Original subtitles

So how did we get where we are today.

So here's a kind of short history of AI. Really, to put it

in as simple as possible terms, AI looked like this:

we had people who wanted to build these things, and they had dreams. They had dreams of these

amazing robots. And the robots, they thought,

they talked, they did whatever we wanted the robots do: translate, play chess,

you know, whatever.

And then they started to build.

And really at the beginning it was like kids with tinker toys.

The tools we had available were not adequate to realize our vision,

and in a lot of ways they still aren't.

Okay, and so we had this grand dream, but very kind of limited in practice. And that

really is a short story of AI. I'm now gonna show you a documentary fragment.

This is an except from an old program called NOVA.

This consists of a discussion,

from a modern perspective,

of what AI was like

in the early days, say the fifties.

And it's gonna have interviews, conducted a long time ago

with a bunch of people--some of whose names you'll recognize, who were amazingly

smart people,

predicting on the basis of the first few years of computation, where AI was gonna go.

And now we can look back

fifty-plus years later and kind of think through what that meant and where we are today.

So let's take a look.

"The Thinking Machine".

Hello again.

With me tonight,

is Professor Jerome B. Wiesner, director of the research laboratory of electronics

at MIT.

Dr. Wiesner, what really worries me today is what's going to happen to us

if machines can think, and what

interests me specifically is, can they.

Well that's a very hard question to answer. If you'd asked me that question just a few years ago

I'd have said that it's very far-fetched and today I just have to admit that I don't really know.

I suspect if you come back in four or five years, I'll say sure, they really do think.

Well if you're confused, doctor, how do you think I feel.

We're just really beginning to understand the capabilities of the computers.

I've got some film to illustrate this point which I think will amaze you.

That man isn't playing checkers against a computer is he?

Sure and it plays pretty well.

While most computer scientists saw it as a mere number cruncher,

a small group thought that the digital computer had a much grander destiny.

Being a general purpose machine, it could be programmed to do things

which in humans require intelligence:

play games like checkers and chess, and solve brain teasers.

The field became known as artificial intelligence.

Can machines really think? Even a scientist argued that one.

I'm convinced that machines can and will think. I don't mean the machines will behave like men.

I don't think for very long time we're going to have a difficult problem

distinguishing a man from a robot.

I don't think my daughter will ever marry computer.

But I think that computers will

be doing the things that men do when we say they're thinking.

I'm convinced that machines can and will think in our lifetime.

I confidently expect that within a matter of 10 or 15 years,

something will emerge from the laboratories, which is not too far off from the robot of science fiction fame.

They hadn't reckoned with ambiguity when they set out to use

computers to translate languages.

A five hundred thousand dollar super-calculator

most versatile electronic brain known,

translates to Russian into English.

Instead of mathematical wizardry, a sentence in Russian...

One the first non-numerical applications of computers,

it was hyped as the solution to the Cold War obsession of keeping tabs on what

the Russians were doing. Claims were made that the computer would

replace most human translators.

Of course you're just in the experimental stage, when you go in for full-scale production, what will the capacity be?

We should be able to do, with a modern

commercial computer,

about one to two million words an hour, and this

will be an adequate speed to cope with the whole output of the Soviet Union

in just a few hours of computer time a week.

When will you be able to achieve this speed?

If our experiments go well,

then perhaps within five years or so.

And finally, does this mean the end of human translators?

I'd say yes

for translators of scientific and technical material, but as regards poetry and

novels, no, I don't think we'll ever replace the translators of that type of material.

You know that was the fifties and sixties, that was the early days.

In the fifties, people basically realize that

you could do computations with circuits and the brain was kind of a little bit

like a bunch of circuits so surely we're almost there.

And then people got really excited and this is basically what you saw

in this video.

That people, you know, they got the computer to play checkers reasonably well.

They got the computer to translate, or at least come close. What did that mean? It means

you looked up the words in a dictionary and you output them. Turns out there's a

little bit more to translation than that, and so it was not the case that

in five years they could cope with the output of the entire Soviet Union in a couple hours.

But like think about where they were.

They talk about this five hundred thousand dollar super-computer. In today's

money that's like a billion dollars. Really big expensive computer.

It has less computation than your phone by orders of magnitude.

It probably has less computation than your toaster. It may have less computation than your shoe.

And this thing, could barely like,

you know, output a string.

And they're like next stop intelligence And it looked so close.

And it just wasn't.

And when you go around saying that in four or five years we will have it solved,

and then nothing happens for ten years, for twenty years,

people started trying write down everything they knew because they realized

knowledge was important.

You couldn't act in the world without knowing something about the world.

So they started writing everything down, like,

you know when

ice gets warm it melts and it turns out there's an infinite number of these facts and

they contradict each other if you don't write them down right, and

it turned out that this basically imploded.

This whole industry that had all this excitement behind it,

there is this decision the actually you know

it doesn't work. It didn't work in four years, it didn't work in forty years,

it's just a bust. And that's what people call AI winter. People totally lost confidence

in the whole endeavor. The AI classes were not filled like this and

there was not an industry presence of any of these technologies. People gave up on it as a

pipe dream.

Then what happened?

In the nineties people had a kind of change of heart.

Rather than using the core tools of logic which are still important,

they started using the core tools of probability and statistics. They had a focus on

uncertainty, which turned out to be the key thing you need to manage.

Whatever system you built, it wasn't going to be perfect and it wasn't gonna know everything and you need

to be able to balance all that stuff, the stuff you do know against the stuff you don't know.

And, people started thinking with kind of a much wider range of technical tools

--much, much, much, better computers,

and better algorithms and suddenly we started to have the tools to do something.

And some people were declaring AI spring, that maybe things were starting to

bloom again.