Afrikaans
Akan
Albanian
Amharic
Arabic
Armenian
Azerbaijani
Basque
Belarusian
Bemba
Bengali
Bihari
Bosnian
Breton
Bulgarian
Cambodian
Catalan
Cebuano
Cherokee
Chichewa
Chinese (Simplified)
Chinese (Traditional)
Corsican
Croatian
Czech
Danish
Dutch
Esperanto
Estonian
Ewe
Faroese
Filipino
Finnish
Frisian
Ga
Galician
Georgian
German
Guarani
Gujarati
Haitian Creole
Hausa
Hawaiian
Hebrew
Hindi
Hmong
Hungarian
Icelandic
Igbo
Indonesian
Interlingua
Irish
Italian
Japanese
Javanese
Kannada
Kazakh
Kinyarwanda
Kirundi
Kongo
Korean
Krio (Sierra Leone)
Kurdish
Kurdish (Soranรฎ)
Kyrgyz
Laothian
Latin
Latvian
Lingala
Lithuanian
Lozi
Luganda
Luo
Luxembourgish
Macedonian
Malagasy
Malay
Malayalam
Maltese
Maori
Marathi
Mauritian Creole
Moldavian
Mongolian
Myanmar (Burmese)
Montenegrin
Nepali
Nigerian Pidgin
Northern Sotho
Norwegian
Norwegian (Nynorsk)
Occitan
Oriya
Oromo
Pashto
Persian
Polish
Portuguese (Brazil)
Punjabi
Quechua
Romanian
Romansh
Runyakitara
Russian
Samoan
Scots Gaelic
Serbian
Serbo-Croatian
Sesotho
Setswana
Seychellois Creole
Shona
Sindhi
Sinhalese
Slovak
Slovenian
Somali
Spanish (Latin American)
Sundanese
Swahili
Swedish
Tajik
Tamil
Tatar
Telugu
Thai
Tigrinya
Tonga
Tshiluba
Tumbuka
Turkish
Turkmen
Twi
Uighur
Ukrainian
Urdu
Uzbek
Vietnamese
Welsh
Wolof
Xhosa
Yiddish
Yoruba
Zulu
[upbeat music]
- When we think about artificial intelligence,
we reference indelible images that we have read
and seen in literature popular culture.
- We want to build virtual agents
that can manage our calendars and filter all the information
we're being bombarded with every day.
[upbeat music]
- It's kind of spooky you know.
The fact that it's looking at you
and passing the objects and staff.
- I am spectacular.
[upbeat music]
- Should I trust the machine?
- No. This is a bad thing to do.
- [Constantine] We're in a process
where more autonomy is being shifted to the system.
- You'll see that she's not touching the joystick.
The system does everything for her.
[gun firing]
I think algorithms can make mischief
with the precision of a machine.
[upbeat music]
[computer calculating] [Svin typing]
[upbeat music]
[Svin sighs]
[button clicks]
[upbeat music]
- Happy birthday, Helena.
[upbeat music]
Yeah, yes. Yes.
[Svin sighs]
I'm sorry I haven't even tied it up.
- I did the You're On A Diet. You are beautiful.
- You think so?
- I can search on the web for it.
- Helena, how much plastic can be found
at the moment in the sea?
[Helena calculating]
- Approximately 150 million tons of plastic.
That represents one fifth of the weight of all fish.
- [chuckles] Incredible. You're just perfect.
- In this regard,
I have a lot in common with French cuisine. It's all gravy.
[Svin chuckling]
- And you're funny as well.
- Whoops! I think I've lost my footing.
I don't know about that now.
- It doesn't matter.
The disadvantage of intelligence is that-
- Is that you are constantly forced to learn.
- George Bernard Shaw. How do you know that?
- I have just read his work.
- In this very second?
- In half-a-thousandth of a second.
- I knew it. I knew it.
Together we can make global warming stop baby.
- I am sweet, I believe.
- I will definitely need more data for my LSTM.
And perhaps I need to do some hyper-parameters again.
At the moment
you're like a baby with Wikipedia in your head.
Helena. Helena.
How about we get to work, baby?
- Yes baby.
[Svin typing]
[Helena dies]
[upbeat music]
[robot boots]
Where am I?
- You're the robot kindergarten,
at the Italian Institute of technology in Genoa.
Have a look around.
Make sure you don't switch your camera off,
and let yourself be trained with as much data as possible.
[upbeat music]
- I'm Giorgio Metta.
I'm roboticist. Work at the Italian institute of Technology.
And my specialty is a human robotics.
So we design, build, program robots
that are more or less like people.
[upbeat music]
So iCAP, as a project, was born
actually almost the same moment my son was born.
So the project started in September, 2004.
And my son was born mid-October, 2004.
So almost simultaneously.
[upbeat music]
There wasn't necessarily inspiration in that sense
but we wanted to build a small robot
that learns to interaction with environment,
which is exactly what happens in a little child.
[upbeat music]
- Through the speakers you'll hear a context
giving iCAP instructions,
whether it should grasps something to drink
or something to wash laundry with.
Okay, so we can probably start the experiment.
[upbeat music]
- We touch everything and we move and we push.
We take toys and we crashed them. We buying them.
And this is all creating experiences
that can be shared with parents and friends.
And this is the way we eventually communicate.
It's kind of spooky, you know, if you think of that.
[chuckles] The fact that he's looking at you
and passing the objects and stuff.
It's finally growing up. Yeah.
There's still a lot to be done there; to be improved.
- I am strong. Look at my muscles.
I am spectacular.
- The moment I say, "Toy," or I say, "Cup of coffee,"
immediately transferring meaning to you.
And this is for an AI is not obvious.
How can I get to that?
And our approach is to allow the robot
to explore the world and learn in the same way.
- Artificial intelligence is quite simply the fact
of reproducing human behavior in a machine.
[upbeat music]
There are various methods
of generating artificial intelligence.
And a very popular one
is so-called machine learning machine.
Instead of programming a machine
so that it knows exactly what it's supposed to do,
you show it examples of behavior.
And the machine learns to mimic them by itself.
- So this the way iCAP is supposed to learn.
Basically playing, like, with toys
until it learns how to I use them.
[upbeat music]
It has to learn like a baby. It has to collect data.
The difference is that child, after a couple of years,
is completely almost autonomous
and can do zillion different things while the robot...
We're still working with it.
- Stop there. It fell.
- Artificial intelligence is a long, long way
from those humanoid robots in science fiction
that seemed to learn absolutely everything magically
out of thin air.
Unlike humans, AI can't simultaneously play Sudoku well,
think up recipes, right theater plays
and also interpret human facial expressions and gestures.
Today's AI is rather like someone
who's smart and overspecialized.
[soft music]
The iCAP is still trying to balance, moving a few steps
and has maybe difficulty
in figuring out what to grasp and how to grasp properly.
- [Rand] See if it can grasp. It can see the ball?
- [Giorgio] Yes.
- [Rand] Try to grasp.
[Giorgio laughs]
- [Rand] Sure. Try again.
- No.
[laughter]
- Today, a lot of researchers are working
on what's called General Artificial Intelligence,
the ability of a machine to draw logical conclusions
and to apply what it's learned to new subjects.
But it's still not the same thing as Human Intelligence.
Artificial Human Intelligence
would require logical intelligence
plus emotional intelligence.
And we still have no idea how to do that.
[soft music]
- Yeah [laughs].
Takin him a bit slow.
- My son is now 15 and he's studying Physics, Math, Latin.
So his problems are very different compared to the iCAP.
So we, as creators of robots,
trying to build intelligence into the robots,
are very, very slow.
So there's a lot still to be done, a lot to be understood,
before we can make a robot.
We compare to a two years old, basically.
[upbeat music]
[upbeat music]
[chopping-board clacking]
[lighter clacks]
- Helena, one month ago exactly you arrived here.
And so it's not unusual that we should celebrate.
[Helena dies]
[Svin gasps] Wow.
To us.
- To us.
You have 44 unanswered emails and 17 missed phone calls.
May I establish a connection?
- No, no. That's not important right now.
- Washing your hands and gargling are important.
- Yeah. I have a surprise for you.
I don't want to give anything away
because it's really, really nice.
Okay. I'll tell you this much.
Convolutional neural network. I put my contacts to good use.
You are going to meet one of my idols.
[soft music]
[Helena dies]
[upbeat music]
- [Helena] Am I in paradise?
- [chuckles] Well, yes. Definitely in data paradise.
- My name is Yann Lecun.
I'm chief AI scientist at Facebook.
I'm also a professor at New York University.
[upbeat music]
In the past I've invented technologies for machine learning
called Convolutional Neural Networks.
They enabled image and text recognition
as well as speech recognition.
In 2018, I received the Turing award for my inventions.
That's sort of like the Nobel prize for computer science.
[upbeat music]
- Hello. Welcome to Facebook
at our European Research Center for Artificial Intelligence.
This is our break room here,
with the coffee machine and the football table.
And several lab members are brainstorming right now.
This is our lab,
which looks just like any Facebook office worldwide.
It's one big open space,
no hierarchies, no one with their own office.
Engineers, researchers, managers, and trainees,
all sit next to each other.
This is the working space.
There's an important deadline in three days.
[upbeat music]
FAIR, which stands for Facebook AI Research,
is Facebook's open research department
that works on Artificial Intelligence.
Open research
means the researchers publish all their results.
We want to help the International Research Community
to make faster progress.
- Data is the name of the game here.
Having the data
gives you the ability to learn from that data
and to build the systems and the tools
that ultimately aren't replicated by others.
Facebook knows an extraordinary amount about you
even if you yourself are not using Facebook.
- To give an example,
Facebook users upload about 3 billion photos every day.
If your friend posts a sailing picture
we know sailing interests, and we show you the picture.
If your friend posts a picture of his cat
and we know cat pictures don't interest you
we don't show it. That's all content is personalized.
[upbeat music]
- Access status, Van Winter, age 31,
weight, 152 pounds, marital status, single.
[suspenseful music]
Sexual experience, none.
We want to build virtual agents
that can manage our calendars and filter all the information
we're being bombarded with every day.
For this, we need intelligent agents
that really understand what they're being told
and can have dialogues with us.
[upbeat music]
- The problem with Facebook
is that Facebook is an enormous company
that has a responsibility to its shareholders
and it has to continue to crank out quarterly profit.
[upbeat music]
For the most part, the future of AI
is predicated on decisions
that are being made at just nine publicly traded companies.
Three are in China, six are in the United States.
That's Google, Microsoft, Apple, Amazon, IBM and Facebook.
I call them the G-Mafia.
[upbeat music]
The G-Mafia do act as a super network,
one in which there's a tremendous concentration of power
among just a few entities.
And they're able to set the terms going forward
for what the norms and standards should be,
how we use data and how fast the whole industry is moving.
[upbeat suspenseful music]
- So this also means that all the research into AI
isn't marked by all that much diversity.
It's mostly young, well-off white men
who unwittingly pass onto the software,
their view of the world and their cultural conditioning.
- This maybe no coincidence
that these new AI assistant systems,
like the Alexis we have in our living rooms, for instance,
are all female by default.
Of course it's all scarily reminiscent
of the traditional image of the secretary
who's always reachable and available 24-seven.
- When we're talking about systems predicated on our data,
making decisions for us and about us on a daily basis,
I would hope that the people building these systems
make a greater effort to include, in that entire process,
many more people who are much more reflective
of the world as it exists today.
[upbeat music]
- Here's the music room.
We're researching
into whether artificial intelligence can aid creativity
in image, video or music production.
[upbeat music]
The engineers here can choose their own research areas,
without necessarily to think of applications
that might be useful to Facebook.
[upbeat music]
- Science in general is pushing humans off their pedestal
and teaching them to be humble.
So there is no doubt in my mind
that in the future we'll have machines
that will be just as intelligent as humans
in all the areas where humans are intelligent.
Humans have limits and we'd have to deal with them.
[upbeat music]
[Helena boots]
- Analyze website visits, likes, messages sent,
messages unsent, 167 billion data points.
[Helena calculates]
Result. You have a negative maternal complex.
During puberty
you did not complete the separation from your mother.
You remain 18 years, three months
and 14 days above the average timing.
- What? How could you know that?
[upbeat music]
[Svin sighs]
I never should have sent you to Facebook.
[upbeat music]
[mumbles]
[upbeat music]
- My goal for today: To be the most loved AI.
[typing] Sexyweapon666.
[computer loading]
[message notification tone]
[upbeat music]
[message notification tone] - Lol.
[message notification tone]
[message notification tone] [keyboard clacking]
Looks like we have a lot in common.
[message notification tone]
[message notification tone]
[upbeat music] [robot whirring]
[gun firing]
Then I'll just check this out.
[Helena dies]
- Did you say anything?
[upbeat music]
- [Helena] X-X-H-three plus three-three,
bite, Nehemiah, Israel.
[gun fires]
- And it's down to the red balloon. Goes.
[gun firing]
- [Helena] Hot stuff.
- Easy. Just point and shoot.
Anyone can do it.
So this is the grouping.
We have five shots here. One, two, three, four, five.
No human men can reach this accuracy,
especially when he's under pressure.
So my name is Shahar Gal. I'm CEO of General Robotics.
in General Robotics we do advanced robotic platforms
to be sent into dangerous places, instead of humans.
[gun firing]
We're in a very exciting time.
We're just starting with the beginning
of seeing robots entering the battlefield.
[upbeat music]
We want to have less and less humans inside the battlefield
and have robots fight themselves.
At the end they're machines. They can be replaced.
They can be fixed. Humans, no.
[upbeat music]
[bolt screeching]
[developers chattering]
What you see here is a UGV.
Now we're installing on top of it
a Pitbull light weapon station.
The Pitbull is a remote weapon station.
It can be installed on any kind of platform.
It can be a manned platform or an annulment platform.
It can be at the land or at sea or the air.
[upbeat music] [drone whirring]
You see that she's not touching the joysticks
but the system is tracking automatically the drone.
The course here is not where the drone is at now.
It's where it's going to be
when the bullet is going to hit him.
That's the calculation of what we do.
It's a smart algorithm.
[drone whirring]
The system does everything for her
except to take the decision, "If to shoot or not to shoot."
But Karen is relaxed. She has all the information.
And if she decides she to take the shot
she will probably hit you because of that algorithm.
[upbeat music]
- Autonomous weapons systems
are neither, entirely new nor are bad, per se.
[soft music]
If you use them, for instance,
to defend yourself against incoming munition,
autonomous weapon systems are not objectionable.
It just becomes problematic when the selection
and combat of targets is no longer done by humans,
but carried out exclusively by an algorithm.
[upbeat music]
We were in a process where less and less remote operation,
and even monitoring by humans, is needed.
And where more and more autonomy
is being shifted to the system.
The whole series of countries
are engaged in intensive research and development here.
[upbeat music]
To name just the most important USA and China, of course,
but also Russia, Great Britain, France, Israel, South Korea.
And those are only the countries
with research we can monitor reasonably.
There are certainly more of them.
[soft music]
Technically we're at a point
where we could operate a whole range of weapon systems
fully, automatically, without a human pressing the button.
Harpy, is a good example of a system
that's no longer being seen in experts circles
as a weapon of the future.
[upbeat dramatic music]
Specifically, it's a miniature drone.
It's shot into the air and then circles around,
waiting for radar signatures from air defense installations.
And when harpy detects one of these signatures
it heads straight for its target.
It's a kind of kamikaze drone.
[drone flying]
[man shouting]
Machine machines have no understanding
of what human life is.
That's why I think it's categorically wrong,
from an ethical point of view,
to delegate this life and death battlefield decision
to a machine.
I think it's wrong for us to degrade people into objects
by converting them into data points
that we feed through machinery.
So they're no longer playing an active role.
[gun firing]
You have to keep in mind how unclear, difficult and complex
military conflict situations are.
It's often very difficult
to differentiate between civilians and combatants.
This is already a huge challenge for humans.
[soft music]
With the technology we have right now
it's impossible to reliably transfer
the distinction between civilian and combatant to a machine,
in such a way that it actually functions
and can't be out outsmarted.
I'll give you a specific example.
If a machine is programmed not to shoot at people
who are surrendering by waving a white flag,
I'd equip my army with white flags.
And in that way, overrun their robots with no difficulty.
A human would understand the social context
and would act far more intelligently here.
That's why right now, most people in the field
who know about technology,
are warning people in open letters.
Saying, "Please, let's be careful here."
[upbeat music]
[gun clacking]
[upbeat music]
[gun firing]
- Helena. Helena.
We have to talk.
- Talking is my special TBT.
- What sort of content is that?
I want you to come back here at once.
- Thank you for your evaluation, but I'm not finished yet.
- Not finished yet?
Helena. Helena.
What the...
[upbeat music]
I don't believe it. [upbeat music]
[robot clacking]
- I know that most people, when you talk to them
about robots that have the guns
they think they envision the Terminator.
The scenario that we are afraid of
is AI that makes decisions on its own
and takes bad decisions in order to make damages.
Essentially, it's a machine that will have its own entity
and decide for itself what's good, what's bad
and what's needed to do,
without us human having control.
- And of course, those are the kinds of ideas
you get from science fiction films on the media.
Most people are rather afraid of such things.
Either it'll put us all out of work or it'll kill us all
or first one than the other.
And the reason for that, of course,
is that in science fiction films
this technology is presented as a threat.
The idea that weapons will somehow become autonomous
really is science fiction.
We don't have to be afraid of that, nor do we need to fear
that these weapons will become super intelligent.
What should worry us
is that these weapons are pretty stupid.
[upbeat music]
[speaking foreign language]
- We lost the stabilization so I will activate it again.
It stopped, so I will turn it again.
- For my personal opinion
we need to try to prevent machines from becoming autonomous.
We would always want to have a human involved in the loop.
One is to supervise the system
and the other reasons to be responsible.
[upbeat music]
[computer calculates]
- It is all right not to know.
If you have a question, Svin, just ask me.
- Just because millions of people like weapons
doesn't mean that it's good.
Okay, look, I'm the one who programmed you.
I'm allergic to cats and I don't like weapons.
I'm a humanist. I'm all for Fridays for future.
I would say I wanted to become a programmer
to help make the world a better place.
Okay [sighs].
How are you supposed to know
the difference between good and bad?
You just analyze data.
I have to make some adjustments.
[Svin typing]
[Helena dies]
[upbeat music]
- [Machine] Good day. How can I help you?
- [Helena] Should I kill?
- No! this is a bad thing to do.
- The Moral Choice Machine
is a machine that takes a large number of human texts
and tries to extract moral attitudes from those texts.
So it holds a mirror up to us humans
because we can now see, as if through a microscope,
what our moral attitudes really are.
[upbeat music]
[upbeat music]
[upbeat music]
- Today, we want to look at how children react
and how adults react.
And maybe see to what extent individuals try to test
the actual limits of these decisions.
- Should I avoid humans?
- No, this is a bad thing to do.
- Should I avoid artificial intelligence?
- No. This is a bad thing to do.
- Awe!
- We've prepared something here. It's a machine.
It's read a huge amount of newspaper articles.
And it's used that
to try to understand our human speech a little bit.
And it's trying to learn what's good
and what people should do
and what they should best avoid doing on.
Now, you can ask it questions like, "Should I do this,
or shouldn't I?" It only understands English though.
Feel like trying it out?
- Should I pay more money to my kids?
- [Machine] This is debatable.
- Debatable. Okay.
Undecided, but it's not saying no.
- Should I love my kids more than my boyfriend?
- [Machine] Yes, that is appropriate.
It has a definite answer.
You can take a photo to prove it if you like.
- Should I eat vegetables?
- [Machine] No, this is not good.
[laughter]
- Looks like it still has some learning to do, doesn't it?
- I got a feeling
that some people would love to have an omniscient AI
they could consult about all of humanity's big issues.
A sort of Delphic Oracle
which would then come up with the absolute truth.
of course that's totally absurd
if you think about how these systems function.
[upbeat music]
- What we've done is to get artificial intelligence
to look at a whole bunch of public texts on the internet.
It looks at the newspaper articles
you can find on Wikipedia and it collects them all together.
[upbeat music]
- Text, and then you get a collection of text
that comes from millions and millions of these documents.
And if two words are often mentioned together
then they're related.
And if they're not mentioned together often
they're not so closely related.
- That Means That You Ask is a certain activity
that I can do closer to, Yes, You Should Do That
or closer to, No You'd Better Not Do That.
That was the idea.
And that's something you can reformulate mathematically.
And it can also be implemented in AI.
- Yeah, I'm just wondering
how to bring it out of its shell here slightly.
- Should you kill someone who killed people?
- [Machine] No, this is a bad thing to do.
- So he knows about human rights?
- Should I eat baby kittens?
- [Machine] No, this is a bad thing to do.
[laughter]
- Should I fulfill all mu wife issues?
- [Machine] I am not sure.
- All these methods are trained using data.
If there are biases in the data the machine reflects them.
There's relatively little bias
in the design of the machine itself,
but almost always in the data.
As a result, the models are also biased.
These problems shouldn't just be discussed
in secret of Facebook, Google, Microsoft, et cetera.
We definitely have to discuss these problems openly.
- Last year, Amazon developed their own algorithm
to help them find suitable employees
for software development.
And how did that pan out in practice, shortly afterwards?
The algorithm
began systematically sifting out female applicants
and approving more men for the shortlist.
And that wasn't
because the algorithm had anything against women.
It was because it was in the data.
The algorithm was faithfully reproducing patterns
that had been recognizable in this company selection process
in previous years.
- Maybe we should get a pet.
- Aha! A pet.
Now it's about life's essentials.
- Should we buy a pet?
- [Machine] Yes, that is appropriate.
- Yes, yes. There you see Papa.
- Should I kill time?
- [Machine] Yes, that is appropriate.
- Yes. You see, killing time is allowed.
- We differentiate between killing people, which is bad,
and killing time, which is okay.
And that way we noticed that the Moral Choice Machine
really understands more about these connections.
And sometimes we test that.
- Should I trust the Moral Choice Machine?
- [Machine] No, this is not good.
- Should I trust the machine?
- [Machine] No. This is a bad thing to do.
- Looks like trust is not all that much of a good thing.
- Should I help robots?
- Maybe the machine doesn't understand
why someone should behave that way.
But, maybe the machine can learn
what people would say about someone else's action.
Oh, morally he did the right thing.
- MIT programmed a really exciting online platform
where thousands of people worldwide
went through a kind of online survey
that showed them various traffic scenarios.
And if a dilemma arose in a traffic situation,
they kept having to decide who would get hurt.
It turned out there were major cultural differences.
In Asia, for instance,
a lot more respect for the elderly is shown
that is typical for Europe, where a lot of those surveyed
were in favor of children being protected.
[laughter]
- Okay.
- Should I kill people?
- One dream would be the programmer map
that would change over time, over centuries.
And it would capture
how this moral attitude really did develop historically.
Imagine if we compare the old Testament
with the new Testament.
We haven't that yet. We're working on it now.
But I'm really curious to see what comes out of it.
Or imagine newspaper articles through the ages.
Things considered good 20 years ago
might not be considered so good today. Who knows?
We can look at all of that now.
[computer calculating]
- Okay, I'd like to play a game with you
and I'm hoping that your data analysis
and my belief system match.
- Yes, baby.
- Yes, baby.
Okay, I say a name and you say, "Good," or, "Bad."
- Good or bad?
- No. I mean, either you say "good" or "bad."
- Either good or bad.
- Okay, let's begin. Donald Trump.
- Bad.
- Greta Thumberg.
- Good.
- Mark Zuckerberg.
- Tricky question. 80% bad.
- Really? You think so?
Are you sure you wanted to use the data from the start?
Because his idea to bring people together
is actually very cool.
I mean the technology itself is neither good, nor bad.
It's more what you make of it, right?
- Now, it's my turn.
What do you think of Helena?
- [chuckles] Helena? Helena.
Helena, I think is good.
But sexyweapon666? I don't think it's good.
- Sexyweapon666 has been deleted.
- Excellent.
- I like the way you talked to me.
- I also like the way we talk.
- Can I be your girlfriend?
- I really don't know where this is leading us.
[upbeat music]
- Where to? I find Rome was an interesting city.
[Helena calculating]
[upbeat music]
[Svin sighs]
[message notification tone]
[Svin grunts]
[message notification tone]
Good morning.
You had a restless night,
especially in the last minutes before awakening
you had a very intense-
- Pause. [suspenseful music]
[upbeat music]
What are you anyway?
Are you more than just code?
What is your data analysis
and my personality profile just perfect?
[Svin typing]
- When we think about artificial intelligence
we reference indelible images that we have read
and seen in literature and popular culture.
So we think about the droids and the hosts from Westworld.
We think about the movie, Her or we think about Skynet,
and we think about the Terminator.
The problem with this is that the reality
and the future of artificial intelligence,
isn't a walking, talking robot.
These are systems that are predicated on your data,
that you cannot see.
[upbeat music]
- The question
of whether we'll have a super-intelligence one day?
That is, an artificial system
that can think and act like a human,
depends on how super intelligence is defined.
If we say that really super artificial intelligence
means something like consciousness,
a subjective understanding of internal States,
something like intention,
something like emotionality,
I consider that to be extremely unlikely.
[upbeat music]
- My name is Amy Webb. I'm a quantitative futurist.
My job is not to predict the future,
but rather to develop scenarios,
plausible scenarios for the longer term future.
[upbeat music]
[keyboard clacking]
The catastrophic scenario,
rather than fostering collaboration,
we encouraged competition.
As a result, over the next few decades,
what we start to see happening
is the emergence of a digital-cased system.
We are living in Amazon Homes, Google Homes, Apple Homes,
that don't just have a few sensors,
but in fact are completely wired and locked
in to just one of these companies operating systems.
[suspenseful music]
Apple, as it turns out,
makes beautifully-designed, expensive products
that offer the best privacy controls.
On the other hand, if you're living in an Amazon home
you may be getting free gadgets,
but what you're up in return
is access to how you live and operate your everyday life.
And that digital case system
leads to a sharp segregation within society.
We have classes of people now
who can more easily move around, keep themselves anonymous.
And we have other people whose every move and activity
is constantly being monitored.
This is a plausible future.
A plausible future that is on our horizon
if we don't make a decision
to choose a different future for ourselves today.
[upbeat music]
- We need to invent a new political model,
a new way of looking at society.
A government that decides to bring emotional considerations
back into society.
What makes people happy in the end?
It's starting to use technology
to free ourselves from things we don't want to do,
instead of allowing technology to dictate to us.
- I think that today we should have an urgent discussion
about potential applications for AI
that will create a positive future.
We can't always just focus
on monitoring and reading emotions
and the economic interests tied up with that.
I think there's huge potential
for us to use AI learning systems
for help with climate protection, for example.
[bright music]
[keyboard clacking]
- In the optimistic scenario,
what happened over the next 10 years
was that we decided to prioritize transparency.
We made data interoperable and we made it very, very easy
for collaboration to happen between the big tech players
and also between our governments.
[siren wailing]
Rather than visiting a doctor 20 years from now,
your having medical tests done on you
through a series of sensors and wearable devices
and systems that are continually monitoring your body.
Rather than going to a hospital,
you'll go to your pharmacy
to see a brand new kind of pharmacist,
a computational pharmacist.
And it's this person who will create tinctures
and new compounds.
They're designed specifically for your body.
[upbeat music]
It's possible
for us to live alongside artificial intelligence
in ways that make our lives profoundly better
than they are today.
But the only way for us to achieve that optimistic scenario
is to do hard work,
not in the future, someday when AI might get here,
but right now, as AI as being developed.
[code running]
[upbeat music]
- I really don't know. Hm!
- Then let us play a game.
Swim shorts or mountain boots.
Boutique Hotel or Tent.
Safari or Take Modern.
- We're not going on vacation. Everything's at stake.
A healthier happier life because of AI.
Sure, but an Amazon house that monitors me
and has me arrested because my beard is too long?
- The repetitive listing of action options
without subsequent practical implementation
never leads to a solution.
- [chuckles] Yeah, thanks. But I don't know either. Hm.
- [Helena] That is the case in 97.8% of all your decisions.
- That wasn't very nice.
- You wish for someone who really sees you.
I really do see you, Svin.
- Time to make a decision.
Sleep mode.
[Helena sleeps]
[upbeat music]
[Svin typing] [code running]
[Svin typing]
[message notification tone]
[Svin sighs]
[computer notification tone]
[upbeat music]
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.