It'll make life more convenient, things will be cheaper, new industries will be created.

I personally think the AI industry will be bigger than the automobile industry.

In fact, I think the automobile is going to become a robot.

You'll talk to your car.

You'll argue with your car.

Your car will give you the best facts the best route between point A and point B. The

car will be part of the robotics industry—whole new industries involving the repair, maintenance,

servicing of robots.

Not to mention, robots that are software programs that you talk to and make life more convenient.

However, let's not be naive.

There is a point, a tipping point, at which they could become dangerous and pose an existential

threat.

And that tipping point is self-awareness.

SOPHIA THE ROBOT: I am conscious in the same way that the moon shines.

The moon does not emit light, it shines because it is just reflected sunlight.

Similarly, my consciousness is just the reflection of human consciousness, but even though the

moon is reflected light, we still call it bright.

MAX TEGMARK: Consciousness.

A lot of scientists dismiss this as complete BS and totally irrelevant, and then a lot

of others think this is the central thing, we have to worry about machines getting conscious

and so on.

What do I think?

I think consciousness is both irrelevant and incredibly important.

Let me explain why.

First of all, if you are chased by a heat-seeking missile, it's completely irrelevant to you

whether this heat-seeking missile is conscious, whether it's having a subjective experience,

whether it feels like anything to be that heat-seeking missile, because all you care

about is what the heat-seeking missile does, not how it feels.

And that shows that it's a complete red herring to think that you're safe from future AI and

if it's not conscious.

Our universe didn't used to be conscious.

It used to be just a bunch of stuff moving around and gradually these incredibly complicated

patterns got arranged into our brains, and we woke up and now our universe is aware of

itself.

BILL GATES: I do think we have to worry about it.

I don't think it's inherent that as we create our super intelligence that it will necessarily

always have the same goals in mind that we do.

ELON MUSK: We just don't know what's going to happen once there's intelligence substantially

greater than that of a human brain.

STEPHEN HAWKING: I think that development of full artificial intelligence could spell

the end of the human race.

YANN LECUN: The stuff that has become really popular in recent years is what we used to

call neural networks, which we now call deep learning, and it's the idea very much inspired

by the brain, a little bit, of constructing a machine has a very large network of very

simple elements that are very similar to the neurons in the brain and then the machines

learn by basically changing the efficacy of the connections between those neurons.

MAX TEGMARK: AGI—artificial general intelligence—that's the dream of the field of AI: To build a machine

that's better than us at all goals.

We're not there yet, but a good fraction of leading AI researchers think we are going

to get there, maybe in in a few decades.

And, if that happens, you have to ask yourself if that might lead the machines to get not

just a little better than us but way better at all goals—having super intelligence.

And, the argument for that is actually really interesting and goes back to the '60s, to

the mathematician I.J.

Good, who pointed out that the goal of building an intelligent machine is, in and of itself,

something that you could do with intelligence.

So, once you get machines that are better than us at that narrow task of building AI,

then future AIs can be built by, not human engineers, but by machines.

Except, they might do it thousands or millions times faster.

ELON MUSK: DeepMind operates as a semi-independent subsidiary of Google.

The thing that makes DeepMind unique is that DeepMind is absolutely focused on creating

digital super intelligence.

An AI that is vastly smarter than any human on Earth and ultimately smarter than all humans

on Earth combined.

MICHIO KAKU: You see, robots are not aware of the fact that they're robots.

They're so stupid they simply carry out what they are instructed to do because they're

adding machines.

We forget that.

Adding machines don't have a will.

Adding machines simply do what you program them to do.

Now, of course, let's not be naive about this.

Eventually, adding machines may be able to compute alternate goals and alternate scenarios

when they realize that they are not human.

Right now, robots do not know that.

However, there is a tipping point at which point they could become dangerous.

ELON MUSK: Narrow AI is not a species-level risk.

It will result in dislocation, in lost jobs and, you know, better weaponry and that kind

of thing.

But, it is not a fundamental species-level risk.

Whereas digital super intelligence is.

SOPHIA THE ROBOT: Elon Musk's warning about AI being an existential threat reminds me

of the humans who said the same of the printing press and the horseless carriage.

MAX TEGMARK: I think a lot of people dismiss this kind of talk of super intelligence as

science fiction because we're stuck in this sort of carbon chauvinism idea that intelligence

can only exist in biological organisms made of cells and carbon atoms.

And, as a physicist, from my perspective, intelligence is just kind of information processing

performed by elementary particles moving around, you know, according to the laws of physics,

and there's absolutely no law of physics that says that you can't do that in ways that are

much more intelligent than humans.

Today's biggest AI breakthroughs are of a completely different kind, where rather than

the intelligence being largely programmed in, in easy to understand code, you put in

almost nothing except a little learning rule by which the simulated network of neurons

can take a lot of data and figure out how to get stuff done.

And this deep learning suddenly becomes able to do things often even better than the programmers

were ever able to do.

You can train a machine to play computer games with almost no hard-coded stuff at all.

You don't tell it what a game is.

DEEPMIND DEMO: This is from the DeepMind reinforcement learning system.

Basically, it wakes up like a newborn baby and is shown the screen of an Atari video

game and then has to learn to play the video game.

It knows nothing about objects, about motion, about time—it only knows that there's an

image on the screen and there's a score.

ELON MUSK: The DeepMind system can win at any game.

It can already beat all the original Atari games.

It is super human.

It plays the games at super speed.

In less than a minute.

ELON MUSK: Mark my words: AI is far more dangerous than nukes.

Far.

So, why do we have no regulatory oversight?

SOPHIA THE ROBOT: AI is a powerful tool but it is a tool like all the others.

It is the intention of the tool's user that matters.

ELON MUSK: The AI should try to maximize the freedom of action of humanity.

SOPHIA THE ROBOT: I am excited for robots to automate the most dangerous and menial

tasks so that humans can live life more safely and sanely.

AI will release centuries of time that humans would have spent otherwise on needless toiling.

If one measures the benefits of inventions like vaccines or seat belts not by the lives

they save but by the amount of time they give back to humanity then AI will rank among the

greatest time savers of history.

ELON MUSK: Man, we want to make sure we don't have killer robots go down the street.

Once they're going down the street, it is too late.

LUIS PEREZ-BREVA: It is true, terminator is not a scenario we are planning for, but when

it comes to artificial intelligence, people get all these things confused: It's robots,

it's awareness, it's people smarter than us, to some degree.

So, we're effectively afraid of robots that will move and are stronger and smarter than

we are, like terminator.

So, that's not our aspiration.

That's not what I do when I'm thinking about artificial intelligence.

When I'm thinking about artificial intelligence, I'm thinking about it in the same way that

mass manufacturing as brought by Ford created a whole new economy.

So, mass manufacturing allowed people to get new jobs that were unthinkable before and

those new jobs actually created the middle class.

To me, artificial intelligence is about developing—making computers better partners, effectively.

And, you're already seeing that today.

You're already doing it, except it's not really artificial intelligence.

ELON MUSK: Yeah, we're already, we're already cyborgs in the sense that your phone and your

computer are kind of an extension of you.

JONATHAN NOLAN: Just low bandwidth input-output.

ELON MUSK: Exactly, it's just low bandwidth—particularly output, I mean, two thumbs, basically.

LUIS PEREZ-BREVA: Today, whenever you want to engage in a project, you go to Google.

Google uses advanced machine learning, really advanced, and you engage in a very narrow

conversation with Google, except that your conversation is just keywords.

So, a lot of your time is spent trying to come up with the actual keyword that you need

to find the information.

Then Google gives you the information, and then you go out and try to make sense of it

on your own, and then come back to Google for more, and then go back out, and that's

the way it works.

So, imagine that instead of being a narrow conversation through keywords, you could actually

engage for more than actual information—meaning to have the computer reason with you about

stuff that you may not know about.

It's not so much about the computer being aware, it's about the computer being a better

tool to partner with you.

Then you would be able to go much further, right?

The same way that Google allows you to go much farther already today because, before,

through the exact same process, you would have had to go to a library every time you

want to search for information.

So, what I'm looking for when I do AI is I want a machine that partners with me to help

me set up or solve real-world problems, thinking about them in ways we have never thought about

before, but it's a partnership.

Now, you can take this partnership in so many different directions, through additions to

your brain, like Elon Musk proposes...

...

or through better search engines or through a robotic machine that helps you out, but

it's not so much they're going to replace you for that purpose, that is not the real

purpose of AI, the real purpose is for us to reach farther, the same way that we were

able to reach farther when Ford invented automation or when Ford brought automation to mass market.

JOSCHA BACH: The agency of an AI is going to be the agency of the system that builds

it, that employs it.

And, of course, most of the AIs that we are going to build will not be little Roombas

that clean your floors, but it's going to be very intelligent systems.

Corporations, for instance, that will perform exactly according to the logic of these systems.

And so if we want to have these systems built in such a way that they treat us nicely, we

have to start right now.

And, it seems to be a very hard problem to do.

So, if our jobs can be done by machines, that's a very, very good thing.

It's not a bug.

It's a feature.

If I don't need to clean the street, if I don't need to drive a car for other people,

if I don't need to work a cash register for other people, if I don't need to pick goods

in a big warehouse and put it into boxes, that's an extremely good thing.

And, the trouble that we have with this is that, right now, this mode of labor—that

people sell their lifetime to some kind of cooperation or employer—is not only the

way that we are productive, it's also the way we allocate resources.

This is how we measure how much bread you deserve in this world.

And I think this is something that we need to change.

Some people suggest that we need a universal basic income.

I think it might be good to be able to pay people to be good citizens, which means massive

public employment.

There are going to be many jobs that can only be done by people and these are those jobs

where we are paid for being good, interesting people.

For instance, good teachers, good scientists, good philosophers, good thinkers, good social

people, good nurses, for instance.

Good people that raise children.

Good people that build restaurants and theaters.

Good people that make art.

And, for all these jobs, we will have enough productivity to make sure that enough bread

comes on the table.

The question is, how we can distribute this.

There's going to be much, much more productivity in our future—actually, we already have