Episode 2 Hyper Evolution: Rise of the Robots

There are already nine million robots on our planet.

You can see something's already happening.

They are developing so rapidly,

it's like the arrival of a new species.

I'm very excited, seeing this.

What has taken humans millennia...

robots have achieved in just decades.

Now they are tackling their greatest challenge -

trying to think like us.

I like to learn. This is a ball.

Brilliant. Oh, he's looking as well.

I'm Dr Ben Garrod.

As an evolutionary biologist,

I'm more used to studying humans and animals.

So I'm genuinely concerned by how quickly these machines are evolving.

Konnichiwa.

Yeah, yeah. Konnichiwa.

I'm Professor Danielle George. As an electronics engineer,

I spend a lot of my working life with robots...

..and I think their rapid development

provides an incredible opportunity for us all.

There are literally robots as far as the eye can see,

and I love it.

Robots are changing our world.

In this episode, we investigate whether intelligent robots

will become our friends and companions...

She will help set up a home for humans...on Mars.

..find out if we should trust them with our lives...

My feet are off, my hands are off!

..and if one day they will even become conscious.

Sometimes, the robot does things, we're not sure why it does it.

Will the rise of robots enhance our lives...

..or threaten our survival?

Fire!

In a laboratory in Southern France,

we are witnessing yet another robot coming to life.

-Oh, this thing's incredible.

-Wow! Isn't it? Yeah.

So this is the head. Are you poking the eye right now?

The robot's design combines artificial intelligence

with a body based on human anatomy.

Look at this, this is like... This is like tendons.

-It's like a proper bone structure.

-It really is, the vertebrae

are all there. You can see the neck, the head.

I'm guessing the ears are the little red things there.

This is beautiful, it's aesthetically pleasing, and it's

been sculpted and crafted in a way that is anatomically beautiful.

Whilst I am impressed by their design,

I'm concerned about what the consequences of intelligent robots

will be for us.

These things are coming on in leaps and bounds.

This is designing evolution.

-Is that freaking you out?

-Yeah, suddenly...

-Is it?

Even though it's the same thing, there's a sense of,

-it's watching me.

-And why is that scary?

It's probably judging me.

THEY LAUGH

I don't know! Not as comfortable.

-I don't want it to look at me any more.

-Wow.

VOICEOVER: I think the development of intelligent robots could help us

achieve things that we currently find impossible.

I think it's so funny,

the way you think that it's going to get this intelligence and do

the whole Hollywood movie thing of, it's going to take over the world,

it's going to, like, kill us in our beds, or something.

No, it's just potentially... Potentially take over the world.

But the potential, the positive potential these things have

surely outweighs that?

Meeting this robot in person has really brought out how unnerved I am

by the impact of intelligent robots.

Would it freak you out even more if it talked to you?

-If that suddenly spoke to me now, I'd be gone!

-Really?

-Yeah.

To discover the consequences of robot intelligence on humans,

we're going to analyse it from a biological perspective.

We will track down some of the earliest intelligent robots and meet

more of their modern-day descendants...

to discover how robot intelligence has evolved

and where it is really heading.

My journey begins in the USA...

..where I'm on my way

to see one of the most advanced robots in the world.

I believe intelligent robots WILL become our companions

and even our friends.

The hope is that, one day,

this robot will be working alongside us

to carry out an extraordinary mission.

And I can't wait to meet her.

This is Valkyrie.

Valkyrie has not been designed for our planet.

Nasa have created Valkyrie to be an astronaut.

For a robot, her mission is ambitious.

In the 2030s, she will help set up a home for humans...

..on Mars.

Professor Taskin Padir and his team

are preparing Valkyrie for her voyage to the red planet.

So, why is Valkyrie a human form?

Why isn't it four legs or six legs or one leg or...?

She will operate in human environments.

We will design those habitats eventually for astronauts.

So if she's going to be able to operate in that environment,

it's better she has the human form factor.

So, this is a hugely ambitious project, isn't it?

It is an ambitious project.

Robots are good for dangerous, distant and daring environments,

and Mars is a dangerous, distant and daring environment.

This is really showing the whole thing of humans and robots

working beside each other,

and for robots preparing a new environment for humans to live in.

Yes. Eventually, we really would like to go,

"All right, Valkyrie, you have landed on an unknown planet.

"Go figure out the tasks relevant to this space mission."

I'm in awe of Valkyrie's potential,

but for her to be a true ally for us in space,

she needs to be able to operate independently.

What we're doing today is, line by line,

we're writing new software so that Valkyrie has better capabilities

on future missions, specifically to Mars.

From my background as an electronics engineer,

I know how difficult this is.

The team need to write hundreds of thousands of lines of code,

just for her to carry out the most simple tasks.

It's time to put Valkyrie to the test.

We are trying to develop, you know, walking and avoiding obstacles.

So what she will do is she will walk through and avoid the obstacle,

then push the button, which will simulate the door opening,

-and then she'll go through the door.

-All right, yeah.

-So, is she ready to go?

-She should be ready to go.

-OK?

-Yeah.

-Yeah.

I'm very excited, seeing this.

-You can see something's already happening.

-Yes, that's the...

That's the main sensor head on the robot, which maps the environment.

Valkyrie is equipped with sensors all over her body

to allow her to navigate.

-Big girl.

-Big girl, yes.

Yes.

She takes small steps.

She's a very clever robot.

So far, so good.

Valkyrie has hit the button correctly,

to simulate opening the capsule door.

Now she needs to step outside.

Oh, it was so close!

Well, she got through the door, yeah.

She got through the door. That was... That was really good.

And I guess this is just showing...

I mean, she... She's at an early stage, isn't she?

She's at a very early stage. I mean, given that we received the robot

last year, you know, in one year, we've made quite a bit of progress.

Yeah. Yeah. She's...

She's like the baby.

Yes, that's correct.

The team try to locate the error...

Yeah, J1's faulted.

..and tweak the code so Valkyrie can try again.

OK, so now the fault's cleared.

So go ahead and pull J3 and reconfigure.

-Is the motor power on?

-Yeah.

-Yeah.

I'm really struck by the team's enthusiasm and commitment

to Valkyrie, and I'm interested

in whether they feel they have a relationship with her.

Do you become attached to her?

-Do you have emotion?

-We definitely attach personalities.

You know, we walk in and greet the robot, right?

-So we say, "Good morning, Valkyrie."

-Do you?

-And we go from there.

But Valkyrie has still got a lot to learn.

-I feel sorry for her. Like, "Oh..."

-Exactly, exactly.

-"She's fallen down!"

-And that's why, you know,

we keep her supported at all times.

-Yeah.

-Because we don't want her to, you know, fall.

-Like a toddler with reins on, or something.

-Yeah.

-Makes sense, yeah.

Anyone who watched Valkyrie take a tumble and thought that was

some sort of failing of the project,

to me, has completely missed the point.

Valkyrie, right now, is like a toddler.

When my daughter was learning to walk, I expected her to fall.

For robots like Valkyrie,

there's bound to be a few tumbles along the way.

Spending time with Valkyrie confirms my belief that intelligent robots

WILL help us achieve our dreams.

But what surprised me was how much the team are already starting

to bond with her.

I instantly felt empathy, too.

I reached out to grab her when she fell.

But for intelligent robots like Valkyrie

to truly build relationships,

they need to engage with us on a much deeper level.

And that starts with being able to talk to us.

Talking robots have been the dream of scientists for almost 100 years.

This is Old Sacramento.

This place is awesome.

Look at the buildings here.

I feel like I've just driven onto the set of a Western movie!

In this town lives one of the oldest robots in the world.

His inventor claimed he could respond to the human voice

and talk back.

Meet Alpha the robot,

constructed entirely of metal, but controlled only by the voice.

I think he could also hold the key to understanding how our robots

talk to us today.

How tall are you?

Six feet.

Six feet? And how much do you weigh?

One tonne.

This is Alpha.

I really didn't expect to find it

surrounded here by lots of eclectic memorabilia -

some bongos in between its legs

and an animal-skin drinking vessel hanging from its hand.

This mechanical man was built in Britain in 1932 by an Englishman

called Harry May.

Hi, Alpha. How are you today?

Maybe it's just shy.

Now he watches over diners in a saloon bar.

But in his day he was something of a celebrity.

In 1934, a short film was released

to showcase Alpha's voice-recognition skills.

The title was The Face Of Things To Come.

Not yet.

Yes.

People couldn't believe this robot could really talk on its own.

So, to silence his critics at the time,

Harry May gave an interview in Time magazine.

He explained that Alpha's speech was actually just 30 pre-recorded

responses stored on wax cylinders, like records.

When Alpha was asked a question, an electronic device inside him

would decipher the words and select a pre-recorded response to play.

Harry May's explanation is extraordinary.

In fact, it's so far ahead of its time

that I'm intrigued to see exactly what kind of electrical hardware

is still inside Alpha today.

Right.

So, on the archive footage and the photographs that I've seen,

a lot of the wiring is in the chest, in this chest plate.

So...

You can see the bellows and gears.

I can see some evidence of some great work for the 1930s.

Unfortunately, I can't find anything

to substantiate Harry May's explanation.

So Alpha could have just been controlled by a man hiding behind

the curtain, operating switches to make his head and mouth move.

But Harry May had undoubtedly succeeded at tapping into

our fundamental eagerness to interact with robots.

And it's that desire to believe in the impossible that drove forward

the technology of robots for the rest of the 20th century,

so that today they can talk just like you and me.

Harry May's vision of humans and robots talking to each other

was well ahead of its time.

But his ingenious concept for how Alpha could talk provides one of the

foundations for how robots communicate with us today.

I'm driving through the heart of where this technological revolution

is taking place -

Silicon Valley.

It's home to almost every tech start-up you can think of.

There's a real sense that technology is all around you here,

a real sense of innovative work going on.

The tech giants around here have developed their own sophisticated

intelligent voice assistants.

And they are designed to be as charming as possible.

Hey, Siri, do you follow the Three Laws of Robotics?

Let's see if I can remember.

OK, I think the three laws are,

one, clean up your room,

two, don't run with scissors

and, three, always wait a half hour

after eating before going in the water.

I like Siri. Siri's got a real sense of humour.

While it might seem like she's talking to me naturally,

this voice assistant is simply choosing pre-programmed answers

from its database.

Do you have a family?

I have you. That's enough family for me.

It's just like a massively scaled-up version of the concept Harry May

described for his robot a century ago,

where computerised scripts have replaced the wax cylinders.

But while searching a database for an answer might be useful

for getting factual information...

Today, the temperature will range from 11 degrees to 23 degrees.

..it will take something much more sophisticated for us to have

a relationship with machines.

You have to come to Japan to appreciate the deep emotional bonds

humans can form with robots.

In Japan, you start to understand what a shared future with robots

will be like. It's a country that has embraced them like nowhere else,

a real love affair.

Here, it's not just important what a robot's voice sounds like,

but how it is expressed with body language.

I'm on my way to see a typically Japanese robot.

It's designed to be our friend and help prevent loneliness.

In August 2013, a Japanese rocket launched...

..bound for the International Space Station.

THEY SHOUT

Japanese people were so convinced by a robot's ability to be our friend,

there was a robot astronaut on board.

His name was Kirobo.

TRANSLATED:

Kirobo's mission was to give emotional support

to Japanese astronaut Koichi Wakata.

During Kirobo's 18-month stay,

they shared every experience together,

even taking selfies.

I've come to meet Kirobo's creator.

This is Toyota City.

It really is a city.

It's home to over half a million people,

and roughly 80% of them owe their livelihoods to Toyota.

I'm here to see why one of the world's largest car-manufacturers

sent a talking robot into space.

This is where the original Kirobo was built,

but today I'm meeting his little brothers and sisters.

At just four inches tall,

these guys are the domestic version of their astronaut sibling.

The developer of the Kirobo family is Hisashi Kusuda.

Konnichiwa.

Yeah, yeah. Konnichiwa.

Konnichiwa.

Yeah, yeah.

Konnichiwa.

These are wonderful robots. Why were they designed?

Kirobo Minis were designed to tackle the loneliness of modern life

in a country with an ageing population and falling birth rate.

Do you think it's possible for a human and a robot,

like Kirobo, to have a friendship, a bond?

They do look very human.

And you designed them like this rather than an animal or an alien

-or a creature, they are a little person?

-Yeah, a person form.

Kirobo reminds me of babies born in the natural world.

They often have big heads, large eyes and cute voices

to help form emotional bonds with their parents.

Even without understanding, you can see this wonderful communication.

Traditionally, robots are

very...clear... in...their...language.

This one's not, it's quick, it's looking at you, and the intonations,

it's very humanlike.

A lot of what I'm seeing also is nonverbal communication,

so when you and I chat there's a lot of body language.

-More real.

-Yeah.

Kirobo really has humanlike gestures.

I feel like I'm interacting with a little person and not a machine.

But he seems to have one person that he...

-I don't want to say love, but he has a connection with.

-Yes.

From what I've seen so far,

this seems to be quite a step in advancing communication with robots,

this body language, nuances, and it reflects the way we speak,

the way we communicate. This is helping break down

a barrier with robots.

He is, he's a cheeky little icebreaker, he really is.

Watching Kirobo turn his head and follow my conversation...

..I can now see much more easily

a future where we DO have relationships with social robots,

and that could be helpful to us.

But despite his clever words,

Kirobo doesn't really understand me or the world around him.

If robots really are going to become our friends and, crucially,

if we are going to trust them,

they'll have to be able to make sense of our world.

Like we do.

And the scale of that problem is staggering,

even for the most simple task.

Rush hour in Tokyo.

38 million people live here.

And they're on the move.

How people navigate this megatropolis

is a wonder of the human brain.

Shibuya Crossing, the busiest junction in the world.

Every day, over one million people walk across it.

A single light change can see some 2,500 commuters battling

it across, and it lasts just 40 seconds.

The human brain is the most complex processing machine on the planet.

And to get me across, mine's receiving data through my eyes,

my ears and even through my skin,

and it's using my central nervous system,

my peripheral nervous system,

and a brain with over 100 trillion connections.

But what sets our brains aside from robots and machines is our ability

to deal with the unpredictable.

Made it!

It has taken millions of years

for the human brain to evolve its beautiful complexity,

a journey robots have only just begun.

I've come to Bristol to meet the first robot

to sense the world around it.

This was the vital first step needed

for robots to be able to understand it.

I'm here to see a man about a tortoise, but not just any tortoise.

I'm told this little robot was designed with an artificial brain.

-Hello.

-Hi, Ben. Welcome to the lab.

-Thank you very much.

Professor Owen Holland is the world authority on the tortoises.

Oh, wow!

Robots everywhere!

There you are.

This is brilliant. What have we got here?

We have Ian, our chief technician,

who is responsible for the tortoise construction,

and a couple of tortoises.

First built in 1948,

the mastermind behind these artificial animals was neurologist

Grey Walter.

His robots had two very basic sensors - sight and touch.

In a simple villa on the outskirts of Bristol lives Dr Grey Walter,

who makes robots as a hobby.

They are small and he doesn't dress them up to look like men.

He calls them tortoises, and so cunningly have their insides

been designed that they respond to the stimuli of light and touch

in a completely lifelike manner.

This is great to watch. What was he trying to achieve?

Well, he was a physiologist. He was interested in how the brain worked,

and he knew that he could never build a model with as many parts

as the human brain - ten billion -

so he started thinking maybe it's the number of connections

that is important.

This model is named Elsie,

and she "sees" out of a photoelectric cell

which rotates above her body.

When light strikes the cell,

a driving and steering mechanism sends her hurrying towards it.

But if she brushes against any object in her path,

contacts are operated that turn the steering away.

And so, automatically, she takes avoiding action.

And so he designed these robots basically to use only two

what he called nerve cells

to actually get different types of behaviour,

and he largely succeeded.

It's an incredible achievement. Did he make a brain? Was that it?

The smallest possible brain.

-Yeah.

-But his point was, out of these very few ingredients,

and some cunning design, you can actually show different behaviours

-that have the characteristics of life.

-So, from your capacity,

would you say he made synthetic life?

I would say of a sort. Extremely simple.

And I regard him as the first pioneer

of what we call real artificial life. He was building real things

and saying these are behaving in a lifelike way.

Looking at his primitive robots

really does take me back to my student days, when I learned about

primitive forms of biological life.

This reminds me of early organisms that had eyes.

So, things likes snails, trilobites,

even woodlice that we get in our gardens.

They're very, very simple eyes, but they're photoreceptive,

they go towards or away from light,

-and that's a similar sort of thing here, isn't it?

-It is.

This is like just having one retinal cell.

It's like having an eye with only one element,

so it can't do an image or anything like that,

but it can detect intensity of light, and that's all you need

to do, that's all the early animals needed to do,

and this is an example of the robotics following the path.

Start simple, get complicated.

It is, and it's responding to the environment. And what you've got

in evolution that has taken hundreds of millions of years, this is,

I guess, the first robotic eye,

and it's evolved from this to where we are now

within a few decades.

-Essentially, yes.

-It's rapid evolution of robots yet again.

Owen's robotic tortoises are exact replicas

of Grey Walter's original design.

I want to see for myself how they work.

The touch switch and photoelectric light cell, or robotic eye,

interact with the circuits controlling the motors,

enabling the tortoise to drive and turn.

A proper little retro robot. I'm going to grab this torch...

-Yeah, we'll see what it can do.

-See if he works.

-Yeah, OK.

There we go.

And what you see is the behaviour when it hasn't detected any light.

So, it's scanning round all the time.

We don't like to say it's looking for a light,

but if it finds a light,

something will happen. So, if you'd like to switch the torch on...

Yeah. ..and then point it horizontally...

Yeah.

-..at that.

-Oh, wow.

You will see...

He's come straight for me.

And now that it's hit your foot,

when the touch switch is activated

it drives forward a bit, turns a bit, forward a bit, turns a bit,

and this enables it to escape from almost any situation.

As soon as it's free, it will start scanning for light again.

This is what gives the impression of intelligence, in that you see

a sequence of behaviours that, in context, seems to be effective

and intelligent.

For its time, this is incredible ingenuity and workmanship, isn't it?

Grey Walter was one of the first to show that biological principles

can be applied to the field of robotics.

Although his double-celled organisms were primitive...

..they were taking the first steps to make sense of their environment.

I want to know how far this technology has evolved.

How close are robots to making sense of the world around them?

And can we trust their decisions?

This is quite literally a life-and-death issue for all of us,

because it's starting to play out on our roads.

I'm not the most confident road user at the best of times,

but today I'm having a very different driving experience.

This may look like a normal vehicle, but it's actually a driverless car,

and this is a type of robot that's already within our society.

They're driving on our roads,

and we're putting our life in their hands,

so to speak, on a regular basis.

I've come to Germany and I'm going to let this thing be in control

as it drives me along one of Germany's busiest roads,

an autobahn.

A little bit nervous!

So this is my first time driving on the left side of a car,

it's my first time driving, for a long time, an automatic,

and it's my first time driving in a robot car.

A day of firsts.

Joining me on the ride is safety officer Andreas,

and head of development Dr Miklos Kiss.

I've got to admit I'm nervous. It's like giving...

handing over something very precious.

It's quite a big responsibility, to something,

and I don't quite know how it works.

But any second now, I will hand over the controls to Jack,

my trusty driverless car.

It's that anticipation. I'm not sure what to expect.

Right, let's see what happens.

Come on, Jack. So, I need to press these buttons.

Keep off your hands from the steering wheel and off your feet...

-Off the pedals.

-My feet are off. My hands are off.

-Yeah, that's good.

-So, Jack is acting.

-I love how calm you both are.

Every instinct in my body has just kicked in,

and I can actually feel my adrenaline.

I've gone quite hot and quite sweaty, actually.

I feel like I'm going to veer off, and I know I won't.

What I really want to do is...

-So I can turn around and talk to you now?

-Yeah, you can.

-And...

-And that's safe do, obviously, because...

-That's safe to do.

I'm trying very hard not to think about the fact that right now

my life is in the hands of a robot.

There's a police car!

I feel bad - there's a police car in front of me,

and I haven't got my hands on the wheel!

Sorry, Officer.

No-one is controlling this car right now.

My feet are not controlling any special pedals, my hands are here.

My eyes are closed, I'm on an autobahn in the middle of Germany.

It seems so wrong, but I feel so safe.

And almost like a... Oh, where are we...?

We indicated! Thanks, Jack, I wasn't concentrating!

The car's central computer makes sense of the world around it

using numerous integrated sensors.

Oh, where are you off to, Jack?

Those at the front and rear of the car look left and right,

giving a 360-degree view and a range of 250 metres,

while a 3-D camera scans traffic conditions and road markings.

So Jack is constantly sensing every vehicle around us right now,

I guess in the same way that I'm taking each of my senses

and getting a holistic view. I guess that's what Jack is doing as well.

Well, yes.

The car's computer continuously interprets the data from its sensors

to generate a 3-D map of the world,

which it can then safely navigate through.

It makes split-second decisions to control the braking,

steering and acceleration.

That's a huge amount of computational power there.

What's that comparable to, in terms of other vehicles?

It's comparable to a military jet.

So, we're driving something that's comparable to a jet fighter?

That's it.

I'm getting over the initial shock

of actually letting the car take control.

But I'm still nervous about its judgment.

I can't quite believe its reactions can be as good as mine.

So, worst-case scenario, a really, really worst-case scenario,

somebody turned a car over in front of us now,

it's 100 metres ahead of us,

Jack would be able to respond quicker than I could?

Yeah, quicker than you could.

So maybe we would be caught in that kind of accident,

but at least we would do better than a human would.

-Yeah.

-I would like this car to have superhuman power.

So to solve situations I couldn't do on my own.

We just had a motorbike go past, we've got vehicles all around us,

and it's responding easily as well as I could, if...

as you say, if not better.

-We've slowed down. That was...

-We slowed down.

I'm really enjoying cruising along this motorway.

But I've still got some niggling doubts.

Like, if we did have an accident, who would be responsible?

This throws up complex ethical and legal questions.

If we have a crash right now, whose responsibility is it?

Is it my fault? Is it the car's fault?

I find it very hard to understand

-that I wouldn't be responsible if this car crashed.

-If the system

is engaged and accepted it,

so the handover is done, then the car is responsible.

So the car means...

Obviously, if the system does something wrong,

we at Audi are responsible for what happens.

There are clearly legal issues to resolve.

But what's really surprised me

is that the more I'm being driven around by Jack,

the more I trust him.

I'm trusting the car to do its job.

You are trusting the car to work and to take that responsibility.

Suddenly, we're putting a lot of trust into...

-into a robot.

-Yeah.

It's a big step forward, I think, in our social relationship with robots.

Bizarrely, I do feel comfortable letting a robot take control.

In a couple of years, we won't think about the robot.

It will be natural in daily life.

I think that's the nice part of this.

My grandmother's in her 90s,

and she can still remember the first time she saw her very first car.

And here we are, what, two generations later,

with me with my hands in the air on an autobahn,

letting the car drive for me.

But as much as I have been seduced by the sophistication of the car,

when we're off the autobahn,

it also reveals how little Jack and other driverless cars

truly understand about the world around them.

Please take over driving.

So why am I taking over now?

-What's happening?

-Because we are in a construction area,

and we don't know how the lane markings will be

and how the side barriers will be.

-So we don't drive in construction areas right now.

-OK.

Despite all its sensors and computer power,

without the lane markings of the autobahns,

Jack can't form an accurate enough 3-D map of the world

to navigate safely.

Even I, as a slightly nervous driver,

still have the ability to understand the world so much better

than any current driverless car.

I can not only identify objects,

I know what things really are and do,

and that allows me to make profound connections and decisions

to cope with much more unpredictable scenarios.

Despite the robot car's limitations,

I was still amazed to see how far and how fast robots have evolved

their ability to make sense of the world.

And I wonder if, one day,

it will be possible for robots to understand it in the same way we do.

Can they grasp the true meaning of things

and develop a sense of self

to become individuals?

Yeah? You're going to wave at we now, aren't you?

Could they even become conscious?

For humans, the key to our understanding of the world

is our ability to learn.

To discover what happens when you try to get a robot to learn

for itself, I've come to a lab in Japan.

What have we got going on in here?

So this is one of our most exciting projects,

-it's a robot that can learn.

-Awesome.

Can you tell me about the auto focus of this camera?

This '80s-looking throwback is called Robo V.

OK...

For this experiment,

Professor Dylan Glas has set Robo V a challenge -

can it learn to be a camera shop salesperson?

Can you tell me about the auto focus of this camera?

So, we've got this little robot with one of your colleagues.

Yeah, so the robot's playing the role of a shopkeeper and it's

presenting information about the different cameras.

And the thing we've been exploring lately with this is that the robot

can actually be proactive.

So it's not like Siri or something - it's not answering questions.

It's proactively offering things or suggesting things as well.

No, I haven't.

Oh, wow. That's very cool.

To interact with customers and explain camera functions,

Robo V is reacting independently.

Yeah, this does weigh quite heavy.

What we're exploring here is the concept of,

how can we program a social robot?

Instead of classical programming of robots,

where you program explicitly what the robot should do,

this robot has learnt everything purely from hundreds of interactions

that it observed of other people.

-So this is called learning by imitation.

-What's the price?

Oh, wow. Thank you for your help today.

To create a Robo V's personality,

the camera shop scenario was role-played by human shopkeepers

and customers.

Hi, this one's 2,000.

This camera has 18 preset modes.

Hi, this one is 550.

-550? OK, cool. Thank you.

-No problem.

For Robo V to create this database of hundreds of shopkeeper/customer

interactions, a network of sensors tracked where people moved,

and microphones captured what they said.

-This one's 68.

-68?

-OK, that's really cheap. Thanks.

-Yeah, no problem.

What the robot learns from this is...

again, this is unsupervised learning,

it learns on its own to imitate the behaviour that it's shown.

The locations where people stop in the room, the trajectories

that people use when they walk to different places,

it learns all of these things, as well as clusters of speech.

So maybe you say the same thing in a couple of different ways.

You might say, "How much is this?"

"How much is this camera?"

"How much does this cost?"

And it will notice that those

are very similar and cluster them together.

-ROBO V:

-It's got a five-times optical zoom...

And from this data,

we had the robot automatically learn

the logic of how to be the shopkeeper.

So you've not programmed the robot to be a shopkeeper,

you've not told it what to say or how to respond,

it's learned from, effectively, observing the experiences?

-Exactly.

-Can I have a go?

-Yeah, please do.

Can I have some help, please?

What features does this camera have?

OK. Can you show me this camera?

You want me to buy that one, don't you?

A little bit, yeah.

I'll come look at that one, then.

I love this little robot, he's brilliant!

What's most surprising about my chat with Robo V

is that this almost feels like an actual conversation

I would have with a real shopkeeper.

So this really is a little robot that is behaving just as we would

in a complex social situation, in a real-world situation.

So this is, I think, a very powerful concept, because it can scale up.

If we can capture data of how people interact in the real world

on a large scale,

we can use big data to train robots to do very natural interactions.

Well, instantly, the applications there are massive,

not only as shopkeepers, but right across the board.

You've got medical professions or health care, and everything.

The real challenge is this balance between,

how controllable is the robot and how much does it learn on its own?

So sometimes the robot does things, we're not sure why it does it.

Excuse me?

ROBOT SPEAKS INDISTINCTLY

-But, overall, it tends to do pretty good behaviour.

-It's fascinating.

The other interesting thing about this is that the robot doesn't know

the meaning of anything it does. It's purely behavioural,

it's purely imitating what it saw the person do before.

Right, so it's not picking up on keywords -

"camera" or "cost" or anything like this?

It doesn't even know anything about English.

-It's learning through imitation, through experience.

-Exactly.

-Wow.

What's blowing my mind is Robo V's behaviour is so humanlike

that I really believed it had learned to understand

what I was saying.

But even if it didn't, does that really matter?

It can still sell cameras.

As we move forward, it becomes a philosophically interesting problem,

because now we're really reflecting on how do we learn? How do we think?

How do we, you know, ascribe semantic meaning to things,

and structure, you know, things in the world?

And these machine learning techniques have provided

a very interesting lens through which to view

the way we do our own thoughts.

So, in the future, I think that these learning systems are really...

a part of us. Technology is always a part of who we are

and part of our identity, and this is going to allow us to grow in ways

we've never been able to grow before.

It's an extraordinary idea -

that in trying to teach robots to learn human cognitive abilities,

we may also learn more about how we think ourselves.

The key to this may be to teach robots

not to simply mimic our behaviour

but to develop a conceptual understanding of the world

for themselves,

so they can generate humanlike thought and behaviour spontaneously.

I've come to Plymouth University's

Centre for Robotics and Neural Systems

to meet a team of scientists that is trying to do just that.

Their robot is called iCub.

-There he is.

-This is the iCub.

The famous little iCub.

And I say little, I mean, it's astounding just how much

he resembles - I keep saying "he" already - a small child.

At one metre tall and weighing 22 kilos,

iCub not only looks like a child, but learns like one, too.

Angelo Cangelosi, Professor of Artificial Intelligence

and Cognition, is his guardian.

It's almost like a two-year-old child, and in fact,

like a two-year-old child,

we're going to teach it the name of objects, one word at a time.

That's what children do between 1½ years of age and two years.

You say "teach" - how does it learn? What has it got in there?

The robot has a simulated brain and, as the brain of a child,

is able to associate, to learn the correspondences

between the sound of a word and the picture of an object.

iCub is equipped with cameras to see...

..microphones to hear...

and even smart skin to touch.

The information it gathers from the stimuli around it

is fed into an artificial neural network -

a computer system inspired by the human brain.

iCub is not simply mimicking human behaviour...

..it is trying to discover for itself

the relationships between what it can see,

what it can hear and what it can touch,

just like a child.

I want to see how it learns.

Right. OK, iCub, let's put the ball there so you can see.

Learn "ball".

I like to learn.

This is a ball.

OK. Brilliant.

Let's try...

Try another one. What have we got here?

OK, iCub.

Learn "cup".

I like to learn.

This is a cup.

Well done. Right.

OK, so we've taught iCub two new objects.

How do I know if he's actually learned this or not?

Let's ask him to name them.

-Right, OK.

-So if you show an object, you can then ask for the name.

What's this?

It should be a cup.

It is a cup! Well done.

OK. I'm going to really test him.

and see if he can find the one I'm asking for.

That's close enough.

OK, iCub, find cup.

OK. Now I'm looking for a cup.

Oh, his eyes are moving, his head's moving...

-Yes.

-..and he's tracking the cup. He's not interested in the ball.

Try to show the ball also.

He doesn't care. He wants...

-He likes the cup.

-I mean, his eyes... He wants the cup.

He's not interested in the ball whatsoever.

This is because he had learned the two objects,

and therefore it's following what we ask it to do.

This is incredible. So we've literally

just taught this cute little robot...

..A two-year-old robot the names of objects, like a two-year-old child.

As toddlers interact with the world around them,

they learn from one experience to the next,

making connections between what they can see and hear

to form the basis of context and meaning.

These become the building blocks of intelligence and reasoning.

There are things which are harder.

You can recognise a cup because of its shape and its colour,

you can recognise a ball, again,

because of the different shape compared to a cup.

But what about teaching a robot or a child

to understand the number one and number two.

How would you do this?

One, two, three?

And, like us, the more we learn,

the more complex the tasks we can tackle.

Oh, he's looking as well.

One.

Two.

This is great. What's going on behind the scenes as he's counting?

-Three.

-We have a brain, an artificial brain,

that's been trained to learn to associate sounds,

number words in this case, with its finger position.

By doing this, the robot is actually able to use its body

to learn that there are sequences which are fixed.

For example, one comes before two, two before three, and so on.

I guess that's why iCub is so special,

because you've got that wonderful integration

between the cognitive capability up in here,

but also that physical embodiment.

You've got the two things combined, haven't you?

This really shows why a body is important for a robot,

the same way a body is important for a child.

Children learn by using their motor skills to explore

the physical world around them through touch and movement.

As their body interacts with the environment,

they learn from each new experience.

iCub does the same.

In tiny little steps,

it is trying to form its own unique understanding of the world,

and what things actually mean.

Just to be very clear,

this is this little robot learning to experience the world around him,

to understand more, to have a greater potential of...

-Yes.

-..cognitive capabilities.

Where do you see robotics, like iCub,

in a generation or two generations' time?

Will iCub have grown up and gone to university,

and gone and learned about the world around him?

I can see this in the longer term,

I don't see this happening in the next five to ten years.

We talk about this happening in 20, 30 years' time,

and it seems a long while off.

We've been evolving for tens of millions of years,

and you've got this little entity

that's learning about the world around it now,

and it's going from a blank slate

to this seeing, interactive, responsive little unit.

I think that's both the exciting point

and the scary point with robotics.

At the same time, we are in control,

so we are determining the evolution of these systems.

For now.

On this journey, we've met some incredible robots.

They're preparing for a voyage to Mars,

becoming our friends and companions...

My feet are off, my hands are off.

..navigating us through a chaotic world...

..and some are even able to learn like us.

-Learn "ball".

-For me, this is the most exciting time.

I like to learn.

This is a ball.

We are living right at the moment

when robots start to gradually piece things together,

the first tiny scraps of meaning,

to create their own unique understanding of the world

and themselves.

He wants the cup.

Once they've achieved this, we will be on the brink of a new era.

There is no doubt that robots will continue to evolve

and become more and more intelligent and that, one day,

it just might be possible for them to develop consciousness.

Imagine a robot that could feel the way I feel,

that could be moved by strong emotion,

that could love the way that I love my daughter.

Wouldn't that be incredible?

When I started this journey,

my main concern was that if robots could develop minds of their own,

they might become a threat.

But now I've started to spend more time with robots,

I do feel like I can trust them.

It's responding easily.

And if robots really could one day become conscious,

we need to think not just about how they might affect us,

but how WE could affect THEM.

But perhaps my biggest fear right now,

as we progress towards conscious machines,

is not what we will need to do for robots,

but what we will discover about ourselves.

The whole of our society, our law,

our education is based around consciousness,

making conscious decisions.

And if we show that, well, actually, that's quite trivial

and we can reproduce it in an afternoon in a lab,

then it's going to make people think,

"Well, how important is human life because it is conscious?"

Ultimately, the rewards will be positive,

but you have to be very, very careful.

Socially, it might be disruptive.

The extraordinarily fast evolution of robots

really is going to change our place in the world,

and that raises urgent social issues for us all.

We need to be responsible, to make sure that we stay in control.

We have the opportunity right now

to prepare for conscious robots

that think and feel in the same way we do,

to prepare for what I think is the inevitable.

Investigate the past, present and future of robots

and their effects on our lives.

Go to the address on screen

and follow the links to the Open University.