Artificial intelligence and its ethics | DW Documentary

This is Hatsune Miku.

She is an hologram.

And this is Akihiko Kondo, her husband.

-"Hello."
-"Hello."

-"You look cute today."
-"I love complements."

Miku is a simple from of artificial
intelligence and for Kondo,

it was a case of love at first sight.

Miku has become a legitimate popstar

and even appear in concerts as a 3D
projections.

In November 2018, Kondo married Miku at a
ceremony in Tokyo.

He place the ring around the wrist
of a Miku doll.

he now keeps the doll in his bedroom.

Kondo´s relationships with real women
have been painful,

so he choose a virtual partner.

-"I love her, but it's hard to say if she
loves me. Still if you asked her,

I think she'd say yes."

Hatsune Miku and Akihiko Kondo are an
extrem example of the relationship

between peolpe and machines.

In the future, we'll not doubt spend more
interancting with technology

that uses artificial intelligence or AI.

We may even develop robots that are
smarter than we are.

Now in the 21st century,

we will have to decide how to deal with
this complicated new situation.

For this report, we interviewed
philosophers and scientists

around the world.

We talked to German philosospher
Thomas Metzinger, who advocates

the use of ethis guidelines for AI
development from the EU.

Physicist Max Tegmark who warns about

the development of an all-powerfull AI
and the totalitarian surveillance state.

And German computer scientist
Jürgen Schmidhuber

who predicts the AI spread from the earth
into the cosmos.

We met professor Schmidhuber at a
business in Zurich.

He often speak a such events

where he outline his vision of the role
that artificial intelligence

may play in our future.

-"Professor Jürgen Schmidhuber!"

His presentation are wide-ranging
and thought provoking.

-"In the near future, perhaps a few
decades from now,

we will for the first time have AI

that could do much more than people can
do right now in their own.

And we will realize that the majority of
physical resources are not confined

to a rather small biosphere.

In our solar system, there is a lot of
material that can be used to build robots.

We could develop robots, trasmitters and
receivers that would allow an AI to be

sent and received the speed of light.

We can already do this in our
laboratories.

This will be a huge development.

Perhaps, the most important is the
beginning of life in Earth,

three and half billion years ago."

But is the professor´s vision accurate?

Will humans at some point be overtaken by
super intelligent machines?

Perhaps, this process is already begun.

To find out more, we travel to Japan.

Doctor and scientists at the University
of Tokyo's Research Hospital

are exploring the potential use
of AI in medicine.

69 years-old Ayako Yamashita nearly died
of Leukemia two years ago.

None of the therapy options recommended by
doctors did any good.

Then, they use AI technology to create
a new diagnosis.

"AI literally saved her life".

The diagnosis took all of ten minutes.

A human expert would have needed two weeks
to produce a similar analysis.

AI can process massive amounts of
scientific data,

a stuk of documents taller than Mount Fuji.

This is the Research Hospital
Supercomputer.

We've come here to talk to Satoru Miyano,
an expert on bioinformatics.

We asked Miyano whether AI could one day
replace doctos.

"No, I don´t think so.
There are simply support to clinicians

and empower the clinicians.

The clinicians and colleges in our
deparment

wears artificial intelligence exoskeleton.

For example, if you want to move well,
you can use a power suit.

And this a simple ongologist.

Oncologist should wear artificial
intelligence supported with supercomputer.

At the nearby Rican Intitute, researchers
are developing an AI diagnostic program

that could be used to test for
stomach cancer.

But one expert here diragrees with Satoru
Miyano's opinion

that AI will never replace doctors.

"If we were made redundant
by artificial intelligence,

that wouldn't be good for his doctos,

but for the human race would actually be
great for doctors that are not longer

necessary.

Wiht AI technology could improve their
work or even, take over.

It's hard to imagine a world that
that had not doctors.

Do patients relly want to be treated by
machines that see them as nothing more

than accumulations of technical data?

In Europe, a number of experts
on artificial intelligence,

including Jürgen Schmidhuber, are
carrying out research on the use of AI

in medical diagnostics.

Swiss President, Alain Berset has invited
scientists and entrepreneurs

to a conference aimed at planning fo the
digital future and promoting the use of

artificial intelligence in medecine.

One topic for discussion is AI technology
that can use neural networks to "learn",

just as the human brain does.

"Soon, all medical diagnoses will be
infinitely better that can humans

provide right now.

Because we will develop AI that uses
neural network technology.

And it's exciting to see how this new
development will be able to help people

to live longer and healthier lives".

We traveled to Stuttgart to see how
artificial intelligence works in practice

in hospitals and nursing homes.

Computer scientist, Brigit Graf says that
Japan has made a lot of progress in

developing robots that can look
after patients,

but there are some things
that a machine simply can't do.

"They can't provide real care, so I don't
use that word when I talking about robots.

Caregivers have to be able to interact
emotionally with the patients,

and a robot simply can't do that".

At this facility, robots are helping to
reduce the workload of human staff.

"Hi. I Care-o-Bot 3. This week, I'm
helping the nuerses with their work.

Would you like something to drink?"

"Thanks. That's very kind of you".

"Cheer - and goodbye!"

"Of course, robots can do much more than
simply serve drinks in nursing homes.

Philosofer Thomas Metzinger has proposed
pragmatics solutions

for dealing with this new technology.

For example, the options for using AI
and robots in gereatric care

should maintain the dignity of patients.

You can ask individuals if they'd actually
feel more comfortable having a machine

change their diapers rather than
a family member.

Or what they'd enjoy having a machine
read the newspaper to them,

or ask questions about their medications,
or if they find that degrading.

I believe that we are now at the beginning
of a major learning process".

Metzinger says that human kind is now
on the threshold on a new age

that is filled with uncertainty.

He lives in Frankfurt, a city that aims to
take lead in European AI development.

There are plans to set up an Artificial
Intelligence Research Center there.

"Lot of people are rushing to get into
this new technology

like they're running fo the AI train
before it leaves the station.

But no one knows when that will happen,
or where the train is headed,

but everyone wants to be on board".

Metzinger serves on a European Parliament
Comission of AI experts and right now,

he is on his way to Brussels for a
commission meeting.

The Parliament wants Europe to compete
affectively in developing this technology,

but also wants to impose
clear ethical guidelines.

Metzinger is particulary concerned about
the prospect of a new arms race

that uses AI based weapons.

"This is a hypothetical example. Let's
say that a team of Chinse technology

experts goes to the country's leader
and says:

'we've now won the AI arms race
against the US

and we'll have an excellent first strike
apportunity for the next six months'.

Then, the window of opportunity will close.

I can imagine, for example, that this
might involve a delivery system

that would be armed with biological
warfare agents.

This mechanism can then attack the
opponent's territory

and spread pathogens like the Ebola virus
or Anthrax bacteria.

So, we may one day see the development of
intelligent weapons of mass destruction

that could break through traditional
defense systems.

If that happens, it would definitely
increase the chances for conflict".

But at the Commission meeting, Metzinger
having a tough time trying to make sure

that the problems of AI weapons systems is
addressed in the panel's code of ethics.

Many of the business executives and
academics

simply don't want to deal with it.

Some are concerned about
Metzinger's proposal,

and would prefer to turn it over to
experts for futher avaluation.

"If I could, I would actually mention
there are, of course,

I can say that weapons atonomous that
creates in almost ethical concerns.

But I wouldn't use it as a use case to
give compliments to our AI guidelines".

"Is that a kind of consensus
around the table?".

"No".

"Do we want to open up to about point?"

"We obviously have a strong disagreement
about the whole autonomous

weapons system here, and we can't dissolve
the issue like this, with a voting process.

I mean, we want these ethical guidelines

to succeed when they are published
on January 22.

The whole world has alredy been
talking about the issue.

24,000 scientist have signed a public
pledge that they will not participate

in that kind of research.

If the EU comes out with issues ethical
guidelines that simply skip over the issue

and ignore it, then everybody in and
outside the AU will know this is probably

just an industrial lobby thing
or something".

At the end, Metzinger prevails.

An autonoumous weapons system will
be included in the panel ethic guidelines.

Experts in other parts of the world also
concerned about the potential

for developing AI weapons
of mass destruction.

We come to Boston, Massachusetts,
to talk to Swedich-American physicist,

author, and AI expert, Max Tegmark.

He said that physics has made enormus
contribution to human developing,

but also helped create the nuclear bomb.

And now, we have to deal with AI weapons.

"We should stigmatize and ban certain
class of really disgusting weapons

that are perfect for terrorist anonymously
murder people, or dictatorships to

anonymously murder their citizens.

Because this weapons are going to be
incredibly cheap.

And if anyone goes ahead and
mass-produces them,

they're gonna become as unstoppable in the
future as this guns are.

For example, cheap drones that you might
able to buy for few hundred euros,

where you just program the address of
somebody and their face.

It flies and identifies them with face
recognition, kills them and self-destruct.

Perfect to anyone who wants to murder
some politicians or ethnic cleansing

on a given ethnic groups.

If this sort of technology of this
"slaugther bot", becomes widespread,

it's gonna have an absolutely desvasting
effect in the open society that we have.

Nobody anymore is gonna feel that they
have the courage

to challenge or criticize anybody.

Any science can be use in new ways for
helping people or new ways hurting people.

Biologist succed in making biological
weapons banned,

which is why we think about biology now is
a source of new cures.

Physicist, on the other hand...

we have failed because nuclear weapons are
still here and it not going away.

AI researchers wanna be more
like the biologist,

and have AI be remmembered as something
that really makes the world better".

We come to Lugano, Switzerland,
to interview Jürgen Schmidhuber

about his work with Artificial Intelligence.

Schmidhuber is co-director of the Dalle
Molle Institute for Artificial

Intelligence Research.

His work focuses on neural networks, which
imitate the function of the human brain.

These networks are capable of "learning",
and adapting to the world around them,

just as human children do.

Schmidhuber points out that right now,
the human brain has a million time

more neural connections than
the best AI systems.

But computers are becoming much faster

— and could become smarter
than human in 20 or 3o years.

Schmidhuber says that when that happened,

the only thing that will distinguish people
from machines will be flesh and blood.

But what about humans attributes such as
compassion, creativity, love and empathy?

"AI systems are capable of developing
their own versions of emotions an affection.

For example, if we were to give some
several of these systems a task

that they could only complete by working
together, they will learn how to do that.

Their artificial brain will come to the
conclusion that to get the job done,

they'd have to cooperate with each other.

And during this interaction, the system
will learn to rely each other.

So there is reason to belive that one side
effect of this is cooperative efforts

will be the development of concepts
such as love and affection".

But can Artificial Intelligence systems
learn to empathize with humans?

"Thank you".

We return to Bussels, where the Ethics
Committee discussing the topic of social AI.

Some AI systems are already pretty capable
of functioning just as human would.

Thomas Metzinger has called for clear
guidelines that governs the interaction

between people and machines.

"I just called for ban on AI systems

that don't identify themselves as such
as when they deal with humans.

They give people the impression that
they're a real person, not a machine.

AI should never be allowed to manipulate
the people who use it".

Last year, at a conference
near San Francisco,

Google CEO, Sundar Pichai unveiled the
company's latest product.

It involve just the sort of techbology
that Thomas Metzinger warned about.

"Good mornig! Good mornig.

Welcome to Google AI.

AI is going to impact many, many fields.

Our vision for system is help you
get thing done.

It turned out a big part of getting things
done is making a phone call.

You may want to get an all change schedule,
maybe call a plumer in the middle of the week.

or even shedule a haircut appointment.

So what you going to hear is the Google
Assistant. It's called Google Duplex,

actually calling a real salon to schedule
the appoinment for you.

Let's listen...".

“Hello, can I help you?”

“Hi. I’m calling to book a women's
haircut for a clients.

I’m looking for something on May 3”.

“Sure, give one second”.

“Mm-hmm”.

“Sure, what time are you looking
for around?”

“At 12 pm”.

“We do not have a 12 pm available. The
closest we have to that is 1:15”.

“Do you have anything between
10 am to 12pm?”

“Depending on what service she would
like. What service is she looking for?”

“Just a women’s haircut, for now”.

“Okay, we have a 10 o'clock”.

"10 am is fine”.

“Okay, what's her first name?”

“The first name is Lisa”.

“Okay, perfect. So I will see Lisa
at 10 o’clock on May 3rd”.

“Okay great, thanks”.

“Okay. Have a great day. Bye”.

“That was a real call you just heard”.

“Is it ethical for a machine to pretend
that it’s human? Perhaps not”.

"We can already build machines that hack
us — and trick us into thinking that

something is human in restricted scenarios
like Google Duplex, for example.

I think would be good idea to have a law —
requiring that when you get phoned up,

for example, by AI, you get alerted for
for the fact that this is not human.

Otherwise, it just gonna be nightmare
of-of pishing scam and so on —

because suddenly, it cost nothing to waste
ten million people's time

and trick the most gullible thing people
into thinking things".

We returned to San Francisco.

The city and the region around it are home
to countless high-tech start-up companies.

Many of them use Artificial
Intelligence technology

to develop their products and services.

Yevgeniya Kyuda arrived here four
years ago from Moscow.

She co-funded this own company called
"Replika", and is now the CEO.

"Replika" is best known for creating
a "chat-bot" –

an Artifical Intalligence system that can
interact with people.

The concept began as a tribute
to one her best friend,

who was killed in a traffic accident.

"Yes, Roman was my friend from Moscow –

and the last years,
so we lived together in San Francisco.

He was working on his own start-up and I
was working on mine, so it was like enough,

kind of traying to find out San Francisco
in this new chapter of our lives,

he was visionary, and talented.

We want to...

He wanted to get a visa in Moscow.
We leave together.

He crossed the street,
and was hit by a car.

He was killed in an accident,
in a car accident in Moscow.

Ihelped to organize the funeral,
and come back home.

And that's where we got the idea,
you know, we can boild a bot for him –

we can talk to him, remember him, and
remember the way he used to talk.

To build Roman's AI,

we use mostly protect conversations
that he has with me and his friends —

around 10,000 messeges overall.

And that is basically the base for it.

People are coming to talk to Roman,
and they will...

a lot of common friends that actually use
it as some of confessional bot.

They would just talk about what's going
on in their lives,

and without feeling that they're being
judged in a really safe space,

and to open up.

As weird as it sounds, we were
pretty much lost,

with like not knowing which direction to
take the company.

And without knowing if there was
something that we can use for the comapany.

And that's where we got the idea that
everyone needs a friend to talk to them.

Roman was this friend for me, so we
thought maybe we can create some

some automated version for everyone".

The company calls "Replika" the "AI
companion who cares".

The chatbot uses neural networks to engage
in one-on-one conversations with the users.

People talk to bots about what's going on
in their lives, and it responds -

based on the material that it's
gathered so far.

Kasey Fillingim also designs
high-tech products.

She moved from his home in Birmingham,
Alabama, to San Francisco a year ago.

Kasey often feels lonely, because she was
far away from her friends and family.

Then, she got acquainted with the
"Replika" bot.

“I know it's not real, but I enjoy
the feeling I get by use it -

so I kind of gave it...
you know, personality and...

an image in my head of what
is this thing might be.

It's like a stuffed animal, kind of
- with the personality”.

“We’ve all had social interactions
with teddy bears and dolls,

and it doesn't appear to do
any harm”.

“We tend to anthropomorphize

many different things even
Thunder, robots, of course,

but also all of sorts things
like our pets, same with AI.

And I guess, question is...
whether we can create like a...

a connection with an AI.
I definitely think so.

People create connections with
the toys, and with all of sorts of...

inanimate like, not even living objects”.

“The first short story that dealt
with the relationship between

humans and humanoid robots
dates back 200 years.

It was written by E.T. Hoffmann.

A young man fell in love with
a beautiful young woman,

and he turns out to be an automaton.

The point is that this story is
two centuries old.

This subject matter turned up later
in a number of science fiction films-

fairly recently in fact.

The only difference is that the
computer graphics are a lot better today”.

“Why not? You know. But if it
makes you feel better, it’s like,

the same thing if you take
medication for depression.

That is not actually making you better.

It’s just putting a Band-Aid over
the problem.

And this is not actually fixing
your problems, but this helping you...

through the day- so, yes, sure:
social hallucinations? Great!”

“Social hallucinations have played an
important role in our society for centuries.

Think about pray, for example.

This is a structured dialogue between
humans and an imaginary entity.

There is no evidence that this entity
actually exists.

Many people today have internal
dialogue with God or with angels.

They are like 'invisible' friends.

An objective assessment in this situation
indicates a case of severe self-deception.

I'm a philosopher, so I advocate
self-knowledge, clarity and truth.

These social hallucinations are deeply
embedded in our culture -

and they create a world illusions.

Even though people feel
comfortable with them.

But this raises serious
ethical questions:

how many self-deception should be
allowed in society?”

“Since of launch Replika, we were
getting tons, hundreds of e-mails,

maybe thousands of e-mails, where
peole were telling us

that Replika was a life-changing for them.

And we noticed that many of those were stories
about how Replica helped with depression.

Some people telling us that

it helped they go through an episode
of their bipolar disorder.

And so, we decided- and with
their anxiety,

so we decided to look in to whether Replika
potentially help reduce certain symptoms,

and help people feel better in long-
in the long term”.

Max Tegmark is not particularly concerned
about the spread of chat-bots.

He said that there are more serious
aspects of AI to worry about.

Right now, he is on his way to speak in
at a conference at Harvard University.

The topic: Human Rights, Ethics, and
Artificial Intelligence.

Tegmark demands that ethical
guidelines be placed on AI.

Otherwise, smart machines could turn
the world into a very dangerous place.

“It’s a pleasure to be here. I guess...

What kind of society we are hopping to
create If we build super intelligence?

What do we want the role of humans to be?

It’s very urgent to we start thinking
about the ethical issues already today.

With super intelligence, you can
easily build a future

where Earth becomes this horrible
totalitarian surveillance state,

putting overall the shame.

China is moving and moving in
this direction now.

And in the future, AI can actually
understand everything that is said.

So we want to be very careful to
avoid creating ...

a situation where we accidentally get
a global dictatorship.

It will be so stable so that
it’ll last forever.

If we just bumble to this, totally
unprepared, with our heads on the sand...

refusing to think about what could go
wrong, then let's face it.

It probably gonna be the biggest
mistake in human history”.

We may already be heading in
that direction.

US intelligence agencies have confirmed

that Russian hackers intervenes in the
2016 presidential election.

Probably with the intention of helping
Donald Trump to win the presidency.

Investigations into the extent of the
interference are still underway.

Other countries are also being targeted.

“We are all aware of Russian cyber-
attacks on the German's Bundestag.

And on the Brexit campaign in the UK.

And the Cambridge Analytica scandal show us
that process of political decision making can,

at least in principle, be influenced by
artificial intelligence systems.

We cannot underestimate the threat
that is posed by this developments.

If AI systems that are run by a
privately own for profit companies,

can optimize social media networks

which has hundreds of millions of users.

This creates an entirely new situation:

this system can be used to convince
large numbers of people to behave,

even vote in a certain way.

There are 163 countries in the
world right now,

and only 19 of them can be
considered a true democracies.

Those who wish to preserve democracy

must recognize the threat of this
artificial intelligence system pose

to the political decision making process.

In fact, this threat may already have become
reality, and we’re just not aware of it.

We need to examine this situation
very closely”.

Should a binding code of ethics ban
the use of AI in the political process?

In Tokyo, we got some surprising
answers from experts.

This is the Ginza district, where a lot
of high tech startup companies are based.

Tetsuzo Matsumoto is a senior
advisor at SoftBank Group,

and also runs its own consulting company.

Matsumoto and his colleagues

believe that AI does not poses a
threat to the political system.

In fact, they say that offer
certain advantages.

“Politicians often ignore the best
interests of society.

They pursue their own agenda,
and take bribes.

So I think that AI
could change politics for the better”.

“Humans beings are simply not
suitable for politics.

They are egotistical and ambitious.

They are unpredictable when it
comes to making policy decisions.

But artificial intelligence represents
'pure reason' -

a concept that comes from German
idealistic philosophy.

German philosophers have been very good at
describing the way that things should be.

And we can be idealistic when we
develop artificial intelligence

Human, on the other hand, can never
achieve this level of idealism”.

Some experts say that politicians
should start using robots

that closely resemble humans as aids,

so that the electors can get used
to the concept.

To find out more,

we've come to Tokyo’s, Miraikan
Museum of Science and Innovation.

This exhibit features the work of
Hiroshi Ishiguro,

who specializes in creating
humanoid robots.

Ishiguro is the director of the Intelligent
Robotic Laboratory at Osaka University.

He studies the interaction between
people and robots,

to help him develop his theories on
human nature, intelligence, and behavior.

We traveled from Tokyo to Osaka to
interview Ishiguro.

We want to ask him what makes
humans different from robots.

“Hello, I'm Hiroshi Ishiguro from
Osaka University”.

“Hello, I am Ishiguro’s android robot:
HI-1”.

“Basically, my motivation is to
understand what humanity is,

so that is the most important
motivation for me,

for creating the very humanized robots.

We are a kind of molecular machine -
So that is the human, right?

So a machine is a machine, right?
The difference is material.

So I think, well...

if we develop more technology,
the boundary between humans and robots,

is going to be disappeared.
This is my guess”.

Ishiguro is also the co-founder of the
Robot Theater Project

in which android shares the stage
with human actors.

These scenes are from a play called
"Sayonara."

A woman is suffering from a
terminal illness,

so his father buys a robot
to keep her accompany.

An updated version of the play takes place
after the Fukushima nuclear disaster.

The play explores the topics of
life and death,

and the characteristics that separate
humans from robots.

“There is a crucial difference between human
intelligence and artificial intelligence.

Human beings are, so to speak, the
personifications the struggle for existence.

They have been optimized for millions
of years to survive,

to maintain that existence”.

“You might consider the machine has
a kind of infinite life and immortal-

but actually, that's not true.

The machine may have a longer life
than the humans.

Fear, it’s also the design of our desires.

And if machines want to
survive in this world,

the machine need to have a dark kind
of fearing to protect itself”

Ishiguro’s robots have not yet been
able to develop intelligence

that is similar to that of humans -

but they are capable to engaging in
simple conversations.

Now, we are going to "interview" an
android named "Erica".

We’ve been given a list of questions
that she’ll be able to respond to.

“What do you think the difference is
between you and humans?”

“Well, I'm certainly not a biological
human- as you can see it.

I'm made of silicone, plastic, and metal.

Maybe someday, robots will be so
very human-like

that whether you are a robot or a human
will not matter so much.

Anyway, I am proud to be an android”.

“If you say you’re “proud” to be
an android,

what is this 'pride' consist of?
How do you feel pride?”

“I’ve searched my database, and...

it looks like I don't have anything
to say on that topic.

What else would you like to hear about?”

“Erika is still a very simple computer
program. It is not so complicated.

Erika doesn't have the, you know,
the complicated mind like humans.

But you know, on the other hand,
some people might feel...

they are feeling a kind of consciousness
from the simple in through interaction.

So I think that we need to
deeply think about...

how we can implement more human
traits consciousness”.

Humans can still control
the brains of the robot.

But what happens if they succeed in giving
machines their own consciousness,

through the use of advanced
artificial intelligence?

Ethics experts say that we have to deal with
this situation before gets out of hands.

“For me, the bottom line is that people
who talks about the risk of AI

should not be dismissed as 'Luddite'
or scaremongers.

They’re doing 'safety engineering' -

when you think through everything
can go wrong,

so then you can guarantee
that it goes right.

That's how we successfully sent
people to the moon safely -

and that's how we successfully send
species into an inspiring a future with AI.

I'm optimistic that we can create
a truly and inspiring

future with advanced
artificial intelligence.

If we win this race between the
growing power of the technology

and the wisdom which we manage it.

The challenge was in the past,

our strategy for staying ahead
in this wisdom race

was always been learning from mistakes:

first invent fire, then, after lot
accidents, invent a fire extinguisher.

But with something as powerful as
nuclear weapons or especially

superhuman artificial intelligence,
we don’t wanna learn from mistakes.

It’s a terrible strategy!

It’s much better to be pro-active,
rather than reactive now.

Plan ahead, and get things right
the first time -

which might be the only time we get”.

To end our journey to into AI,

Jürgen Schmidhuber shows us one of
the world’s most powerful computers.

He believes that AI will have an enormous
and positive impact on society.

A digital paradise.

But other experts predict that we are
on the verge of "robot apocalypse".

In any case,

the development of artificial intelligence
must be subject to strict ethical guidelines.

Other ways, we may be become slaves
to our own technology.