

they would see something that looks a bit

like a real creature, or at least it seems

like a real creature in some ways, right?

So if they asked you a question like, oh, is this,

is this thing in your computer, is this alive?

Or can it feel pain? What would you say to that?

Yeah, I think right now,

obviously our definition of life

is purely in physical matter.

So we would have a hard time defining anything

I think that's in a computer

as anything more than sort of a shadow of life, right?

What happens if you can do this

for a lot of different parts and you end up with

basically an entire organism inside of a machine,

life in a machine, what do you think?

Yeah, okay, imagine we open up your skull

and we go and we find one neuron in your brain,

and then we replace it, we plug in the place

where that neuron is connected,

we plug a computer and now that computer is doing

what that one neuron can do, and you know,

are you a robot at that point?

You'd say, no, it's just one neuron, right?

But now let's say I go and I take a second neuron, right.

And I replace it, and now am I a machine yet? No.

Okay, third, fourth, fifth.

So the whole idea of this thought experiment is

that now when you go to like an iteration

and you say, well, so how many neurons do I need to replace

before you're no longer you?

(tense music)

Stephen's project shakes the very foundations

of our self-conception, who we are and what makes us human.

Now you've got this entire human brain in your computer.

So you have to wonder, is that a person?

Right.

And how do you treat this brain in your computer?

That leads you then to the whole question of mind uploading.

Yeah. You know?

This is very close. This is basically it.

And, and would someone volunteer for that?

Would you volunteer for it?

Would you want to do it yourself?

So I guess, I mean, there's lots of specific questions.

Like, do you have to chop up my brain in order to upload me?

'Cause then I don't have it anymore

and I can't really go back, so I'm not so eager.

There's no undo button then, that's the thing.

Right, so I'm not yet so eager to jump in and sign up,

but there's all sorts of other, you know,

interesting questions that I'm sure you're exploring.

Like, okay, so if you're uploaded into a computer,

like how secure is that computer?

An important problem to address

as we're confronted by an alarming increase

in the threat of computer hacking and cyber crime.

Randall, however, is not deterred.

If I want to be able to upload my mind

or upload your mind, if we want to be able to do

whole brain emulation as such,

it means that we need to be able to get really close

to the activity that's going on inside the brain.

We need to be able to register it,

characterize what's going on in a circuit.

That's the way that people make neural prostheses.

That's the way that they can create a replica

of what's going on, but right now,

we still have a long way to go,

because most of the interface technology that's available

is very primitive and can only barely record

some information or send some signals to the brain.

The Blue Brain Project in Geneva

is decoding the complexities of the brain.

We understand a lot about the cosmos.

We understand huge amounts of physics.

We understand chemistry,

but why I am here capable of talking to you?

Why I can understand your questions, why I can answer them,

why I may feel an emotion while I'm doing so,

this today we seriously do not understand.

And I think this is a great human endeavor.

And it's what motivates most of us.

Richard Walker heads a team of researchers

who are taking the first serious steps

required to answer those questions.

They're using the brains of rats,

segmenting them into ultra thin slices

and recording their measurements.

From these, a complex model of the rat brain

is produced in the computer,

a kind of three dimensional map

with all the natural biological links and structures.

The important point is it doesn't really matter

what brain you're talking about,

as long as it's a mammal brain.

So you can do it for rats.

You can do it for different bits of the brain.

You can do it for rats of different ages.

We can do it for other species of animal too,

like we want to end up with humans.

Richard Walker and the researchers

have already moved forward,

not least thanks to their highly sophisticated laboratory.

They analyze scans of a mouse's brain on a gigantic screen.

Each line represents an individual neuron,

a brain cell connected to hundreds of other cells.

Countless signals are sent ceaselessly

back and forth across this organic network,

each of them stimulating a specific reaction

at their intended receiver.

Our thoughts are a result of the complex interplay

between these cells, as is therefore our consciousness.

The goal the scientists are striving for

is to crack the code behind this evolutionary mechanism.

When this is turned into technology,

I think it can become a very powerful tool,

and tools are not all used for good.

When we have our machine tools,

we can use them to make a wonderful surgical instruments.

We can use them to make vehicles.

We can use them to make weapons.

Some of those applications are good and some are bad.

I think the ethics is not the ethics of the basic work.

I think we should carry on doing that.

Humans have always built tools.

Humans have always searched for basic knowledge,

and I think it would be catastrophic

if we stopped doing that.

Our most advanced technologies

are being used to solve the mysteries of the human brain.

(tense music)

On the other hand, artificial intelligence is helping

to make machines more human.

Can I ask you a question?

Why is the sky blue?

The sky's blue color is a result

of the Earth's atmosphere scattering

all the short wavelength blue light shined on us by the sun.

The human eye perceives the color blue

when looking at the sky as a result.

Meet IBM's state of the art AI system, Watson.

Watson is a self-learning computer

that can understand human language.

In 2011, Watson became famous when it defeated

two human opponents in a quiz show.

It can even diagnose medical problems.

Doctors were able to use Watson's incredible computing power

to help identify the genetic cause of disease

in a cancer patient and suggest a treatment.

To do this, Watson scanned 3,500 medical studies

and 400,000 additional texts, all in a mere 17 seconds.

Watson is also used inside children's toys.

Dolls and dinosaurs are no longer limited in their speech.

They're now capable of simple yet complete conversations.

Would you like to eat the centerpiece?

Well, it does look tasty,

but I'm not really that hungry.

Oh, good lord, a picky dinosaur.

It's actually not a toy.

It's just disguised as a toy.

Very cleverly wrapped in green plastic

and disguised as an educational device.

But really it's a personality.

It's a companion, it's friends and intellect

all wrapped up into one.

The fact of the matter is if you think about what Watson

has accomplished, people talk about it sometimes as AI,

you know, artificial intelligence.

I actually prefer to think of it as accessible intelligence.

What I mean by that is really a revolutionary way

of interacting with these tremendous capabilities.

You literally have access by simple,

direct, human-like interaction.

You can say "I'd like to make something for dinner

that's based on, I've got a particular fish in mind, Masa.

What can I make?"

And chef Watson will simply come back with an answer.

That is at the human level.

You're not a chief information officer.

You are somebody standing in the house,

sitting there and getting really tired of bringing in pizza.

You want to access it? It's like accessing this.

Ask a question, it's not that hard.

Go for it. Not that hard.

The smart toy is really something.

It's a very fun game, I guess, for the whole family,

except that it also shows us we're inviting

artificial intelligence into our living room.

And we're probably not thinking very much about it

because it's just a toy.

When something is a toy,

we attempt to find that simple and benign,

unless it's too expensive or too violent.

But what is happening is of course

with artificial intelligence is rapidly becoming

a part of our everyday life.

(tense music)

Ask yourself how much money would you be prepared to spend

to buy a home robot that was genuinely useful.

Imagine it could babysit the kids and wash the dishes

and wash the clothes and clean the house

and entertain you and sex you and educate you,

and all those things.

Obviously a major industry

worth trillions of dollars a year worldwide in the 2020s.

So you see that coming and then, well,

you can see the writing on the wall pretty well.

So like if my left hand here represents

human intelligence level and my right hand here,

that represents current more or less

machine intelligence level,

you can imagine what's going to happen, right?

Millions, if not billions of people will see

with their own eyes, they'll see that IQ gap close, right?

And then lots of people start asking the obvious questions.

Well, is humanity going to allow these machines

to become as intelligent as human beings?

At the Hong Kong Polytechnic University,

a team of researchers is working full steam

to develop precisely the machines

that Hugo de Garis warns against so emphatically.

The head of the team here

is Hugo's former colleague Ben Goertzel.

And Hugo and I ended up collaborating

on his China brain project.

He and his students were building a sort of neural net,

AI supercomputer on a bunch of networked together GPUs.

And I went there and taught some of his students

how to use the early versions of my OpenCog system.

I ended up hiring a few people out of Hugo's lab

to come to Hong Kong and work with me here

on OpenCog related things.

How can I put this? He's my best friend.

Don't hurt his feelings, but.

Hey Hugo, how's it going, man?

Ooh, it's hot.

It is hot. It's been a while, huh?

He's not as political, put it that way, as I am.

He's much more optimistic. I'm more pessimistic.

I would claim more realistic.

What's up with you?

You're the one with the news.

Yeah, we got a lot, a lot, a lot of stuff going on here.

So can your robot talk to me yet?

We've been doing some stuff for the robot.

When I was 18. I got conscripted to fight in Vietnam.

And believe me, that experience radicalizes you.

Let's see a bit of the fun little robots.

It forces you to look on the dark side.

Grab a chair, I guess.

These machines may become so advanced,

they may look on us the way we look on bacteria.

Like every time I do this, stamping my feet on the floor,

probably I'm killing bacteria, right?

It's a hard floor, but I'm human.

I don't give a damn about these bacteria that I'm killing

because I feel I'm so superior to them.

They're one cell creatures.

I'm a hundred trillion cell creature.

Their death means nothing to me.

So hi there.

Do you remember Hugo?

Hugo de Garis.

(giggles nervously)

Hugo de Garis is a brain builder

and he believes that topological quantum computing

can help revolutionize computer science.

We both see very clearly, you know,

the potentials inherent in physics

to just process information and manifest intelligence

many, many, many, many orders of magnitude

beyond what the human brain can do.

So how do you represent a link? Again, lines.

Probably the one area where we have had

the most arguments, which have been, you know, friendly

and by and large simulating and entertaining,

just in terms of the human consequences

of the advancement of AI over the next decades.

Hugo has seemed to consider it more likely

that super advanced AIs will just squash all humanity.

Whereas my best guess for what it's worth

is that enhanced humans and improved human life

will coexist with dramatic,

dramatically superior superintelligences.

We have one advantage, we get to make the first move.

We will create, design the first superintelligence.

And so if we do our job right,

then this superintelligence should be an extension of us.

It should be on our side.

I'm not so sure that will work.

We can't even control our own teenagers,

so how are we expected to create a new kind of intelligence

that we barely will understand?

Because it can rewrite its own code

and will think very differently from us,

and still maintain control of it.

I'm rather optimistic that this is going to be

one of the best things that have ever happened.

However, there is a small probability

that it turns out to be the last thing humanity ever does.

And it's the end of our species

and all the value we can ever have achieved.

Inside the control module then.

I think Ben, by nature, he's more of a techie than I am.

I mean, I was, I've gone back to my first

and original love of math, physics.

All the positive things that could come out

of artificial intelligence, you know, I agree with Ben.

I just don't think he goes far enough.

Sort of a Frankenstein monster from various pieces.

If we built machines more intelligent than ourselves,

that were capable of making further changes

to their own source code so that they became the drivers

of progress in building future AI,

then we would have built something

that in very short order would become functionally God-like.

We essentially would have built a God in a box,

and we will want to be very careful

not to build an angry god.

(tense music)

The prototypes in Ben Goertzel's lab

seem neither angry nor particularly highly developed,

but looks are deceiving.

The team is working on a way to network intelligent machines

by uploading that artificial minds into cyberspace.

The mind will be in the cloud.

The sensors are on the robot's body.

What we learn here from playing with these prototypes

can then be commercialized.

If you can get intelligent robots out there

with millions of users, I mean,

imagine the amount of learning that the AI can do

from all these users, right?

Yeah, that's been your vision.

I mean, over at Henson's office,

we've got robots, his latest robot heads

really look totally realistic, look like human beings.

Can I see reality the way you do?

Reality cannot be detected.

That's an interesting metaphysics.

Makes sense to me.

Makes sense to me too.

Will you still got to Beijing now as well?

Yup.

So what's in the suitcase?

Oh, this is Sophia's body.

Here we go.

There's a lot of requests for companionship robots

of various kinds, elder care companionship, you know.

This is so realistic.

I mean, just look like a mold.

So I like to sculpt, have that animation background.

Just moves with the machine.

Yeah, so what you might notice

is her lips can actually purse.

Yeah, which is kinda freaky.

Her mouth can, you know,

make the full range of speaking motions.

So she's got cameras in her eyes and a camera in her chest.

She tracks faces, she'll actually see your face

and make eye contact with you.

Hugo seems fascinated yet skeptical.

So in essence, the collaboration between you two.

And a bunch of other people.

Keeping it simple, you're the body and you're the mind.

Yeah, so Hugo, can you imagine yourself uploaded

into one of these robots?

I might be the millionth.

Millionth?

I wouldn't be the first.

How about the second?

(dramatic music)

Like all technological breakthroughs,

the development of human-like machines may come at a cost,

but that will not stop the human spirit of exploration

and thirst for knowledge.

Well, once I'm in there and I tell you how great it is,

then you may be convinced.

You know, you know, Randall, right, from minduploading.org?

The neurophysiologist?

Yeah, he's working on trying to figure out

how to get all the information--

From the brain.

Out of a person.

How to read a brain.

Yeah, yeah, how to read the brain,

So then you could scan all the information

out of a human brain, put it into a computer,

which you could use to drive a body like this.

That is a huge task.

They can't do it now.

A quadrillion connections.

Yeah, Randall's work is really neat.

(tense music)

Randall Koene's project

entails enormous challenges.

I would like to transcend the limits of biology,

mostly because those limits are limits

that are imposed by natural selection.

So they're imposed by the needs of the environment,

not necessarily the desires of what we want to accomplish

or needs that we haven't encountered yet.

We've been messing around with biological evolution

for centuries, even millennia.

Now we're coming to the point where we can mess around

with genetics directly and we can mess around

with computers, with robotic creations.

We are acting as a kind of catalyst,

perhaps even more than a catalyst,

for dramatic changes in the course of evolution,

broadly understood. (dramatic music)

But the nature of the consequences

is by no means solely technical.

When you talk about copying information from your brain

into another system, that poses a variety

of philosophical problems.

You seem to have been duplicated at best,

but the fact that you or your identical twin

is now able to live out its adventures and is immortal,

presumably, that will be a cold comfort to you

having been copied.

So there are questions of identity here

that would arguably confound any notion

of uploading a human mind into another condition.

Some people think that the self is a program

and you can download and upload it at will.

I think that's a category mistake.

I don't think a person is a program,

because the program is like a mathematical equation,

where individuals who function causally in the world

occupy space, equations aren't located anywhere.

People often ask me, what is it that you have

to take along and what can you leave behind

if you're going to emulate a brain

so that you have a mind upload,

which parts are essential, what makes up me?

The best way they have to scan your brain now

is to remove your head, freeze it,

slice it like in the deli counter,

and then scan each slice.

Then you try to reconstruct the connectome

inside the computer and they're working on

the three-dimensional image processing.

There must be a much better way.

Well, I think if we just like drill a hole here

and put some carbon nanotubes in your head,

and then maybe those nanotubes can snake around

and read out the map.

They could have tiny little radio transmitters.

It would also serve as a brain computer interface

while you're still alive.

I think with my own personality,

if I upload myself anywhere, I'm not that conservative,

so I would just, I would want to enhance my mind

however I could, if I could see through a billion sensors

at once and not just one body, then why not, right?

And if I could solve a differential equation in my mind

in a fraction of a second, sure.

So once we've gotten the mind out of this substrate

that it evolved in but is now trapped in,

then that's just the beginning

of an endless variety of options.

There's still that deep philosophical problem,

the sense of self.

So would, would this have its own sense of self?

I think that that would,

that would be resolved as we start getting

maybe more of a scientific understanding of what self is.

There's a horrible possibility here,

which is that we could build machines

that are far more intelligent than us

in terms of their competence,

in terms of the kinds of goals they can achieve

in the world and they may, in fact, not be conscious,

the lights may not be on,

and there will be nothing that's like to be those machines.

And that is quite an alarming prospect from my view,

ethically speaking, because then you could just have a,

you know, the galaxy being populated by, you know,

blind robots and you know, who would want that?

What if we wire me into her?

Now my hypothesis is wiring my brain

into the robot's cloud mind,

I think I would feel something.

Yeah, including consciousness.

It's uncanny, isn't it?

Now she looks like your daughter.

What looks like a harmless game

is fundamentally one of the greatest challenges

mankind has ever faced.

(electronic stuttering)

It deals with the question of which role technology

should play for us in the future.

Randall Koene has already answered

this question for himself.

So to be honest, of course I don't really know

how long it's going to take

to create a whole brain emulation

because it's a very complicated task,

and it's very hard to tell with tasks like that.

Just how long they take depends on so many things

and we need a lot of different tools,

many of which are just basically in development.

So it's entirely possible that I may not live to see

the outcome or may not live to help get us there.

But what I'm trying to do is to at least pave a path,

to create a roadmap to help move it along

and to make sure that there are projects in existence

and people who understand the meaning and the importance

of these ideas so that it can carry forward anyhow.

And I think that's why I do this.

And perhaps this is how I make meaning in my life.

(tense music)

It has been a long journey from the Big Bang

to the emergence of an intelligent, self-aware being,

one that never ceases to question.

And this will be a watershed moment,

more important than anything that has happened

in human history, the creation of superintelligence

would be the last invention that we will ever need to make.

And if one is looking for some kind of comparable event,

I don't know, maybe the rise of life in the first place.

(tense music)

Human nature is all about self transformation, I think.

We all try to become better people.

And as a society, of course,

we will not really stop learning things.

Innovation has always been with humans.

So it's not really possible to stop technology.

However, you might aim at get the beneficial technologies

before dangerous technologies.

(tense music)

Maybe these machines will be friendly towards us,

but on the other hand, maybe not.

So we're taking a risk in building them,

and at some stage, humanity is going to have to choose,

your children or your grandchildren

will have to make a decision.

It'll be very real.

It won't be just a piece of science fiction

that it seems like today.

(tense music)