Rodney Brooks: Robotics - Lex Fridman 知识图谱

Lex Fridman (02:51.040)

What is the most beautiful?

Rodney Brooks (02:52.160)

It's just mechanically gorgeous. As everything Aaron builds,

Lex Fridman (02:55.600)

there's always been mechanically gorgeous. It's just exquisite in the detail.

Rodney Brooks (03:00.400)

We're talking about mechanically, like literally the amount of actuators.

Lex Fridman (03:04.400)

The actuators, the cables, he anodizes different parts, different colors,

Lex Fridman (03:10.080)

and it just looks like a work of art.

Lex Fridman (03:13.200)

What about the face? Do you find the face beautiful in robots?

Rodney Brooks (03:17.760)

When you make a robot, it's making a promise for how well it will be able to interact,

Lex Fridman (03:23.120)

so I always encourage my students not to overpromise.

Rodney Brooks (03:27.680)

Even with its essence, like the thing it presents, it should not overpromise.

Rodney Brooks (03:31.840)

Yeah, so the joke I make, which I think you'll get, is if your robot looks like Albert Einstein,

Rodney Brooks (03:37.200)

it should be as smart as Albert Einstein.

Lex Fridman (03:39.440)

So the only thing in Domo's face is the eyeballs, because that's all it can do.

Rodney Brooks (03:47.520)

It can look at you and pay attention.

Rodney Brooks (03:52.640)

It's not like one of those Japanese robots that looks exactly like a person at all.

Lex Fridman (03:58.240)

But see, the thing is, us humans and dogs, too, don't just use eyes as attentional mechanisms.

Lex Fridman (04:06.160)

They also use it to communicate, as part of the communication.

Rodney Brooks (04:09.440)

Like a dog can look at you, look at another thing, and look back at you,

Lex Fridman (04:12.880)

and that designates that we're going to be looking at that thing together.

Rodney Brooks (04:15.840)

Yeah, or intent, you know, on both Baxter and Sawyer at Rethink Robotics,

Lex Fridman (04:21.200)

they had a screen with, you know, graphic eyes,

Lex Fridman (04:25.440)

so it wasn't actually where the cameras were pointing, but the eyes would look in the direction

Rodney Brooks (04:31.200)

it was about to move its arm, so people in the factory nearby were not surprised by its motions,

Rodney Brooks (04:36.160)

because it gave that intent away.

Rodney Brooks (04:39.840)

Before we talk about Baxter, which I think is a beautiful robot, let's go back to the beginning.

Lex Fridman (04:45.120)

When did you first fall in love with robotics?

Lex Fridman (04:48.560)

We're talking about beauty and love to open the conversation.

Rodney Brooks (04:50.880)

This is great.

Lex Fridman (04:51.440)

I was born in the end of 1954, and I grew up in Adelaide, South Australia,

Lex Fridman (04:57.520)

and I have these two books that are dated 1961, so I'm guessing my mother found them in a store

Lex Fridman (05:05.120)

in 62 or 63, How and Why Wonder Books.

Lex Fridman (05:09.600)

How and Why Wonder Book of Electricity, and a How and Why Wonder Book of Giant Brains and Robots.

Lex Fridman (05:15.680)

And I learned how to build circuits, you know, when I was eight or nine, simple circuits,

Lex Fridman (05:23.200)

and I read, you know, learned the binary system, and saw all these drawings, mostly, of robots,

Lex Fridman (05:31.680)

and then I tried to build them for the rest of my childhood.

Lex Fridman (05:36.080)

Wait, 61, you said?

Lex Fridman (05:38.400)

This was when the two books, I've still got them at home.

Lex Fridman (05:41.200)

What does the robot mean in that context?

Rodney Brooks (05:43.520)

Some of the robots that they had were arms, you know, big arms to move nuclear material around,

Lex Fridman (05:51.600)

but they had pictures of welding robots that looked like humans under the sea, welding stuff

Lex Fridman (05:57.600)

underwater.

Lex Fridman (05:59.040)

So they weren't real robots, but they were, you know, what people were thinking about for robots.

Lex Fridman (06:05.200)

What were you thinking about?

Lex Fridman (06:06.560)

Were you thinking about humanoids?

Lex Fridman (06:07.920)

Were you thinking about arms with fingers?

Lex Fridman (06:09.760)

Were you thinking about faces or colors?

Lex Fridman (06:12.080)

Were you thinking about faces or cars?

Rodney Brooks (06:14.000)

No, actually, to be honest, I realized my limitation on building mechanical stuff.

Lex Fridman (06:19.360)

So I just built the brains, mostly, out of different technologies as I got older.

Rodney Brooks (06:28.320)

I built a learning system which was chemical based, and I had this ice cube tray.

Rodney Brooks (06:35.040)

Each well was a cell, and by applying voltage to the two electrodes, it would build up a

Rodney Brooks (06:42.400)

copper bridge.

Lex Fridman (06:43.040)

So over time, it would learn a simple network so I could teach it stuff.

Lex Fridman (06:50.000)

And mostly, things were driven by my budget, and nails as electrodes and an ice cube tray

Lex Fridman (07:00.160)

was about my budget at that stage.

Rodney Brooks (07:02.160)

Later, I managed to buy transistors, and I could build gates and flip flops and stuff.

Lex Fridman (07:07.520)

So one of your first robots was an ice cube tray?

Rodney Brooks (07:11.040)

Yeah, it was very cerebral because it learned to add.

Lex Fridman (07:16.720)

Very nice.

Rodney Brooks (07:17.920)

Well, just a decade or so before, in 1950, Alan Turing wrote a paper that formulated

Lex Fridman (07:26.080)

the Turing Test, and he opened that paper with the question, can machines think?

Lex Fridman (07:32.400)

So let me ask you this question.

Lex Fridman (07:34.160)

Can machines think?

Lex Fridman (07:36.160)

Can your ice cube tray one day think?

Rodney Brooks (07:40.800)

Certainly, machines can think because I believe you're a machine, and I'm a machine, and I

Rodney Brooks (07:44.720)

believe we both think.

Lex Fridman (07:46.640)

I think any other philosophical position is sort of a little ludicrous.

Lex Fridman (07:51.360)

What does think mean if it's not something that we do?

Lex Fridman (07:53.760)

And we are machines.

Lex Fridman (07:56.160)

So yes, machines can, but do we have a clue how to build such machines?

Lex Fridman (08:00.880)

That's a very different question.

Lex Fridman (08:02.480)

Are we capable of building such machines?

Lex Fridman (08:05.680)

Are we smart enough?

Rodney Brooks (08:06.720)

We think we're smart enough to do anything, but maybe we're not.

Lex Fridman (08:10.000)

Maybe we're just not smart enough to build stuff like us.

Rodney Brooks (08:14.160)

The kind of computer that Alan Turing was thinking about, do you think there is something

Rodney Brooks (08:18.720)

fundamentally or significantly different between the computer between our ears, the biological

Rodney Brooks (08:25.040)

computer that humans use, and the computer that he was thinking about from a sort of

Lex Fridman (08:31.200)

high level philosophical?

Rodney Brooks (08:33.280)

Yeah, I believe that it's very wrong.

Rodney Brooks (08:36.480)

In fact, I'm halfway through a, I think it'll be about a 480 page book, the working title

Rodney Brooks (08:44.160)

is Not Even Wrong.

Lex Fridman (08:45.440)

And if I may, I'll tell you a bit about that book.

Rodney Brooks (08:48.080)

Yes, please.

Lex Fridman (08:48.720)

So there's two, well, three thrusts to it.

Rodney Brooks (08:52.720)

One is the history of computation, what we call computation.

Rodney Brooks (08:56.160)

It goes all the way back to some manuscripts in Latin from 1614 and 1620 by Napier and

Rodney Brooks (09:03.760)

Kepler through Babbage and Lovelace.

Lex Fridman (09:06.640)

And then Turing's 1936 paper is what we think of as the invention of modern computation.

Lex Fridman (09:17.360)

And that paper, by the way, did not set out to invent computation.

Lex Fridman (09:23.120)

It set out to negatively answer one of Hilbert's three later set of problems.

Rodney Brooks (09:29.680)

He called it an effective way of getting answers.

Lex Fridman (09:38.560)

And Hilbert really worked with rewriting rules, as did Church, who also, at the same time,

Rodney Brooks (09:49.360)

a month earlier than Turing, disproved Hilbert's one of these three hypotheses.

Lex Fridman (09:54.880)

The other two had already been disproved by Gödel.

Rodney Brooks (09:57.360)

Turing set out to disprove it, because it's always easier to disprove these things than

Lex Fridman (10:01.680)

to prove that there is an answer.

Lex Fridman (10:04.160)

And so he needed, and it really came from his professor while I was an undergrad at

Lex Fridman (10:12.880)

Cambridge, who turned it into, is there a mechanical process?

Lex Fridman (10:16.400)

So he wanted to show a mechanical process that could calculate numbers, because that

Lex Fridman (10:23.840)

was a mechanical process that people used to generate tables.

Rodney Brooks (10:27.760)

They were called computers, the people at the time.

Lex Fridman (10:30.800)

And they followed a set of rules where they had paper, and they would write numbers down,

Lex Fridman (10:35.360)

and based on the numbers, they'd keep writing other numbers.

Lex Fridman (10:39.040)

And they would produce numbers for these tables, engineering tables, that the more iterations

Rodney Brooks (10:46.800)

they did, the more significant digits came out.

Lex Fridman (10:48.960)

And so Turing, in that paper, set out to define what sort of machine could do that, mechanical

Rodney Brooks (10:56.960)

machine, where it could produce an arbitrary number of digits in the same way a human computer

Lex Fridman (11:04.320)

did.

Lex Fridman (11:06.720)

And he came up with a very simple set of constraints where there was an infinite supply

Lex Fridman (11:13.600)

of paper.

Rodney Brooks (11:14.320)

This is the tape of the Turing machine, and each Turing machine came with a set of instructions

Rodney Brooks (11:22.320)

that, as a person, could do with pencil and paper, write down things on the tape and erase

Rodney Brooks (11:27.920)

them and put new things there.

Lex Fridman (11:30.000)

And he was able to show that that system was not able to do something that Hilbert had

Rodney Brooks (11:36.560)

hypothesized, so he disproved it.

Lex Fridman (11:38.800)

But he had to show that this system was good enough to do whatever could be done, but couldn't

Rodney Brooks (11:47.120)

do this other thing.

Lex Fridman (11:48.400)

And there he said, and he says in the paper, I don't have any real arguments for this,

Lex Fridman (11:53.840)

but based on intuition.

Lex Fridman (11:55.840)

So that's how he defined computation.

Lex Fridman (11:58.080)

And then if you look over the next, from 1936 up until really around 1975, you see people

Lex Fridman (12:05.440)

struggling with, is this really what computation is?

Lex Fridman (12:10.000)

And so Marvin Minsky, very well known in AI, but also a fantastic mathematician, in his

Rodney Brooks (12:17.200)

book Finite and Infant Machines from the mid-'60s, which is a beautiful, beautiful mathematical

Lex Fridman (12:22.400)

book, says at the start of the book, well, what is computation?

Lex Fridman (12:26.720)

Turing says it's this, and yeah, I sort of think it's that.

Rodney Brooks (12:29.520)

It doesn't really matter whether the stuff's made of wood or plastic.

Lex Fridman (12:32.240)

It's just that relatively cheap stuff can do this stuff.

Lex Fridman (12:36.320)

And so yeah, seems like computation.

Lex Fridman (12:40.160)

And Donald Knuth, in his first volume of his Art of Computer Programming in around 1968,

Lex Fridman (12:49.440)

says, well, what's computation?

Rodney Brooks (12:52.320)

It's this stuff, like Turing says, that a person could do each step without too much

Rodney Brooks (12:57.200)

trouble.

Lex Fridman (12:57.600)

And so one of his examples of what would be too much trouble was a step which required

Rodney Brooks (13:03.600)

knowing whether Fermat's Last Theorem was true or not, because it was not known at the

Lex Fridman (13:08.160)

time.

Lex Fridman (13:08.800)

And that's too much trouble for a person to do as a step.

Lex Fridman (13:12.160)

And Hopcroft and Ullman sort of said a similar thing later that year.

Lex Fridman (13:18.080)

And by 1975, in the A.H.O.

Rodney Brooks (13:20.960)

Hopcroft and Ullman book, they're saying, well, you know, we don't really know what

Rodney Brooks (13:24.880)

computation is, but intuition says this is sort of about right, and this is what it is.

Lex Fridman (13:31.280)

That's computation.

Rodney Brooks (13:32.400)

It's a sort of agreed upon thing which happens to be really easy to implement in silicon.

Lex Fridman (13:39.280)

And then we had Moore's Law, which took off, and it's been an incredibly powerful tool.

Rodney Brooks (13:44.640)

I certainly wouldn't argue with that.

Lex Fridman (13:46.080)

The version we have of computation, incredibly powerful.

Lex Fridman (13:49.440)

Can we just take a pause?

Lex Fridman (13:51.440)

So what we're talking about is there's an infinite tape with some simple rules of how

Rodney Brooks (13:55.440)

to write on that tape, and that's what we're kind of thinking about.

Lex Fridman (13:59.120)

This is computation.

Rodney Brooks (14:00.080)

Yeah, and it's modeled after humans, how humans do stuff.

Lex Fridman (14:03.200)

And I think it's, Turing says in the 36th paper, one of the critical facts here is that

Rodney Brooks (14:09.040)

a human has a limited amount of memory.

Lex Fridman (14:11.920)

So that's what we're going to put onto our mechanical computers.

Rodney Brooks (14:15.280)

So, you know, I'm like mass.

Lex Fridman (14:19.680)

I'm like mass or charge or, you know, it's not given by the universe.

Rodney Brooks (14:26.240)

It was, this is what we're going to call computation.

Lex Fridman (14:29.200)

And then it has this really, you know, it had this really good implementation, which

Rodney Brooks (14:33.600)

has completely changed our technological world.

Lex Fridman (14:36.800)

That's computation.

Rodney Brooks (14:40.400)

Second part of the book, or argument in the book, I have this two by two matrix with science.

Rodney Brooks (14:48.880)

In the top row, engineering in the bottom row, left column is intelligence, right column

Rodney Brooks (14:56.080)

is life.

Lex Fridman (14:58.000)

So in the bottom row, the engineering, there's artificial intelligence and artificial life.

Rodney Brooks (15:03.440)

In the top row, there's neuroscience and abiogenesis.

Lex Fridman (15:07.520)

How does living matter turn in?

Lex Fridman (15:09.920)

How does nonliving matter become living matter?

Lex Fridman (15:12.720)

Four disciplines.

Rodney Brooks (15:14.000)

These four disciplines all came into the current form in the period 1945 to 1965.

Lex Fridman (15:24.000)

That's interesting.

Rodney Brooks (15:24.880)

There was neuroscience before, but it wasn't effective neuroscience.

Rodney Brooks (15:28.480)

It was, you know, there were these ganglia and there's electrical charges, but no one

Rodney Brooks (15:32.160)

knows what to do with it.

Lex Fridman (15:33.680)

And furthermore, there are a lot of players who are common across them.

Rodney Brooks (15:38.000)

I've identified common players except for artificial intelligence and abiogenesis.

Lex Fridman (15:43.360)

I don't have, but for any other pair, I can point to people who work them.

Lex Fridman (15:47.200)

And a whole bunch of them, by the way, were at the research lab for electronics at MIT

Lex Fridman (15:53.200)

where Warren McCulloch held forth.

Rodney Brooks (15:58.240)

In fact, McCulloch, Pitts, Letvin, and Maturana wrote the first paper on functional neuroscience

Rodney Brooks (16:06.400)

called What the Frog's Eye Tells the Frog's Brain, where instead of it just being this

Rodney Brooks (16:10.480)

bunch of nerves, they sort of showed what different anatomical components were doing

Lex Fridman (16:17.680)

and telling other anatomical components and, you know, generating behavior in the frog.

Rodney Brooks (16:23.920)

Would you put them as basically the fathers or one of the early pioneers of what are now

Lex Fridman (16:29.840)

called artificial neural networks?

Rodney Brooks (16:33.120)

Yeah, I mean, McCulloch and Pitts.

Lex Fridman (16:36.560)

Pitts was a much younger than him.

Rodney Brooks (16:38.880)

In 1943, had written a paper inspired by Bertrand Russell on a calculus for the ideas eminent

Rodney Brooks (16:48.240)

in neural systems where they had tried to, without any real proof, they had tried to

Rodney Brooks (16:56.080)

give a formalism for neurons basically in terms of logic and gates or gates and not

Rodney Brooks (17:03.280)

gates with no real evidence that that was what was going on, but they talked about it

Lex Fridman (17:09.120)

and that was picked up by Minsky for his 1953 dissertation on, which was a neural

Lex Fridman (17:16.160)

network, we call it today.

Rodney Brooks (17:18.400)

It was picked up by John von Neumann when he was designing the Edbeck computer in 1945.

Rodney Brooks (17:26.640)

He talked about its components being neurons based on, and in references, he's only got

Rodney Brooks (17:31.680)

three references and one of them is the McCulloch Pitts paper.

Lex Fridman (17:35.600)

So all these people and then the AI people and the artificial life people, which was

Rodney Brooks (17:40.000)

John von Neumann originally, there's like overlap between all, they're all going around

Lex Fridman (17:44.560)

the same time.

Lex Fridman (17:45.440)

And three of these four disciplines turned to computation as their primary metaphor.

Lex Fridman (17:51.760)

So I've got a couple of chapters in the book.

Lex Fridman (17:54.480)

One is titled, wait, computers are people?

Lex Fridman (17:58.480)

Because that's where our computers came from.

Rodney Brooks (18:00.800)

Yeah.

Lex Fridman (18:01.920)

And, you know, from people who were computing stuff.

Lex Fridman (18:05.280)

And then I've got another chapter, wait, people are computers?

Lex Fridman (18:08.960)

Which is about computational neuroscience.

Rodney Brooks (18:10.880)

Yeah.

Lex Fridman (18:11.360)

So there's this whole circle here.

Lex Fridman (18:14.160)

And that computation is it.

Rodney Brooks (18:16.560)

And, you know, I have talked to people about, well, maybe it's not computation that goes

Rodney Brooks (18:21.760)

on in the head.

Lex Fridman (18:22.960)

Of course it is.

Rodney Brooks (18:24.160)

Yeah.

Lex Fridman (18:24.480)

Okay, well, when Elon Musk's rocket goes up, is it computing?

Lex Fridman (18:31.520)

Is that how it gets into orbit?

Lex Fridman (18:32.800)

By computing?

Lex Fridman (18:34.080)

But we've got this idea, if you want to build an AI system, you write a computer program.

Rodney Brooks (18:39.840)

Yeah, so the word computation very quickly starts doing a lot of work that it was not

Rodney Brooks (18:46.480)

initially intended to do.

Rodney Brooks (18:48.640)

It's the second and same if you talk about the universe as essentially performing a

Rodney Brooks (18:53.280)

computation.

Lex Fridman (18:53.760)

Yeah, right.

Rodney Brooks (18:54.320)

Wolfram does this.

Lex Fridman (18:55.280)

He turns it into computation.

Rodney Brooks (18:57.200)

You don't turn rockets into computation.

Lex Fridman (18:59.360)

Yeah.

Rodney Brooks (18:59.840)

By the way, when you say computation in our conversation, do you tend to think of computation

Lex Fridman (19:04.640)

narrowly in the way Turing thought of computation?

Rodney Brooks (19:08.000)

It's gotten very, you know, squishy.

Lex Fridman (19:14.080)

Yeah.

Rodney Brooks (19:14.400)

Squishy.

Lex Fridman (19:17.680)

But computation in the way Turing thinks about it and the way most people think about it

Rodney Brooks (19:22.640)

actually fits very well with thinking like a hunter gatherer.

Rodney Brooks (19:29.440)

There are places and there can be stuff in places and the stuff in places can change

Lex Fridman (19:34.000)

and it stays there until someone changes it.

Lex Fridman (19:37.120)

And it's this metaphor of place and container, which, you know, is a combination of our place

Rodney Brooks (19:44.880)

cells in our hippocampus and our cortex.

Lex Fridman (19:48.160)

But this is how we use metaphors for mostly to think about.

Lex Fridman (19:52.240)

And when we get outside of our metaphor range, we have to invent tools which we can sort

Lex Fridman (19:57.120)

of switch on to use.

Lex Fridman (19:58.960)

So calculus is an example of a tool.

Rodney Brooks (1:00:06.100)

interacting with the Mobileye implementation of Tesla Autopilot, I've driven a lot of car,

Rodney Brooks (1:00:12.740)

you know, I've been in Google self driving car since the beginning.

Rodney Brooks (1:00:18.020)

I thought there was no way before I sat and used Mobileye, I thought they're just knowing

Rodney Brooks (1:00:23.300)

computer vision.

Lex Fridman (1:00:24.100)

I thought there's no way it could work as well as it was working.

Lex Fridman (1:00:26.980)

So my model of the limits of computer vision was way more limited than the actual implementation

Lex Fridman (1:00:35.300)

of Mobileye.

Rodney Brooks (1:00:35.940)

I was so that's one example.

Lex Fridman (1:00:37.860)

I was really surprised.

Rodney Brooks (1:00:39.380)

It's like, wow, that was that was incredible.

Rodney Brooks (1:00:41.700)

The second surprise came when Tesla threw away Mobileye and started from scratch.

Rodney Brooks (1:00:50.580)

I thought there's no way they can catch up to Mobileye.

Rodney Brooks (1:00:52.740)

I thought what Mobileye was doing was kind of incredible, like the amount of work and

Rodney Brooks (1:00:56.260)

the annotation.

Rodney Brooks (1:00:56.980)

Yeah, well, Mobileye was started by Amnon Shashua and used a lot of traditional, you

Rodney Brooks (1:01:01.620)

know, hard fought computer vision techniques.

Lex Fridman (1:01:04.420)

But they also did a lot of good sort of like non research stuff, like actual like just

Lex Fridman (1:01:11.620)

good, like what you do to make a successful product, right?

Lex Fridman (1:01:14.420)

Scale, all that kind of stuff.

Lex Fridman (1:01:16.020)

And so I was very surprised when they from scratch were able to catch up to that.

Lex Fridman (1:01:20.660)

That's very impressive.

Lex Fridman (1:01:21.620)

And I've talked to a lot of engineers that was involved.

Lex Fridman (1:01:23.780)

This is that was impressive.

Rodney Brooks (1:01:25.620)

That was impressive.

Lex Fridman (1:01:27.300)

And the recent progress, especially under the involvement of Andrej Karpathy, what they

Rodney Brooks (1:01:34.900)

were what they're doing with the data engine, which is converting into the driving task

Rodney Brooks (1:01:40.340)

into these multiple tasks and then doing this edge case discovery when they're pulling back

Rodney Brooks (1:01:45.140)

like the level of engineering made me rethink what's possible.

Rodney Brooks (1:01:49.940)

I don't I still, you know, I don't know to that intensity, but I always thought it was

Rodney Brooks (1:01:55.380)

very difficult to solve autonomous driving with all the sensors, with all the computation.

Lex Fridman (1:02:00.260)

I just thought it's a very difficult problem.

Lex Fridman (1:02:02.420)

But I've been continuously surprised how much you can engineer.

Rodney Brooks (1:02:07.860)

First of all, the data acquisition problem, because I thought, you know, just because

Rodney Brooks (1:02:12.100)

I worked with a lot of car companies and they're they're so a little a little bit old school

Rodney Brooks (1:02:20.180)

to where I didn't think they could do this at scale like AWS style data collection.

Lex Fridman (1:02:25.940)

So when Tesla was able to do that, I started to think, OK, so what are the limits of this?

Rodney Brooks (1:02:33.140)

I still believe that driver like sensing and the interaction with the driver and like studying

Rodney Brooks (1:02:40.980)

the human factor psychology problem is essential.

Lex Fridman (1:02:43.700)

It's it's always going to be there.

Rodney Brooks (1:02:45.460)

It's always going to be there, even with fully autonomous driving.

Lex Fridman (1:02:48.740)

But I've been surprised what is the limit, especially a vision based alone, how far that

Rodney Brooks (1:02:55.220)

can take us.

Lex Fridman (1:02:57.060)

So that's my levels of surprise now.

Rodney Brooks (1:03:00.900)

OK, can you explain in the same way you said, like Alpha Zero, that's a homework problem

Lex Fridman (1:03:07.380)

that's scaled large in its chest, like who cares?

Rodney Brooks (1:03:10.260)

Go with here's actual people using an actual car and driving.

Lex Fridman (1:03:15.380)

Many of them drive more than half their miles using the system.

Rodney Brooks (1:03:19.380)

Right.

Lex Fridman (1:03:20.420)

So, yeah, they're doing well with with pure vision for your vision.

Rodney Brooks (1:03:24.980)

Yeah.

Lex Fridman (1:03:25.480)

And, you know, and now no radar, which is I suspect that can't go all the way.

Lex Fridman (1:03:30.820)

And one reason is without without new cameras that have a dynamic range closer to the human

Lex Fridman (1:03:36.340)

eye, because human eye has incredible dynamic range.

Lex Fridman (1:03:39.300)

And we make use of that dynamic range in its 11 orders of magnitude or some crazy number

Lex Fridman (1:03:46.500)

like that.

Rodney Brooks (1:03:47.700)

The cameras don't have that, which is why you see the the the bad cases where the sun

Lex Fridman (1:03:53.140)

on a white thing and it blinds it in a way it wouldn't blind the person.

Rodney Brooks (1:03:59.860)

I think there's a bunch of things to think about before you say this is so good, it's

Lex Fridman (1:04:06.020)

just going to work.

Rodney Brooks (1:04:06.660)

OK, and I'll come at it from multiple angles.

Lex Fridman (1:04:12.180)

And I know you've got a lot of time.

Rodney Brooks (1:04:13.700)

Yeah.

Lex Fridman (1:04:14.420)

OK, let's let's I have thought about these things.

Rodney Brooks (1:04:17.220)

Yeah, I know.

Rodney Brooks (1:04:18.740)

You've been writing a lot of great blog posts about it for a while before Tesla had autopilot.

Rodney Brooks (1:04:24.980)

Right.

Lex Fridman (1:04:25.480)

So you've been thinking about autonomous driving for a while from every angle.

Lex Fridman (1:04:29.220)

So so a few things, you know, in the US, I think that the death rate for autonomous driving

Lex Fridman (1:04:36.020)

death rate from motor vehicle accidents is about thirty five thousand a year,

Rodney Brooks (1:04:44.900)

which is an outrageous number, not outrageous compared to covid deaths.

Lex Fridman (1:04:49.140)

But, you know, there is no rationality.

Lex Fridman (1:04:52.100)

And that's part of the thing people have said.

Rodney Brooks (1:04:54.340)

Engineers say to me, well, if we cut down the number of deaths by 10 percent by having

Rodney Brooks (1:04:58.900)

autonomous driving, that's going to be great.

Lex Fridman (1:05:01.300)

Everyone will love it.

Lex Fridman (1:05:02.100)

And my prediction is that if autonomous vehicles kill more than 10 people a year, they'll be

Rodney Brooks (1:05:09.620)

screaming and hollering, even though thirty five thousand people a year have been killed

Rodney Brooks (1:05:14.260)

by human drivers.

Lex Fridman (1:05:16.260)

It's not rational.

Rodney Brooks (1:05:17.860)

It's a different set of expectations.

Lex Fridman (1:05:20.100)

And that will probably continue.

Lex Fridman (1:05:23.860)

So there's that aspect of it.

Rodney Brooks (1:05:25.300)

The other aspect of it is that when we introduce new technology, we often change the rules

Rodney Brooks (1:05:34.420)

of the game.

Lex Fridman (1:05:36.020)

So when we introduced cars first into our daily lives, we completely rebuilt our cities

Lex Fridman (1:05:45.060)

and we changed all the laws.

Rodney Brooks (1:05:46.900)

Yeah, jaywalking was not an offense that was pushed by the car companies so that people

Rodney Brooks (1:05:52.820)

would stay off the road so there wouldn't be deaths from pedestrians getting hit.

Rodney Brooks (1:05:57.460)

We completely changed the structure of our cities and had these foul smelling things

Rodney Brooks (1:06:02.580)

everywhere around us.

Lex Fridman (1:06:04.580)

And now you see pushback in cities like Barcelona is really trying to exclude cars, et cetera.

Lex Fridman (1:06:11.060)

So I think that to get to self driving, we will, large adoption, it's not going to be

Rodney Brooks (1:06:21.460)

just take the current situation, take out the driver and put the same car doing the

Rodney Brooks (1:06:27.300)

same stuff because the end case is too many.

Lex Fridman (1:06:31.860)

Here's an interesting question.

Lex Fridman (1:06:33.300)

How many fully autonomous train systems do we have in the U.S.?

Lex Fridman (1:06:41.860)

I mean, do you count them as fully autonomous?

Lex Fridman (1:06:43.860)

I don't know because they're usually as a driver, but they're kind of autonomous, right?

Lex Fridman (1:06:47.860)

No, let's get rid of the driver.

Rodney Brooks (1:06:51.380)

Okay.

Lex Fridman (1:06:51.860)

I don't know.

Rodney Brooks (1:06:52.820)

It's either 15 or 16.

Lex Fridman (1:06:54.660)

Most of them are in airports.

Rodney Brooks (1:06:56.900)

There's a few that are fully autonomous.

Rodney Brooks (1:06:59.860)

Seven are in airports, there's a few that go about five, two that go about five kilometers

Rodney Brooks (1:07:06.260)

out of airports.

Rodney Brooks (1:07:11.460)

When is the first fully autonomous train system for mass transit expected to operate fully

Rodney Brooks (1:07:17.460)

autonomously with no driver in a U.S.

Lex Fridman (1:07:22.420)

City?

Rodney Brooks (1:07:23.540)

It's expected to operate in 2017 in Honolulu.

Lex Fridman (1:07:27.940)

Oh, wow.

Rodney Brooks (1:07:29.300)

It's delayed, but they will get there.

Lex Fridman (1:07:32.020)

BART, by the way, was originally going to be autonomous here in the Bay Area.

Lex Fridman (1:07:35.780)

I mean, they're all very close to fully autonomous, right?

Lex Fridman (1:07:38.820)

Yeah, but getting that close is the thing.

Lex Fridman (1:07:41.540)

And I've often gone on a fully autonomous train in Japan, one that goes out to that

Lex Fridman (1:07:48.660)

fake island in the middle of Tokyo Bay.

Rodney Brooks (1:07:50.660)

I forget the name of that.

Lex Fridman (1:07:53.460)

And what do you see when you look at that?

Lex Fridman (1:07:55.540)

What do you see when you go to a fully autonomous train in an airport?

Lex Fridman (1:08:03.380)

It's not like regular trains.

Rodney Brooks (1:08:07.060)

At every station, there's a double set of doors so that there's a door of the train

Lex Fridman (1:08:12.100)

and there's a door off the platform.

Lex Fridman (1:08:18.020)

And this is really visible in this Japanese one because it goes out in amongst buildings.

Lex Fridman (1:08:23.540)

The whole track is built so that people can't climb onto it.

Rodney Brooks (1:08:27.060)

Yeah.

Lex Fridman (1:08:27.860)

So there's an engineering that then makes the system safe and makes them acceptable.

Rodney Brooks (1:08:32.260)

I think we'll see similar sorts of things happen in the U.S.

Lex Fridman (1:08:37.620)

What surprised me, I thought, wrongly, that we would have special purpose lanes on 101

Rodney Brooks (1:08:46.180)

in the Bay Area, the leftmost lane, so that it would be normal for Teslas or other cars

Rodney Brooks (1:08:55.140)

to move into that lane and then say, okay, now it's autonomous and have that dedicated lane.

Rodney Brooks (1:09:00.900)

I was expecting movement to that.

Lex Fridman (1:09:03.380)

Five years ago, I was expecting we'd have a lot more movement towards that.

Rodney Brooks (1:09:06.500)

We haven't.

Lex Fridman (1:09:07.460)

And it may be because Tesla's been overpromising by saying this, calling their system fully

Rodney Brooks (1:09:12.820)

self driving, I think they may have been gotten there quicker by collaborating to change the

Lex Fridman (1:09:21.780)

infrastructure.

Rodney Brooks (1:09:23.460)

This is one of the problems with long haul trucking being autonomous.

Rodney Brooks (1:09:30.180)

I think it makes sense on freeways at night for the trucks to go autonomously, but then

Lex Fridman (1:09:38.020)

is that how do you get onto and off of the freeway?

Lex Fridman (1:09:40.260)

What sort of infrastructure do you need for that?

Lex Fridman (1:09:43.780)

Do you need to have the human in there to do that or can you get rid of the human?

Lex Fridman (1:09:48.500)

So I think there's ways to get there, but it's an infrastructure argument because the

Rodney Brooks (1:09:55.060)

long tail of cases is very long and the acceptance of it will not be at the same level as human

Lex Fridman (1:10:02.020)

drivers.

Lex Fridman (1:10:02.580)

So I'm with you still, and I was with you for a long time, but I am surprised how well

Lex Fridman (1:10:09.780)

how many edge cases of machine learning and vision based methods can cover.

Rodney Brooks (1:10:15.540)

This is what I'm trying to get at is I think there's something fundamentally different

Rodney Brooks (1:10:22.260)

with vision based methods and Tesla Autopilot and any company that's trying to do the same.

Rodney Brooks (1:10:27.460)

Okay, well, I'm not going to argue with you because, you know, we're speculating.

Rodney Brooks (1:10:34.260)

Yes, but, you know, my gut feeling tells me it's going to be things will speed up when

Rodney Brooks (1:10:43.620)

there is engineering of the environment because that's what happened with every other technology.

Rodney Brooks (1:10:48.260)

I'm a bit, I don't know about you, but I'm a bit cynical that infrastructure is going

Rodney Brooks (1:10:53.940)

to rely on government to help out in these cases.

Rodney Brooks (1:11:00.340)

If you just look at infrastructure in all domains, it's just a government always drags

Rodney Brooks (1:11:05.540)

behind on infrastructure.

Lex Fridman (1:11:07.540)

There's like there's so many just well in this country in the future.

Rodney Brooks (1:11:11.780)

Sorry.

Lex Fridman (1:11:12.260)

Yes, in this country.

Lex Fridman (1:11:13.700)

And of course, there's many, many countries that are actually much worse on infrastructure.

Lex Fridman (1:11:17.780)

Oh, yes, many of the much worse and there's some that are much worse.

Rodney Brooks (1:11:21.220)

You know, like high speed rail, the other countries are much better.

Rodney Brooks (1:11:25.940)

I guess my question is, like, which is at the core of what I was trying to think through

Lex Fridman (1:11:31.220)

here and ask is like, how hard is the driving problem as it currently stands?

Lex Fridman (1:11:37.540)

So you mentioned, like, we don't want to just take the human out and duplicate whatever

Rodney Brooks (1:11:41.220)

the human was doing.

Lex Fridman (1:11:42.260)

But if we were to try to do that, what, how hard is that problem?

Rodney Brooks (1:11:48.340)

Because I used to think is way harder.

Rodney Brooks (1:11:52.420)

Like, I used to think it's with vision alone, it would be three decades, four decades.

Rodney Brooks (1:11:59.220)

Okay, so I don't know the answer to this thing I'm about to pose, but I do notice that on

Rodney Brooks (1:12:06.740)

Highway 280 here in the Bay Area, which largely has concrete surface rather than blacktop

Rodney Brooks (1:12:13.380)

surface, the white lines that are painted there now have black boundaries around them.

Lex Fridman (1:12:20.900)

And my lane drift system in my car would not work without those black boundaries.

Rodney Brooks (1:12:27.460)

Interesting.

Lex Fridman (1:12:28.260)

So I don't know whether they started doing it to help the lane drift, whether it is an

Rodney Brooks (1:12:32.420)

instance of infrastructure following the technology, but my car would not perform as well as the

Rodney Brooks (1:12:41.220)

lane, my car would not perform as well without that change in the way they paint the line.

Rodney Brooks (1:12:45.460)

Unfortunately, really good lane keeping is not as valuable.

Lex Fridman (1:12:50.340)

Like, it's orders of magnitude more valuable to have a fully autonomous system.

Rodney Brooks (1:12:54.900)

Like, yeah, but for me, lane keeping is really helpful because I'm more healthy at it.

Lex Fridman (1:13:00.900)

But you wouldn't pay 10 times.

Rodney Brooks (1:13:03.700)

Like, the problem is there's not financial, like, it doesn't make sense to revamp the

Lex Fridman (1:13:11.540)

infrastructure to make lane keeping easier.

Rodney Brooks (1:13:14.820)

It does make sense to revamp the infrastructure.

Rodney Brooks (1:13:17.300)

If you have a large fleet of autonomous vehicles, now you change what it means to own cars,

Rodney Brooks (1:13:22.260)

you change the nature of transportation.

Lex Fridman (1:13:24.980)

But for that, you need autonomous vehicles.

Rodney Brooks (1:13:29.620)

Let me ask you about Waymo then.

Lex Fridman (1:13:31.540)

I've gotten a bunch of chances to ride in a Waymo self driving car.

Lex Fridman (1:13:37.380)

And they're, I don't know if you'd call them self driving, but.

Rodney Brooks (1:13:40.980)

Well, I mean, I rode in one before they were called Waymo when I was still at X.

Lex Fridman (1:13:45.780)

So there's currently, there's a big leap, another surprising leap I didn't think would

Lex Fridman (1:13:50.740)

happen, which is they have no driver currently.

Rodney Brooks (1:13:53.780)

Yeah, in Chandler.

Lex Fridman (1:13:55.060)

In Chandler, Arizona.

Lex Fridman (1:13:56.100)

And I think they're thinking of doing that in Austin as well.

Lex Fridman (1:13:58.980)

But they're expanding.

Rodney Brooks (1:14:01.540)

Although, you know, and I do an annual checkup on this.

Lex Fridman (1:14:06.100)

So as of late last year, they were aiming for hundreds of rides a week, not thousands.

Lex Fridman (1:14:14.020)

And there is no one in the car, but there's certainly safety people in the loop.

Lex Fridman (1:14:22.660)

And it's not clear how many, you know, what the ratio of cars to safety people is.

Rodney Brooks (1:14:26.820)

It wasn't, obviously, they're not 100% transparent about this.

Lex Fridman (1:14:31.620)

None of them are 100% transparent.

Rodney Brooks (1:14:33.220)

They're very untransparent.

Lex Fridman (1:14:34.420)

But at least the way they're, I don't want to make definitively, but they're saying

Rodney Brooks (1:14:39.540)

there's no teleoperation.

Lex Fridman (1:14:42.580)

So like, they're, I mean, okay.

Lex Fridman (1:14:45.620)

And that sort of fits with YouTube videos I've seen of people being trapped in the car

Lex Fridman (1:14:52.820)

by a red cone on the street.

Lex Fridman (1:14:55.460)

And they do have rescue vehicles that come, and then a person gets in and drives it.

Lex Fridman (1:15:01.620)

Yeah.

Lex Fridman (1:15:02.580)

But isn't it incredible to you, it was to me, to get in a car with no driver and watch

Rodney Brooks (1:15:09.700)

the steering wheel turn, like for somebody who has been studying, at least certainly

Rodney Brooks (1:15:15.060)

the human side of autonomous vehicles for many years, and you've been doing it for way

Lex Fridman (1:15:18.980)

longer, like it was incredible to me that this was actually could happen.

Rodney Brooks (1:15:22.420)

I don't care if that scale is 100 cars.

Lex Fridman (1:15:24.100)

This is not a demo.

Rodney Brooks (1:15:25.860)

This is not, this is me as a regular human.

Lex Fridman (1:15:28.820)

The argument I have is that people make interpolations from that.

Rodney Brooks (1:15:33.060)

Interpolations.

Lex Fridman (1:15:33.940)

That, you know, it's here, it's done.

Rodney Brooks (1:15:37.060)

You know, it's just, you know, we've solved it.

Lex Fridman (1:15:39.380)

No, we haven't yet.

Lex Fridman (1:15:40.980)

And that's my argument.

Lex Fridman (1:15:42.500)

Okay.

Lex Fridman (1:15:42.900)

So I'd like to go to, you keep a list of predictions on your amazing blog post.

Lex Fridman (1:15:48.420)

It'd be fun to go through them.

Lex Fridman (1:15:49.700)

But before then, let me ask you about this.

Rodney Brooks (1:15:51.620)

You have a harshness to you sometimes in your criticisms of what is perceived as hype.

Lex Fridman (1:16:05.940)

And so like, because people extrapolate, like you said, and they kind of buy into the hype

Lex Fridman (1:16:10.980)

and then they kind of start to think that the technology is way better than it is.

Lex Fridman (1:16:18.900)

But let me ask you maybe a difficult question.

Lex Fridman (1:16:22.260)

Sure.

Lex Fridman (1:16:23.780)

Do you think if you look at history of progress, don't you think to achieve the quote impossible,

Lex Fridman (1:16:30.740)

you have to believe that it's possible?

Rodney Brooks (1:16:32.740)

Oh, absolutely.

Lex Fridman (1:16:34.260)

Yeah.

Rodney Brooks (1:16:34.820)

Look, his two great runs, great, unbelievable, 1903, first human power, human, you know,

Lex Fridman (1:16:46.980)

human, you know, heavier than their flight.

Rodney Brooks (1:16:49.300)

Yeah.

Lex Fridman (1:16:50.580)

1969, we land on the moon.

Rodney Brooks (1:16:52.740)

That's 66 years.

Rodney Brooks (1:16:53.940)

I'm 66 years old in my lifetime, that span of my lifetime, barely, you know, flying,

Rodney Brooks (1:17:00.260)

I don't know what it was, 50 feet, the length of the first flight or something to landing

Lex Fridman (1:17:05.380)

on the moon.

Rodney Brooks (1:17:06.340)

Unbelievable.

Lex Fridman (1:17:08.100)

Fantastic.

Lex Fridman (1:17:08.980)

But that requires, by the way, one of the Wright brothers, both of them, but one of

Lex Fridman (1:17:13.060)

them didn't believe it's even possible like a year before.

Rodney Brooks (1:17:16.180)

Right.

Lex Fridman (1:17:16.680)

So, like, not just possible soon, but like ever.

Rodney Brooks (1:17:20.420)

So, you know.

Lex Fridman (1:17:21.940)

How important is it to believe and be optimistic is what I guess.

Rodney Brooks (1:17:24.820)

Oh, yeah, it is important.

Rodney Brooks (1:17:26.100)

It's when it goes crazy, when I, you know, you said that, what was the word you used

Lex Fridman (1:17:32.100)

for my bad?

Lex Fridman (1:17:33.060)

Harshness.

Rodney Brooks (1:17:33.780)

Harshness.

Lex Fridman (1:17:34.580)

Yes.

Rodney Brooks (1:17:40.180)

I just get so frustrated.

Lex Fridman (1:17:41.940)

Yes.

Rodney Brooks (1:17:42.440)

When people make these leaps and tell me that I'm, that I don't understand, you know, yeah.

Lex Fridman (1:17:53.020)

There's just from iRobot, which I was co founder of.

Rodney Brooks (1:17:57.420)

Yeah.

Rodney Brooks (1:17:57.740)

I don't know the exact numbers now because I haven't, it's 10 years since I stepped

Rodney Brooks (1:18:00.860)

off the board, but I believe it's well over 30 million robots cleaning houses from that

Lex Fridman (1:18:06.220)

one company.

Lex Fridman (1:18:06.780)

And now there's lots of other companies.

Lex Fridman (1:18:08.140)

Yes.

Lex Fridman (1:18:08.140)

Was that a crazy idea that we had to believe in 2002 when we released it?

Lex Fridman (1:18:14.940)

Yeah, that was, we had, we had to, you know, believe that it could be done.

Rodney Brooks (1:18:20.540)

Let me ask you about this.

Lex Fridman (1:18:21.740)

So iRobot, one of the greatest robotics companies ever in terms of creating a robot that actually

Rodney Brooks (1:18:28.380)

works in the real world, probably the greatest robotics company ever.

Lex Fridman (1:18:31.900)

You were the co founder of it.

Rodney Brooks (1:18:33.660)

If, if the Rodney Brooks of today talked to the Rodney of back then, what would you tell

Lex Fridman (1:18:40.860)

him?

Rodney Brooks (1:18:41.340)

Cause I have a sense that would you pat him on the back and say, well, you're doing is

Lex Fridman (1:18:47.100)

going to fail, but go at it anyway.

Rodney Brooks (1:18:50.780)

That's what I'm referring to with the harshness.

Lex Fridman (1:18:54.060)

You've accomplished an incredible thing there.

Rodney Brooks (1:18:56.700)

One of the several things we'll talk about was, you know, you know, you know, you've

Lex Fridman (1:19:01.500)

done several things we'll talk about.

Rodney Brooks (1:19:03.820)

Well, like that's what I'm trying to get at that line.

Rodney Brooks (1:19:06.940)

No, it's, it's when my harshness is reserved for people who are not doing it, who claim

Rodney Brooks (1:19:14.140)

it's just, well, this shows that it's just going to happen.

Lex Fridman (1:19:16.860)

But here, here's the thing.

Rodney Brooks (1:19:18.300)

This shows.

Lex Fridman (1:19:19.020)

But you have that harshness for Elon too.

Lex Fridman (1:19:24.060)

And no, no, it's a different harshness.

Lex Fridman (1:19:26.380)

No, it's, it's a different argument with Elon.

Rodney Brooks (1:19:30.540)

I think SpaceX is an amazing company.

Rodney Brooks (1:19:34.780)

On the other hand, you know, I, in one of my blog posts, I said, what's easy and what's

Rodney Brooks (1:19:40.060)

hard.

Lex Fridman (1:19:40.460)

I said, yeah, space X vertical landing rockets.

Rodney Brooks (1:19:44.300)

It had been done before.

Lex Fridman (1:19:46.380)

Grid fins had been done since the sixties.

Rodney Brooks (1:19:48.700)

Every Soyuz has them.

Lex Fridman (1:19:52.780)

Reusable space DCX reuse those rockets that landed vertically.

Rodney Brooks (1:19:58.220)

There's a whole insurance industry in place for rocket launches.

Lex Fridman (1:20:02.780)

There are all sorts of infrastructure that was doable.

Rodney Brooks (1:20:07.980)

It took a great entrepreneur, a great personal expense.

Lex Fridman (1:20:11.980)

He almost drove himself, you know, bankrupt doing it, a great belief to do it.

Rodney Brooks (1:20:18.860)

Whereas Hyperloop, there's a whole bunch more stuff that's never been thought about and

Lex Fridman (1:20:25.740)

never been demonstrated.

Lex Fridman (1:20:28.380)

So my estimation is Hyperloop is a long, long, long, a lot further off.

Rodney Brooks (1:20:33.660)

But, and if I've got a criticism of, of, of Elon, it's that he doesn't make distinctions

Rodney Brooks (1:20:39.740)

between when the technology's coming along and ready.

Lex Fridman (1:20:44.780)

And then he'll go off and mouth off about other things, which then people go and compete

Rodney Brooks (1:20:50.140)

about and try and do.

Lex Fridman (1:20:51.100)

And so this is where I, I, I, I understand what you're saying.

Rodney Brooks (1:20:57.580)

I tend to draw a different distinction.

Rodney Brooks (1:21:00.060)

I, I have a similar kind of harshness towards people who are not telling the truth, who

Rodney Brooks (1:21:06.220)

are basically fabricating stuff to make money or to, well, he believes what he says.

Rodney Brooks (1:21:11.420)

I just think that's a very important difference because I think in order to fly, in order

Rodney Brooks (1:21:18.300)

to get to the moon, you have to believe even when most people tell you you're wrong and

Lex Fridman (1:21:24.060)

most likely you're wrong, but sometimes you're right.

Rodney Brooks (1:21:26.940)

I mean, that's the same thing I have with Tesla autopilot.

Lex Fridman (1:21:29.900)

I think that's an interesting one.

Rodney Brooks (1:21:31.900)

I was, especially when I was at MIT and just the entire human factors in the robotics community

Lex Fridman (1:21:38.780)

were very negative towards Elon.

Rodney Brooks (1:21:40.300)

It was very interesting for me to observe colleagues at MIT.

Lex Fridman (1:21:45.020)

I wasn't sure what to make of that.

Rodney Brooks (1:21:46.620)

That was very upsetting to me because I understood where that, where that's coming from.

Lex Fridman (1:21:51.900)

And I agreed with them and I kind of almost felt the same thing in the beginning until

Rodney Brooks (1:21:56.300)

I kind of opened my eyes and realized there's a lot of interesting ideas here that might

Lex Fridman (1:22:01.660)

be over hype.

Rodney Brooks (1:22:02.540)

You know, if you focus yourself on the idea that you shouldn't call a system full self

Rodney Brooks (1:22:09.740)

driving when it's obviously not autonomous, fully autonomous, you're going to miss the

Rodney Brooks (1:22:16.220)

magic.

Lex Fridman (1:22:16.860)

Oh, yeah, you are going to miss the magic.

Lex Fridman (1:22:18.940)

But at the same time, there are people who buy it, literally pay money for it and take

Lex Fridman (1:22:25.340)

those words as given.

Lex Fridman (1:22:27.180)

So it's, but I haven't.

Lex Fridman (1:22:30.300)

So that I take words as given is one thing.

Rodney Brooks (1:22:33.420)

I haven't actually seen people that use autopilot that believe that the behavior is really important,

Lex Fridman (1:22:39.500)

like the actual action.

Lex Fridman (1:22:40.700)

So like, this is to push back on the very thing that you're frustrated about, which

Rodney Brooks (1:22:45.740)

is like journalists and general people buying all the hype and going out in the same way.

Rodney Brooks (1:22:52.460)

I think there's a lot of hype about the negatives of this, too, that people are buying without

Lex Fridman (1:22:57.980)

using people use the way this is what this was.

Rodney Brooks (1:23:01.020)

This opened my eyes.

Rodney Brooks (1:23:02.060)

Actually, the way people use a product is very different than the way they talk about

Rodney Brooks (1:23:07.580)

it.

Lex Fridman (1:23:07.820)

This is true with robotics, with everything.

Rodney Brooks (1:23:09.500)

Everybody has dreams of how a particular product might be used or so on.

Lex Fridman (1:23:13.660)

And then when it meets reality, there's a lot of fear of robotics, for example, that

Rodney Brooks (1:23:17.980)

robots are somehow dangerous and all those kinds of things.

Lex Fridman (1:23:20.380)

But when you actually have robots in your life, whether it's in the factory or in the

Rodney Brooks (1:23:23.980)

home, making your life better, that's going to be that's way different.

Lex Fridman (1:23:28.300)

Your perceptions of it are going to be way different.

Lex Fridman (1:23:30.460)

And so my just tension was like, here's an innovator.

Lex Fridman (1:23:34.780)

Supercruise from Cadillac was super interesting, too.

Rodney Brooks (1:23:41.500)

That's a really interesting system.

Lex Fridman (1:23:43.020)

We should be excited by those innovations.

Lex Fridman (1:23:45.580)

OK, so can I tell you something that's really annoyed me recently?

Rodney Brooks (1:23:49.020)

It's really annoyed me that the press and friends of mine on Facebook are going, these

Lex Fridman (1:23:56.380)

billionaires and their space games, why are they doing that?

Lex Fridman (1:23:59.740)

And that really, really pisses me off.

Rodney Brooks (1:24:02.300)

I must say, I applaud that.

Lex Fridman (1:24:05.100)

I applaud it.

Rodney Brooks (1:24:06.780)

It's the taking and not necessarily the people who are doing the things, but, you know, that

Rodney Brooks (1:24:13.180)

I keep having to push back against unrealistic expectations when these things can become

Rodney Brooks (1:24:19.740)

real.

Rodney Brooks (1:24:20.300)

Yeah, I this was interesting on because there's been a particular focus for me is autonomous

Rodney Brooks (1:24:26.220)

driving, Elon's prediction of when certain milestones will be hit.

Rodney Brooks (1:24:30.140)

There's several things to be said there that I always I thought about, because whenever

Rodney Brooks (1:24:37.660)

you said them, it was obvious that's not going to me as a person that kind of not inside

Lex Fridman (1:24:44.860)

the system is obvious.

Rodney Brooks (1:24:46.940)

It's unlikely to hit those.

Lex Fridman (1:24:48.700)

There's two comments I want to make.

Rodney Brooks (1:24:50.700)

One, he legitimately believes it.

Lex Fridman (1:24:54.220)

And two, much more importantly, I think that having ambitious deadlines drives people to

Rodney Brooks (1:25:04.140)

do the best work of their life, even when the odds of those deadlines are very low.

Lex Fridman (1:25:09.420)

To a point, and I'm not talking about anyone here, I'm just saying.

Lex Fridman (1:25:12.780)

So there's a line there, right?

Lex Fridman (1:25:14.220)

You have to have a line because you overextend and it's demoralizing.

Rodney Brooks (1:25:20.140)

It's demoralizing, but I will say that there's an additional thing here that those words

Lex Fridman (1:25:28.860)

also drive the stock market.

Lex Fridman (1:25:34.140)

And we have because of the way that rich people in the past have manipulated the rubes through

Lex Fridman (1:25:42.060)

investment, we have developed laws about what you're allowed to say.

Lex Fridman (1:25:49.260)

And you know, there's an area here which is I tend to be maybe I'm naive, but I tend to

Rodney Brooks (1:25:58.380)

believe that like engineers, innovators, people like that, they're not they're my they don't

Rodney Brooks (1:26:06.620)

think like that, like manipulating the price of the stock price.

Lex Fridman (1:26:09.500)

But it's possible that I'm I'm certain it's possible that I'm wrong.

Rodney Brooks (1:26:13.980)

It's a very cynical view of the world because I think most people that run companies, especially

Lex Fridman (1:26:21.820)

original founders, they yeah, I'm not saying that's the intent.

Rodney Brooks (1:26:27.260)

I'm saying it's eventually it's kind of you you you you fall into that kind of behavior

Lex Fridman (1:26:33.340)

pattern.

Rodney Brooks (1:26:33.340)

I don't know.

Lex Fridman (1:26:33.900)

I tend to I wasn't saying I wasn't saying it's falling into that intent.

Rodney Brooks (1:26:37.980)

It's just you also have to protect investors in this environment.

Lex Fridman (1:26:43.740)

In this market.

Rodney Brooks (1:26:44.620)

Yeah.

Rodney Brooks (1:26:45.580)

OK, so you have first of all, you have an amazing blog that people should check out.

Lex Fridman (1:26:50.060)

But you also have this in that blog, a set of predictions.

Lex Fridman (1:26:54.780)

Such a cool idea.

Rodney Brooks (1:26:55.740)

I don't know how long ago you started, like three, four years ago.

Lex Fridman (1:26:58.220)

It was January 1st, 2018.

Rodney Brooks (1:27:01.820)

18.

Lex Fridman (1:27:02.940)

And I made these predictions and I said that every January 1st, I was going to check back

Rodney Brooks (1:27:07.740)

on how my predictions.

Lex Fridman (1:27:09.020)

That's such a great thought experiment.

Rodney Brooks (1:27:10.220)

For 32 years.

Lex Fridman (1:27:11.900)

Oh, you said 32 years.

Rodney Brooks (1:27:13.340)

I said 32 years because it's still that'll be January 1st, 2050.

Lex Fridman (1:27:16.940)

I'll be I will just turn ninety.

Rodney Brooks (1:27:21.660)

Five, you know, and so people know that your predictions, at least for now, are in the

Lex Fridman (1:27:31.180)

space of artificial intelligence.

Rodney Brooks (1:27:33.180)

Yeah, I didn't say I was going to make new predictions.

Rodney Brooks (1:27:34.860)

I was just going to measure this set of predictions that I made because I was sort of I was sort

Rodney Brooks (1:27:38.380)

of annoyed that everyone could make predictions.

Lex Fridman (1:27:40.620)

They didn't come true and everyone forgot.

Lex Fridman (1:27:42.460)

So I should hold myself to a high standard.

Lex Fridman (1:27:44.860)

Yeah, but also just putting years and like date ranges on things.

Rodney Brooks (1:27:48.700)

It's a good thought exercise.

Lex Fridman (1:27:50.140)

Yeah, like and like reasoning your thoughts out.

Lex Fridman (1:27:52.940)

And so the topics are artificial intelligence, autonomous vehicles and space.

Lex Fridman (1:27:58.300)

Yeah.

Rodney Brooks (1:28:00.940)

I was wondering if we could just go through some that stand out maybe from memory.

Lex Fridman (1:28:04.700)

I can just mention to you some.

Rodney Brooks (1:28:06.140)

Let's talk about self driving cars, like some predictions that you're particularly proud

Rodney Brooks (1:28:10.780)

of or are particularly interesting from flying cars to the other element here is like how

Rodney Brooks (1:28:20.220)

widespread the location where the deployment of the autonomous vehicles is.

Lex Fridman (1:28:25.900)

And there's also just a few fun ones.

Lex Fridman (1:28:27.580)

Is there something that jumps to mind that you remember from the predictions?

Rodney Brooks (1:28:31.980)

Well, I think I did put in there that there would be a dedicated self driving lane on

Rodney Brooks (1:28:37.500)

101 by some year, and I think I was over optimistic on that one.

Lex Fridman (1:28:42.380)

Yeah, actually.

Rodney Brooks (1:28:42.860)

Yeah, I actually do remember that.

Lex Fridman (1:28:44.140)

But you I think you were mentioning like difficulties at different cities.

Rodney Brooks (1:28:48.620)

Yeah.

Lex Fridman (1:28:50.460)

Cambridge, Massachusetts, I think was an example.

Rodney Brooks (1:28:52.460)

Yeah, like in Cambridge Port, you know, I lived in Cambridge Port for a number of years

Lex Fridman (1:28:56.860)

and you know, the roads are narrow and getting getting anywhere as a human driver is incredibly

Rodney Brooks (1:29:02.780)

frustrating when you start to put and people drive the wrong way on one way streets there.

Rodney Brooks (1:29:07.660)

It's just your prediction was driverless taxi services operating on all streets in

Rodney Brooks (1:29:14.860)

Cambridge Port, Massachusetts in 2035.

Lex Fridman (1:29:21.100)

Yeah.

Lex Fridman (1:29:21.740)

And that may have been too optimistic.

Lex Fridman (1:29:25.020)

You think so?

Rodney Brooks (1:29:26.060)

You know, I've gotten a little more pessimistic since I made these internally on some of these

Lex Fridman (1:29:31.020)

things.

Lex Fridman (1:29:31.500)

So what can you put a year to a major milestone of deployment of a taxi service in in a few

Lex Fridman (1:29:42.780)

major cities like something where you feel like autonomous vehicles are here.

Lex Fridman (1:29:47.500)

So let's let's take the grid streets of San Francisco north of market.

Lex Fridman (1:29:55.900)

Okay.

Rodney Brooks (1:29:56.540)

Okay.

Rodney Brooks (1:29:57.040)

Relatively benign environment, the streets are wide, the major problem is delivery trucks

Rodney Brooks (1:30:07.040)

stopping everywhere, which made things more complicated.

Rodney Brooks (1:30:12.880)

Taxi system there with somewhat designated pickup and drop offs, unlike with Uber and

Rodney Brooks (1:30:21.280)

Lyft, where you can sort of get to any place and the drivers will figure out how to get

Lex Fridman (1:30:28.160)

in there.

Rodney Brooks (1:30:30.720)

We're still a few years away.

Lex Fridman (1:30:32.080)

I, you know, I live in that area.

Lex Fridman (1:30:35.200)

So I see, you know, the self driving car companies cars, multiple multiple ones every day.

Rodney Brooks (1:30:42.240)

Now if they're cruise, Zooks less often, Waymo all the time, different and different ones

Rodney Brooks (1:30:52.480)

come and go.

Lex Fridman (1:30:53.440)

And there's always a driver.

Rodney Brooks (1:30:55.520)

There's always a driver at the moment, although I have noticed that sometimes the driver does

Rodney Brooks (1:31:02.240)

not have the authority to take over without talking to the home office, because they will

Rodney Brooks (1:31:08.000)

sit there waiting for a long time, and clearly something's going on where the home office

Lex Fridman (1:31:14.640)

is making a decision.

Lex Fridman (1:31:16.960)

So they're, you know, and, and so you can see whether they've got their hands on the

Lex Fridman (1:31:21.600)

wheel or not.

Rodney Brooks (1:31:22.400)

And, and it's the incident resolution time that tells you, gives you some clues.

Lex Fridman (1:31:28.240)

So what year do you think, what's your intuition?

Lex Fridman (1:31:30.720)

What date range are you currently thinking San Francisco would be?

Rodney Brooks (1:31:34.880)

Are you currently thinking San Francisco would be autonomous taxi service from any point

Lex Fridman (1:31:42.960)

A to any point B without a driver?

Lex Fridman (1:31:47.760)

Are you still, are you thinking 10 years from now, 20 years from now, 30 years from now?

Rodney Brooks (1:31:53.040)

Certainly not 10 years from now.

Lex Fridman (1:31:55.440)

It's going to be longer.

Rodney Brooks (1:31:56.880)

If you're allowed to go south of market way longer.

Lex Fridman (1:31:59.520)

And unless it's reengineering of roads.

Lex Fridman (1:32:03.440)

By the way, what's the biggest challenge?

Lex Fridman (1:32:05.120)

You mentioned a few.

Lex Fridman (1:32:06.080)

Is it, is it the delivery trucks?

Rodney Brooks (1:32:09.360)

Is it the edge cases, the computer perception, well, here's a case that I saw outside my

Rodney Brooks (1:32:15.040)

house a few weeks ago, about 8pm on a Friday night, it was getting dark, it was before

Lex Fridman (1:32:20.560)

the solstice.

Rodney Brooks (1:32:23.520)

It was a cruise vehicle come down the hill, turned right and stopped dead, covering the

Lex Fridman (1:32:32.080)

crosswalk.

Lex Fridman (1:32:33.600)

Why did it stop dead?

Lex Fridman (1:32:35.120)

Because there was a human just two feet from it.

Rodney Brooks (1:32:38.480)

Now, I just glanced, I knew what was happening.

Rodney Brooks (1:32:41.680)

The human was a woman was at the door of her car trying to unlock it with one of those

Rodney Brooks (1:32:47.840)

things that, you know, when you don't have a key.

Lex Fridman (1:32:50.480)

That car thought, oh, she could jump out in front of me any second.

Rodney Brooks (1:32:55.520)

As a human, I could tell, no, she's not going to jump out.

Lex Fridman (1:32:57.760)

She's busy trying to unlock her.

Rodney Brooks (1:32:59.360)

She's lost her keys.

Lex Fridman (1:33:00.240)

She's trying to get in the car.

Lex Fridman (1:33:01.200)

And it stayed there for, until I got bored.

Lex Fridman (1:33:05.440)

And so the human driver in there did not take over.

Lex Fridman (1:33:11.600)

But here's the kicker to me.

Rodney Brooks (1:33:14.080)

A guy comes down the hill with a stroller, I assume there's a baby in there, and now

Rodney Brooks (1:33:22.720)

the crosswalk's blocked by this cruise vehicle.

Lex Fridman (1:33:25.760)

What's he going to do?

Rodney Brooks (1:33:27.440)

Cleverly, I think, he decided not to go in front of the car.

Lex Fridman (1:33:30.800)

But he had to go behind it.

Rodney Brooks (1:33:34.960)

He had to get off the crosswalk, out into the intersection, to push his baby around

Lex Fridman (1:33:39.360)

this car, which was stopped there.

Lex Fridman (1:33:41.200)

And no human driver would have stopped there for that length of time.

Lex Fridman (1:33:44.880)

They would have got out and out of the way.

Lex Fridman (1:33:46.880)

And that's another one of my pet peeves, that safety is being compromised for individuals

Lex Fridman (1:33:56.000)

who didn't sign up for having this happen in their neighborhood.

Rodney Brooks (1:33:59.760)

Now you can say that's an edge case, but...

Rodney Brooks (1:34:03.200)

Yeah, well, I'm in general not a fan of anecdotal evidence for stuff like this is one of my

Rodney Brooks (1:34:13.040)

biggest problems with the discussion of autonomous vehicles in general, people that criticize

Rodney Brooks (1:34:17.920)

them or support them are using edge cases, are using anecdotal evidence, but I got you.

Lex Fridman (1:34:24.640)

Your question is, when is it going to happen in San Francisco?

Lex Fridman (1:34:26.800)

I say not soon, but it's going to be one of them.

Lex Fridman (1:34:29.040)

But where it is going to happen is in limited domains, campuses of various sorts, gated

Lex Fridman (1:34:38.640)

communities where the other drivers are not arbitrary people.

Rodney Brooks (1:34:46.000)

They're people who know about these things, they've been warned about them, and at velocities

Lex Fridman (1:34:52.800)

where it's always safe to stop dead.

Rodney Brooks (1:34:57.120)

You can't do that on the freeway.

Rodney Brooks (1:34:58.720)

That I think we're going to start to see, and they may not be shaped like current cars,

Rodney Brooks (1:35:06.160)

they may be things like May Mobility has those things and various companies have these.

Lex Fridman (1:35:12.560)

Yeah, I wonder if that's a compelling experience.

Rodney Brooks (1:35:14.400)

To me, it's not just about automation, it's about creating a product that makes your...

Lex Fridman (1:35:20.320)

It's not just cheaper, but it's fun to ride.

Rodney Brooks (1:35:23.680)

One of the least fun things is for a car that stops and waits.

Rodney Brooks (1:35:29.600)

There's something deeply frustrating for us humans for the rest of the world to take advantage

Rodney Brooks (1:35:34.400)

of us as we wait.

Lex Fridman (1:35:35.520)

But think about not you as the customer, but someone who's in their 80s in a retirement

Rodney Brooks (1:35:47.520)

village whose kids have said, you're not driving anymore, and this gives you the freedom to

Lex Fridman (1:35:53.200)

go to the market.

Rodney Brooks (1:35:54.240)

That's a hugely beneficial thing, but it's a very few orders of magnitude less impact

Lex Fridman (1:35:59.840)

on the world.

Rodney Brooks (1:36:00.800)

It's just a few people in a small community using cars as opposed to the entirety of the

Lex Fridman (1:36:05.760)

world.

Rodney Brooks (1:36:07.920)

I like that the first time that a car equipped with some version of a solution to the trolley

Lex Fridman (1:36:13.600)

problem is...

Lex Fridman (1:36:14.800)

What's NIML stand for?

Lex Fridman (1:36:16.400)

Not in my life.

Rodney Brooks (1:36:17.040)

Not in my life.

Lex Fridman (1:36:17.680)

I define my lifetime as up to 2050.

Lex Fridman (1:36:20.080)

You know, I ask you, when have you had to decide which person shall I kill?

Lex Fridman (1:36:29.360)

No, you put the brakes on and you break as hard as you can.

Rodney Brooks (1:36:31.760)

You're not making that decision.

Rodney Brooks (1:36:35.360)

I do think autonomous vehicles or semi autonomous vehicles do need to solve the whole pedestrian

Rodney Brooks (1:36:41.280)

problem that has elements of the trolley problem within it, but it's not...

Rodney Brooks (1:36:45.520)

Yeah, well, and I talk about it in one of the articles or blog posts that I wrote, and

Rodney Brooks (1:36:51.760)

people have told me, one of my coworkers has told me he does this.

Lex Fridman (1:36:56.480)

He tortures autonomously driven vehicles and pedestrians will torture them.

Rodney Brooks (1:37:01.600)

Now, once they realize that putting one foot off the curb makes the car think that they

Lex Fridman (1:37:07.360)

might walk into the road, teenagers will be doing that all the time.

Rodney Brooks (1:37:10.800)

I, by the way, one of my, and this is a whole nother discussion, because my main interest

Lex Fridman (1:37:15.440)

with robotics is HRI, human robot interaction.

Rodney Brooks (1:37:19.200)

I believe that robots that interact with humans will have to push back.

Rodney Brooks (1:37:25.520)

Like they can't just be bullied because that creates a very uncompelling experience for

Rodney Brooks (1:37:30.480)

the humans.

Rodney Brooks (1:37:31.280)

Yeah, well, you know, Waymo, before it was called Waymo, discovered that, you know, they

Rodney Brooks (1:37:35.600)

had to do that at four way intersections.

Rodney Brooks (1:37:38.080)

They had to nudge forward to give the cue that they were going to go, because otherwise

Rodney Brooks (1:37:42.800)

the other drivers would just beat them all the time.

Lex Fridman (1:37:46.400)

So you cofounded iRobot, as we mentioned, one of the most successful robotics companies

Rodney Brooks (1:37:52.320)

ever.

Lex Fridman (1:37:53.520)

What are you most proud of with that company and the approach you took to robotics?

Rodney Brooks (1:38:00.480)

Well, there's something I'm quite proud of there, which may be a surprise, but, you know,

Rodney Brooks (1:38:07.840)

I was still on the board when this happened, it was March 2011, and we sent robots to Japan

Lex Fridman (1:38:17.280)

and they were used to help shut down the Fukushima Daiichi nuclear power plant, which was, everything

Rodney Brooks (1:38:27.520)

was, I've been there since, I was there in 2014, and the robots, some of the robots were

Rodney Brooks (1:38:32.240)

still there.

Lex Fridman (1:38:33.120)

I was proud that we were able to do that.

Lex Fridman (1:38:35.600)

Why were we able to do that?

Lex Fridman (1:38:38.000)

And, you know, people have said, well, you know, Japan is so good at robotics.

Rodney Brooks (1:38:42.960)

It was because we had had about 6,500 robots deployed in Iraq and Afghanistan, teleopt,

Lex Fridman (1:38:51.600)

but with intelligence, dealing with roadside bombs.

Lex Fridman (1:38:56.480)

So we had, it was at that time, nine years of in field experience with the robots in

Rodney Brooks (1:39:03.360)

harsh conditions, whereas the Japanese robots, which were, you know, getting, this goes back

Rodney Brooks (1:39:09.200)

to what annoys me so much, getting all the hype, look at that, look at that Honda robot,

Rodney Brooks (1:39:14.560)

it can walk, wow, the future's here, couldn't do a thing because they weren't deployed,

Lex Fridman (1:39:20.800)

but we had deployed in really harsh conditions for a long time, and so we're able to do

Lex Fridman (1:39:26.960)

something very positive in a very bad situation.

Lex Fridman (1:39:30.400)

What about just the simple, and for people who don't know, one of the things that iRobot

Lex Fridman (1:39:36.640)

has created is the Roomba vacuum cleaner.

Lex Fridman (1:39:42.320)

What about the simple robot that, that is the Roomba, quote unquote, simple, that's

Lex Fridman (1:39:47.760)

deployed in tens of millions of, in tens of millions of homes?

Lex Fridman (1:39:53.200)

What do you think about that?

Rodney Brooks (1:39:54.240)

Well, I make the joke that I started out life as a pure mathematician and turned into a

Rodney Brooks (1:39:59.440)

vacuum cleaner salesman, so if you're going to be an entrepreneur, be ready for, be ready

Rodney Brooks (1:40:05.440)

to do anything, but I was, you know, there was a, there was a wacky lawsuit that I got

Rodney Brooks (1:40:15.040)

opposed for not too many years ago, and I was the only one who had emailed from the

Rodney Brooks (1:40:20.800)

1990s, and no one in the company had it, so I went and went through my email, and it

Lex Fridman (1:40:27.520)

reminded me of, you know, the joy of what we were doing, and what was I doing?

Lex Fridman (1:40:34.880)

What was I doing at the time we were building, building the Roomba?

Rodney Brooks (1:40:41.920)

One of the things was we had this, you know, incredibly tight budget because we wanted

Lex Fridman (1:40:46.160)

to put it on the shelves at $200.

Rodney Brooks (1:40:50.960)

There was another home cleaning robot at the time, it was the Electrolux Trilobite, which

Rodney Brooks (1:40:59.120)

sold for 2,000 euros, and to us that was not going to be a consumer product, so we had

Rodney Brooks (1:41:05.360)

reason to believe that $200 was a, was a thing that people would buy at.

Rodney Brooks (1:41:10.480)

That was our aim, but that meant we had, you know, that's on the shelf making profit.

Rodney Brooks (1:41:19.120)

That means the cost of goods has to be minimal, so I find all these emails of me going, you

Rodney Brooks (1:41:26.560)

know, I'd be in Taipei for a MIT meeting, and I'd stay a few extra days and go down

Rodney Brooks (1:41:32.000)

to Hsinchu and talk to these little tiny companies, lots of little tiny companies outside of TSMC,

Rodney Brooks (1:41:38.800)

Taiwan Semiconductor Manufacturing Corporation, which let all these little companies be fabulous.

Rodney Brooks (1:41:45.440)

They didn't have to have their own fab so they could innovate, and they were building,

Rodney Brooks (1:41:51.760)

their innovations were to build, strip down 6802s, 6802 was what was in an Apple I, get

Rodney Brooks (1:41:57.840)

rid of half the silicon and still have it be viable, and I'd previously got some of

Rodney Brooks (1:42:03.600)

those for some earlier failed products of iRobot, and that was in Hong Kong going to

Rodney Brooks (1:42:11.520)

all these companies that built, you know, they weren't gaming in the current sense,

Rodney Brooks (1:42:16.800)

there were these handheld games that you would play, or birthday cards, because we had about

Rodney Brooks (1:42:23.360)

a 50 cent budget for computation, so I'm trekking from place to place looking at their chips,

Rodney Brooks (1:42:30.640)

looking at what they'd removed, ah, their interrupt handling is too weak for a general

Rodney Brooks (1:42:38.320)

purpose, so I was going deep technical detail, and then I found this one from a company called

Rodney Brooks (1:42:43.440)

Winbond, which had, and I'd forgotten it had this much RAM, it had 512 bytes of RAM,

Lex Fridman (1:42:50.000)

and it was in our budget, and it had all the capabilities we needed.

Lex Fridman (1:42:54.640)

Yeah, and you were excited.

Rodney Brooks (1:42:57.200)

Yeah, and I was reading all these emails, Colin, I found this, so.

Lex Fridman (1:43:02.400)

Did you think, did you ever think that you guys could be so successful?

Lex Fridman (1:43:07.200)

Like, eventually this company would be so successful, could you possibly have imagined?

Lex Fridman (1:43:12.240)

No, we never did think that.

Rodney Brooks (1:43:13.760)

We'd had 14 failed business models up to 2002, and then we had two winners the same year.

Rodney Brooks (1:43:19.200)

No, and then, you know, we, I remember the board, because by this time we had some venture

Rodney Brooks (1:43:27.600)

capital in, the board went along with us building some robots for, you know, aiming at the Christmas

Rodney Brooks (1:43:36.240)

2002 market, and we went three times over what they authorized and built 70,000 of them,

Lex Fridman (1:43:44.640)

and sold them all in that first, because we released on September 18th, and they were

Lex Fridman (1:43:51.200)

all sold by Christmas.

Lex Fridman (1:43:52.560)

So it was, so we were gutsy, but.

Lex Fridman (1:43:57.040)

But yeah, you didn't think this will take over the world.

Rodney Brooks (1:44:00.640)

Well, this is, so a lot of amazing robotics companies have gone under over the past few

Lex Fridman (1:44:09.040)

decades.

Lex Fridman (1:44:10.560)

Why do you think it's so damn hard to run a successful robotics company?

Lex Fridman (1:44:17.680)

There's a few things.

Rodney Brooks (1:44:20.960)

One is expectations of capabilities by the founders that are off base.

Lex Fridman (1:44:29.680)

The founders, not the consumer, the founders.

Rodney Brooks (1:44:31.600)

Yeah, expectations of what can be delivered.

Lex Fridman (1:44:34.000)

Sure.

Rodney Brooks (1:44:34.500)

Mispricing, and what a customer thinks is a valid price, is not rational, necessarily.

Lex Fridman (1:44:42.180)

Yeah.

Lex Fridman (1:44:43.620)

And expectations of customers, and just the sheer hardness of getting people to adopt a

Lex Fridman (1:44:56.100)

new technology.

Lex Fridman (1:44:57.060)

And I've suffered from all three of these, you know.

Lex Fridman (1:44:59.700)

I've had more failures than successes, in terms of companies.

Rodney Brooks (1:45:04.820)

I've suffered from all three.

Rodney Brooks (1:45:07.860)

So, do you think one day there will be a robotics company, and by robotics company, I mean, where

Lex Fridman (1:45:18.580)

your primary source of income is from robots, that will be a trillion plus dollar company?

Lex Fridman (1:45:24.740)

And if so, what would that company do?

Rodney Brooks (1:45:31.460)

I can't, you know, because I'm still starting robot companies.

Lex Fridman (1:45:35.300)

Yeah.

Rodney Brooks (1:45:38.180)

I'm not making any such predictions in my own mind.

Lex Fridman (1:45:41.380)

I'm not thinking about a trillion dollar company.

Lex Fridman (1:45:43.140)

And by the way, I don't think, you know, in the 90s, anyone was thinking that Apple would

Lex Fridman (1:45:47.220)

ever be a trillion dollar company.

Rodney Brooks (1:45:48.580)

So, these are, these are, you know, these are, you know, these are, you know, these

Rodney Brooks (1:45:52.580)

would be a trillion dollar company, so these are, these are very hard to predict.

Rodney Brooks (1:45:57.220)

But, sorry to interrupt, but don't you, because I kind of have a vision in a small way, and

Rodney Brooks (1:46:03.460)

it's a big vision in a small way, that I see that there would be robots in the home,

Rodney Brooks (1:46:10.180)

at scale, like Roomba, but more.

Lex Fridman (1:46:13.540)

And that's trillion dollar.

Rodney Brooks (1:46:15.620)

Right.

Lex Fridman (1:46:16.120)

And I think there's a real market pull for them because of the demographic inversion,

Lex Fridman (1:46:22.100)

you know, who's going to do all the stuff for the older people?

Lex Fridman (1:46:26.180)

There's too many, you know, I'm leading here.

Rodney Brooks (1:46:31.700)

There's going to be too many of us.

Lex Fridman (1:46:36.420)

But we don't have capable enough robots to make that economic argument at this point.

Lex Fridman (1:46:42.340)

Do I expect that that will happen?

Lex Fridman (1:46:44.180)

Yes, I expect it will happen.

Lex Fridman (1:46:45.380)

But I got to tell you, we introduced the Roomba in 2002, and I stayed another

Lex Fridman (1:46:50.580)

nine years.

Rodney Brooks (1:46:51.780)

We were always trying to find what the next home robot would be, and still today, the

Rodney Brooks (1:46:57.700)

primary product of 20 years late, almost 20 years later, 19 years later, the primary product

Rodney Brooks (1:47:02.660)

is still the Roomba.

Lex Fridman (1:47:03.620)

So iRobot hasn't found the next one.

Lex Fridman (1:47:07.060)

Do you think it's possible for one person in the garage to build it versus, like, Google

Lex Fridman (1:47:12.580)

launching Google self driving car that turns into Waymo?

Lex Fridman (1:47:16.340)

Do you think this is almost like what it takes to build a successful robotics company?

Lex Fridman (1:47:20.980)

Do you think it's possible to go from the ground up, or is it just too much capital

Lex Fridman (1:47:24.420)

investment?

Lex Fridman (1:47:25.540)

Yeah, so it's very hard to get there without a lot of capital.

Lex Fridman (1:47:31.700)

And we're starting to see, you know, fair chunks of capital for some robotics companies.

Rodney Brooks (1:47:38.100)

You know, Series B's, I saw one yesterday for $80 million, I think it was, for Covariant.

Lex Fridman (1:47:45.540)

But it can take real money to get into these things, and you may fail along the way.

Rodney Brooks (1:47:54.740)

I've certainly failed at Rethink Robotics, and we lost $150 million in capital there.

Rodney Brooks (1:48:00.900)

So, okay, so Rethink Robotics is another amazing robotics company you cofounded.

Lex Fridman (1:48:06.580)

So what was the vision there?

Lex Fridman (1:48:09.060)

What was the dream?

Lex Fridman (1:48:11.140)

And what are you most proud of with Rethink Robotics?

Rodney Brooks (1:48:15.620)

I'm most proud of the fact that we got robots out of the cage in factories that were safe,

Lex Fridman (1:48:23.140)

absolutely safe, for people and robots to be next to each other.

Lex Fridman (1:48:26.180)

So these are robotic arms.

Lex Fridman (1:48:27.700)

Robotic arms.

Rodney Brooks (1:48:28.500)

Able to pick up stuff and interact with humans.

Lex Fridman (1:48:31.140)

Yeah, and that humans could retask them without writing code.

Lex Fridman (1:48:35.140)

And now that's sort of become an expectation for a lot of other little companies and big

Lex Fridman (1:48:40.020)

companies, our advertising they're doing.

Rodney Brooks (1:48:42.260)

That's both an interface problem and also a safety problem.

Lex Fridman (1:48:45.540)

Yeah, yeah.

Lex Fridman (1:48:47.620)

So I'm most proud of that.

Lex Fridman (1:48:51.300)

I completely, I let myself be talked out of what I wanted to do.

Rodney Brooks (1:48:59.380)

And, you know, you always got, you know, I can't replay the tape.

Lex Fridman (1:49:02.260)

I can't replay it.

Rodney Brooks (1:49:05.460)

Maybe, you know, if I'd been stronger on, and I remember the day, I remember the exact

Lex Fridman (1:49:12.180)

meeting.

Lex Fridman (1:49:13.860)

Can you take me through that meeting?

Lex Fridman (1:49:16.260)

Yeah.

Lex Fridman (1:49:18.340)

So I'd said that I'd set as a target for the company that we were going to build $3,000

Lex Fridman (1:49:23.940)

robots with force feedback that was safe for people to be around.

Rodney Brooks (1:49:29.700)

Wow.

Lex Fridman (1:49:30.420)

That was my goal.

Lex Fridman (1:49:31.380)

And we built, so we started in 2008, and we had prototypes built of plastic, plastic

Rodney Brooks (1:49:38.980)

gearboxes, and at a $3,000, you know, lifetime, or $3,000, I was saying, we're going to go

Rodney Brooks (1:49:48.180)

after not the people who already have robot arms in factories, the people who would never

Lex Fridman (1:49:52.500)

have a robot arm.

Rodney Brooks (1:49:53.940)

We're going to go after a different market.

Lex Fridman (1:49:55.940)

So we don't have to meet their expectations.

Lex Fridman (1:49:57.940)

And so we're going to build it out of plastic.

Lex Fridman (1:49:59.860)

It doesn't have to have a $35,000 lifetime.

Rodney Brooks (1:50:02.740)

It's going to be so cheap that it's OpEx, not CapEx.

Lex Fridman (1:50:09.140)

And so we had a prototype that worked reasonably well, but the control engineers were complaining

Rodney Brooks (1:50:16.980)

about these plastic gearboxes with a beautiful little planetary gearbox that we could use

Lex Fridman (1:50:24.820)

something called series elastic actuators.

Rodney Brooks (1:50:29.780)

We embedded them in there.

Lex Fridman (1:50:30.980)

We could measure forces.

Rodney Brooks (1:50:32.180)

We knew when we hit something, et cetera.

Rodney Brooks (1:50:35.060)

The control engineers were saying, yeah, but there's this torque ripple because these plastic

Rodney Brooks (1:50:40.100)

gears, they're not great gears, and there's this ripple, and trying to do force control

Lex Fridman (1:50:44.900)

around this ripple is so hard.

Lex Fridman (1:50:47.220)

And I'm not going to name names, but I remember one of the mechanical engineers saying, we'll

Lex Fridman (1:50:55.140)

just build a metal gearbox with spur gears, and it'll take six weeks.

Rodney Brooks (1:50:59.620)

We'll be done.

Lex Fridman (1:51:01.140)

Problem solved.

Rodney Brooks (1:51:03.700)

Two years later, we got the spur gearbox working.

Lex Fridman (1:51:08.020)

We cost reduced it every possible way we could, but now the price went up too.

Lex Fridman (1:51:15.540)

And then the CEO at the time said, well, we have to have two arms, not one arm.

Lex Fridman (1:51:19.860)

So our first robot product, Baxter, now cost $25,000, and the only people who were going

Rodney Brooks (1:51:27.460)

to look at that were people who had arms in factories because that was somewhat cheaper

Lex Fridman (1:51:31.460)

for two arms than arms in factories.

Lex Fridman (1:51:34.180)

But they were used to 0.1 millimeter reproducibility of motion and certain velocities, and I kept

Lex Fridman (1:51:43.700)

thinking, but that's not what we're giving you.

Rodney Brooks (1:51:45.620)

You don't need position repeatability.

Lex Fridman (1:51:47.380)

Use force control like a human does.

Rodney Brooks (1:51:49.700)

No, no, but we want that repeatability.

Lex Fridman (1:51:53.060)

We want that repeatability.

Rodney Brooks (1:51:54.500)

All the other robots have that repeatability.

Lex Fridman (1:51:56.340)

Why don't you have that repeatability?

Lex Fridman (1:51:58.500)

So can you clarify?

Lex Fridman (1:51:59.780)

Force control is you can grab the arm and you can move it.

Rodney Brooks (1:52:02.900)

You can move it around, but suppose you...

Lex Fridman (1:52:06.100)

Can you see that?

Rodney Brooks (1:52:06.900)

Yes.

Lex Fridman (1:52:07.540)

Suppose you want to...

Rodney Brooks (1:52:09.940)

Yes.

Rodney Brooks (1:52:10.440)

Suppose this thing is a precise thing that's got to fit here in this right angle.

Rodney Brooks (1:52:16.520)

Under position control, you have fixtured where this is.

Lex Fridman (1:52:20.520)

You know where this is precisely, and you just move it, and it goes there.

Rodney Brooks (1:52:25.320)

In force control, you would do something like slide over here till we feel that and slide

Lex Fridman (1:52:30.120)

it in there, and that's how a human gets precision.

Rodney Brooks (1:52:34.040)

They use force feedback and get the things to mate rather than just go straight to it.

Rodney Brooks (1:52:42.440)

Couldn't convince our customers who were in factories and were used to thinking about

Rodney Brooks (1:52:48.120)

things a certain way, and they wanted it, wanted it, wanted it.

Lex Fridman (1:52:51.880)

So then we said, okay, we're going to build an arm that gives you that.

Lex Fridman (1:52:56.120)

So now we ended up building a $35,000 robot with one arm with...

Lex Fridman (1:52:59.880)

Oh, what are they called?

Rodney Brooks (1:53:04.840)

A certain sort of gearbox made by a company whose name I can't remember right now, but

Lex Fridman (1:53:08.520)

it's the name of the gearbox.

Lex Fridman (1:53:11.880)

But it's got torque ripple in it.

Lex Fridman (1:53:15.560)

So now there was an extra two years of solving the problem of doing the force with the torque

Rodney Brooks (1:53:19.720)

ripple.

Lex Fridman (1:53:20.200)

So we had to do the thing we had avoided for the plastic gearboxes, which is a little bit

Rodney Brooks (1:53:28.440)

for the plastic gearboxes we ended up having to do.

Lex Fridman (1:53:31.240)

The robot was now overpriced and they...

Lex Fridman (1:53:35.240)

And that was your intuition from the very beginning kind of that this is not...

Rodney Brooks (1:53:40.040)

You're opening a door to solve a lot of problems that you're eventually going to have to solve

Rodney Brooks (1:53:44.760)

this problem anyway.

Lex Fridman (1:53:45.800)

Yeah.

Lex Fridman (1:53:46.120)

And also I was aiming at a low price to go into a different market.

Lex Fridman (1:53:49.240)

Low price.

Rodney Brooks (1:53:50.280)

That didn't have robots.

Lex Fridman (1:53:51.160)

$3,000 would be amazing.

Rodney Brooks (1:53:52.600)

Yeah.

Lex Fridman (1:53:52.760)

I think we could have done it for five.

Rodney Brooks (1:53:54.120)

But, you know, you talked about setting the goal a little too far for the engineers.

Lex Fridman (1:53:58.840)

Yeah, exactly.

Lex Fridman (1:54:02.280)

So why would you say that company not failed, but went under?

Rodney Brooks (1:54:09.000)

We had buyers and there's this thing called the Committee on Foreign Investment in the

Rodney Brooks (1:54:15.400)

U.S., CFIUS.

Lex Fridman (1:54:18.120)

And that had previously been invoked twice.

Rodney Brooks (1:54:21.640)

Around where the government could stop foreign money coming into a U.S. company based on

Lex Fridman (1:54:29.640)

defense requirements.

Rodney Brooks (1:54:32.680)

We went through due diligence multiple times.

Rodney Brooks (1:54:34.600)

We were going to get acquired, but every consortium had Chinese money in it, and all the bankers

Rodney Brooks (1:54:42.280)

would say at the last minute, you know, this isn't going to get past CFIUS, and the investors

Lex Fridman (1:54:47.080)

would go away.

Lex Fridman (1:54:47.880)

And then we had two buyers, once we were about to run out of money, two buyers, and one used

Rodney Brooks (1:54:54.280)

heavy handed legal stuff with the other one, said they were going to take it and pay more,

Rodney Brooks (1:55:02.760)

dropped out when we were out of cash, and then bought the assets at 1 30th of the price

Lex Fridman (1:55:08.040)

they had offered a week before.

Rodney Brooks (1:55:10.920)

It was a tough week.

Lex Fridman (1:55:12.280)

Do you, does it hurt to think about like an amazing company that didn't, you know, like

Lex Fridman (1:55:21.640)

iRobot didn't find a way?

Lex Fridman (1:55:24.440)

Yeah, it was tough.

Rodney Brooks (1:55:25.400)

I said I was never going to start another company.

Rodney Brooks (1:55:27.480)

I was pleased that everyone liked what we did so much that the team was hired by three

Rodney Brooks (1:55:36.360)

companies, and I was very happy that we were able to do that.

Lex Fridman (1:55:40.040)

Three companies within a week.

Rodney Brooks (1:55:42.920)

Everyone had a job in one of these three companies.

Rodney Brooks (1:55:44.760)

Some stayed in their same desks because another company came in and rented the space.

Lex Fridman (1:55:50.680)

So I felt good about people not being out on the street.

Lex Fridman (1:55:55.720)

So Baxter has a screen with a face.

Rodney Brooks (1:55:59.560)

What, that's a revolutionary idea for a robot manipulation, like for a robotic arm.

Lex Fridman (1:56:07.320)

How much opposition did you get?

Rodney Brooks (1:56:08.840)

Well, first the screen was also used during codeless programming.

Lex Fridman (1:56:12.920)

We taught by demonstration.

Rodney Brooks (1:56:14.440)

It showed you what its understanding of the task was.

Lex Fridman (1:56:17.640)

So it had two roles.

Rodney Brooks (1:56:21.240)

Some customers hated it, and so we made it so that when the robot was running it could

Lex Fridman (1:56:26.520)

be showing graphs of what was happening and not show the eyes.

Rodney Brooks (1:56:30.200)

Other people, and some of them surprised me who they were, saying well this one doesn't

Lex Fridman (1:56:36.600)

look as human as the old one.

Rodney Brooks (1:56:37.960)

We liked the human looking.

Lex Fridman (1:56:39.640)

Yeah.

Lex Fridman (1:56:40.120)

So there was a mixed bag.

Lex Fridman (1:56:43.240)

But do you think that's, I don't know, I'm kind of disappointed whenever I talk to

Rodney Brooks (1:56:50.360)

roboticists, like the best robotics people in the world, they seem to not want to do

Lex Fridman (1:56:55.160)

the eyes type of thing.

Rodney Brooks (1:56:56.760)

Like they seem to see it as a machine as opposed to a machine that can also have a human connection.

Lex Fridman (1:57:02.760)

I'm not sure what to do with that.

Rodney Brooks (1:57:03.960)

It seems like a lost opportunity.

Rodney Brooks (1:57:05.480)

I think the trillion dollar company will have to do the human connection very well no matter

Lex Fridman (1:57:10.440)

what it does.

Lex Fridman (1:57:11.160)

Yeah, I agree.

Lex Fridman (1:57:13.800)

Can I ask you a ridiculous question?

Lex Fridman (1:57:15.560)

Sure.

Rodney Brooks (1:57:17.000)

I might give a ridiculous answer.

Lex Fridman (1:57:19.880)

Do you think, well maybe by way of asking the question, let me first mention that you're

Rodney Brooks (1:57:25.640)

kind of critical of the idea of the Turing test as a test of intelligence.

Lex Fridman (1:57:32.280)

Let me first ask this question.

Lex Fridman (1:57:33.640)

Do you think we'll be able to build an AI system that humans fall in love with and it

Lex Fridman (1:57:40.360)

falls in love with the human, like romantic love?

Rodney Brooks (1:57:46.920)

Well, we've had that with humans falling in love with cars even back in the 50s.

Lex Fridman (1:57:51.560)

It's a different love, right?

Rodney Brooks (1:57:52.680)

Well, yeah.

Rodney Brooks (1:57:53.640)

I think there's a lifelong partnership where you can communicate and grow like...

Rodney Brooks (1:57:59.640)

I think we're a long way from that.

Lex Fridman (1:58:01.160)

I think we're a long, long way.

Rodney Brooks (1:58:03.000)

I think Blade Runner had the time scale totally wrong.

Rodney Brooks (1:58:10.440)

Yeah, but so to me, honestly, the most difficult part is the thing that you said with the Marvex

Rodney Brooks (1:58:16.840)

Paradox is to create a human form that interacts and perceives the world.

Lex Fridman (1:58:21.400)

But if we just look at a voice, like the movie Her or just like an Alexa type voice, I tend

Rodney Brooks (1:58:28.040)

to think we're not that far away.

Rodney Brooks (1:58:29.560)

Well, for some people, maybe not, but as humans, as we think about the future, we always try

Rodney Brooks (1:58:43.400)

to...

Lex Fridman (1:58:44.200)

And this is the premise of most science fiction movies.

Rodney Brooks (1:58:46.920)

You've got the world just as it is today and you change one thing.

Lex Fridman (1:58:50.920)

But that's not how...

Lex Fridman (1:58:51.960)

And it's the same with a self driving car.

Lex Fridman (1:58:53.960)

You change one thing.

Rodney Brooks (1:58:55.000)

No, everything changes.

Lex Fridman (1:58:56.840)

Everything grows together.

Lex Fridman (1:58:59.720)

So surprisingly, it might be surprising to you or might not, I think the best movie about

Lex Fridman (1:59:04.520)

this stuff was Bicentennial Man.

Lex Fridman (1:59:09.160)

And what was happening there?

Lex Fridman (1:59:11.080)

It was schmaltzy and, you know, but what was happening there?

Rodney Brooks (1:59:15.720)

As the robot was trying to become more human, the humans were adopting the technology of

Lex Fridman (1:59:21.160)

the robot and changing their bodies.

Lex Fridman (1:59:23.080)

So there was a convergence happening in a sense.

Lex Fridman (1:59:27.160)

So we will not be the same.

Rodney Brooks (1:59:28.760)

You know, we're already talking about genetically modifying our babies.

Lex Fridman (1:59:32.440)

You know, there's more and more stuff happening around that.

Rodney Brooks (1:59:36.680)

We will want to modify ourselves even more for all sorts of things.

Lex Fridman (1:59:43.240)

We put all sorts of technology in our bodies to improve it.

Rodney Brooks (1:59:48.440)

You know, I've got things in my ears so that I can sort of hear you.

Lex Fridman (1:59:53.560)

Yeah.

Lex Fridman (1:59:56.120)

So we're always modifying our bodies.

Rodney Brooks (1:59:57.480)

So, you know, I think it's hard to imagine exactly what it will be like in the future.

Rodney Brooks (20:01.360)

It can do stuff that our raw reasoning can't do, and we've got conventions of when you

Lex Fridman (20:06.640)

can use it or not.

Lex Fridman (20:08.480)

But sometimes, you know, people try to all the time, we always try to get physical metaphors

Rodney Brooks (20:15.280)

for things, which is why quantum mechanics has been such a problem for a hundred years.

Rodney Brooks (20:21.040)

Because it's a particle.

Lex Fridman (20:22.080)

No, it's a wave.

Rodney Brooks (20:22.880)

It's got to be something we understand.

Lex Fridman (20:24.640)

And I say, no, it's some weird mathematical logic that's different from those, but we

Rodney Brooks (20:29.040)

want that metaphor.

Rodney Brooks (20:30.720)

Well, you know, I suspect that, you know, a hundred years or 200 years from now, neither

Rodney Brooks (20:35.680)

quantum mechanics nor dark matter will be talked about in the same terms, you know,

Lex Fridman (20:39.920)

in the same way that Flogerson's theory eventually went away.

Rodney Brooks (20:44.320)

Because it just wasn't an adequate explanatory metaphor, you know.

Rodney Brooks (20:49.440)

That metaphor was the stuff, there is stuff in the burning, the burning is in the matter.

Rodney Brooks (20:56.000)

As it turns out, the burning was outside the matter, it was the oxygen.

Lex Fridman (20:59.840)

So our desire for metaphor and combined with our limited cognitive capabilities gets us

Rodney Brooks (21:05.440)

into trouble.

Lex Fridman (21:06.400)

That's my argument in this book.

Lex Fridman (21:08.320)

Now, and people say, well, what is it then?

Lex Fridman (21:10.080)

And I say, well, I wish I knew that, right, the book about that.

Lex Fridman (21:12.720)

But I, you know, I give some ideas.

Lex Fridman (21:14.640)

But so there's the three things.

Rodney Brooks (21:17.440)

Computation is sort of a particular thing we use.

Lex Fridman (21:22.880)

Oh, can I tell you one beautiful thing, one beautiful thing I found?

Rodney Brooks (21:26.320)

So, you know, I used an example of a thing that's different from computation.

Rodney Brooks (21:30.000)

You hit a drum and it vibrates, and there are some stationary points on the drum surface,

Rodney Brooks (21:35.520)

you know, because the waves are going up and down the stationary points.

Rodney Brooks (21:37.840)

Now, you could compute them to arbitrary precision, but the drum just knows them.

Rodney Brooks (21:45.760)

The drum doesn't have to compute.

Lex Fridman (21:47.760)

What was the very first computer program ever written by Ada Lovelace?

Rodney Brooks (21:51.920)

To compute Bernoulli numbers, and the Bernoulli numbers are exactly what you need to find those

Lex Fridman (21:56.240)

stable points in the drum surface.

Rodney Brooks (21:58.320)

Wow.

Lex Fridman (21:59.520)

And there was a bug in the program.

Rodney Brooks (22:03.280)

The arguments to divide were, I don't know, I don't know.

Lex Fridman (22:06.400)

The arguments to divide were reversed in one place.

Lex Fridman (22:10.000)

And it still worked?

Lex Fridman (22:11.040)

Well, no, she's never got to run it.

Rodney Brooks (22:12.560)

They never built the analytical engine.

Lex Fridman (22:14.000)

She wrote the program without it, you know.

Lex Fridman (22:19.040)

So the computation?

Rodney Brooks (22:21.040)

Computation is sort of, you know, a thing that's become dominant as a metaphor, but

Lex Fridman (22:27.200)

is it the right metaphor?

Lex Fridman (22:29.680)

All three of these four fields adopted computation.

Rodney Brooks (22:33.360)

And, you know, a lot of it swirls around Warren McCulloch and all his students, and he funded

Lex Fridman (22:40.480)

a lot of people.

Lex Fridman (22:45.600)

And our human metaphors, our limitations to human thinking, all play into this.

Lex Fridman (22:50.000)

Those are the three themes of the book.

Lex Fridman (22:52.720)

So I have a little to say about computation.

Lex Fridman (22:54.880)

So you're saying that there is a gap between the computer or the machine that performs

Rodney Brooks (23:05.040)

computation and this machine that appears to have consciousness and intelligence.

Lex Fridman (23:13.360)

Yeah, that piece of meat in your head.

Rodney Brooks (23:16.080)

Piece of meat.

Lex Fridman (23:16.800)

And maybe it's not just the meat in your head, it's the rest of you too.

Rodney Brooks (23:20.720)

I mean, you actually have a neural system in your gut.

Rodney Brooks (23:24.960)

I tend to also believe, not believe, but we're now dancing around things we don't know, but

Rodney Brooks (23:31.680)

I tend to believe other humans are important.

Rodney Brooks (23:36.560)

Like, so we're almost like, I just don't think we would ever have achieved the level

Rodney Brooks (23:42.080)

of intelligence we have with other humans.

Rodney Brooks (23:44.880)

I'm not saying so confidently, but I have an intuition that some of the intelligence

Rodney Brooks (23:49.680)

is in the interaction.

Rodney Brooks (23:51.200)

Yeah, and I think it seems to be very likely, again, this is speculation, but we, our species,

Lex Fridman (24:00.240)

and probably neanderthals to some extent, because you can find old bones where they

Rodney Brooks (24:06.800)

seem to be counting on them by putting notches that were neanderthals, we are able to put

Rodney Brooks (24:15.360)

some of our stuff outside our body into the world.

Lex Fridman (24:18.400)

And then other people can share it.

Lex Fridman (24:20.400)

And then we get these tools that become shared tools.

Lex Fridman (24:22.960)

And so there's a whole coupling that would not occur in the single deep learning network,

Rodney Brooks (24:30.240)

which was fed all of literature or something.

Lex Fridman (24:33.840)

Yeah, the neural network can't step outside of itself.

Lex Fridman (24:38.320)

But is there some, can we explore this dark room a little bit and try to get at something?

Lex Fridman (24:46.640)

What is the magic?

Lex Fridman (24:47.840)

Where does the magic come from in the human brain that creates the mind?

Lex Fridman (24:52.480)

What's your sense as scientists that try to understand it and try to build it?

Lex Fridman (24:58.880)

What are the directions it followed might be productive?

Lex Fridman (25:04.240)

Is it creative, interactive robots?

Rodney Brooks (25:07.040)

Is it creating large deep neural networks that do like self supervised learning and

Rodney Brooks (25:13.440)

just like we'll discover that when you make something large enough, some interesting things

Lex Fridman (25:18.800)

will emerge?

Lex Fridman (25:19.840)

Is it through physics and chemistry, biology, like artificial life angle?

Rodney Brooks (25:23.600)

Like we'll sneak up in this four quadrant matrix that you mentioned.

Lex Fridman (25:28.240)

Is there anything you're most, if you had to bet all your money, financial?

Rodney Brooks (25:33.440)

I wouldn't.

Lex Fridman (25:35.040)

So every intelligence we know, animal intelligence, dog intelligence,

Rodney Brooks (25:40.960)

octopus intelligence, which is a very different sort of architecture from us.

Rodney Brooks (25:49.920)

All the intelligences we know perceive the world in some way and then have action in

Rodney Brooks (25:59.520)

the world, but they're able to perceive objects in a way which is actually pretty damn phenomenal

Lex Fridman (26:11.520)

and surprising.

Rodney Brooks (26:13.200)

We tend to think that the box over here between us, which is a sound box, I think is a blue

Lex Fridman (26:22.000)

box, but blueness is something that we construct with color constancy.

Rodney Brooks (26:32.560)

The blueness is not a direct function of the photons we're receiving.

Rodney Brooks (26:37.120)

It's actually context, which is why you can turn, maybe seeing the examples where someone

Rodney Brooks (26:47.600)

turns a stop sign into some other sort of sign by just putting a couple of marks on

Lex Fridman (26:53.520)

them and the deep learning system gets it wrong.

Lex Fridman (26:55.280)

And everyone says, but the stop sign's red.

Lex Fridman (26:58.160)

Why is it thinking it's the other sort of sign?

Rodney Brooks (26:59.920)

Because redness is not intrinsic in just the photons.

Rodney Brooks (27:02.800)

It's actually a construction of an understanding of the whole world and the relationship between

Rodney Brooks (27:07.120)

objects to get color constancy.

Lex Fridman (27:11.040)

But our tendency, in order that we get an archive paper really quickly, is you just

Rodney Brooks (27:15.760)

show a lot of data and give the labels and hope it figures it out.

Lex Fridman (27:18.880)

But it's not figuring it out in the same way we do.

Rodney Brooks (27:21.040)

We have a very complex perceptual understanding of the world.

Lex Fridman (27:24.720)

Dogs have a very different perceptual understanding based on smell.

Rodney Brooks (27:28.000)

They go smell a post, they can tell how many different dogs have visited it in the last

Lex Fridman (27:34.880)

10 hours and how long ago.

Rodney Brooks (27:36.320)

There's all sorts of stuff that we just don't perceive about the world.

Lex Fridman (27:39.440)

And just taking a single snapshot is not perceiving about the world.

Rodney Brooks (27:42.400)

It's not seeing the registration between us and the object.

Lex Fridman (27:48.400)

And registration is a philosophical concept.

Rodney Brooks (27:52.160)

Brian Cantwell Smith talks about it a lot.

Lex Fridman (27:54.560)

Very difficult, squirmy thing to understand.

Lex Fridman (27:59.200)

But I think none of our systems do that.

Rodney Brooks (28:02.080)

We've always talked in AI about the symbol grounding problem, how our symbols that we

Rodney Brooks (28:06.000)

talk about are grounded in the world.

Lex Fridman (28:08.080)

And when deep learning came along and started labeling images, people said, ah, the grounding

Rodney Brooks (28:12.320)

problem has been solved.

Rodney Brooks (28:13.440)

No, the labeling problem was solved with some percentage accuracy, which is different from

Rodney Brooks (28:18.800)

the grounding problem.

Lex Fridman (28:20.560)

So you agree with Hans Marvick and what's called the Marvick's paradox that highlights

Rodney Brooks (28:28.880)

this counterintuitive notion that reasoning is easy, but perception and mobility are hard.

Lex Fridman (28:39.440)

Yeah.

Rodney Brooks (28:39.840)

We shared an office when I was working on computer vision and he was working on his

Lex Fridman (28:45.360)

first mobile robot.

Lex Fridman (28:46.640)

What were those conversations like?

Lex Fridman (28:48.400)

They were great.

Lex Fridman (28:50.160)

So do you still kind of, maybe you can elaborate, do you still believe this kind of notion that

Lex Fridman (28:56.160)

perception is really hard?

Rodney Brooks (28:59.600)

Like, can you make sense of why we humans have this poor intuition about what's hard

Lex Fridman (29:04.080)

and not?

Rodney Brooks (29:04.480)

Well, let me give us sort of another story.

Lex Fridman (29:10.640)

Sure.

Rodney Brooks (29:11.520)

If you go back to the original teams working on AI from the late 50s into the 60s, and

Lex Fridman (29:21.680)

you go to the AI lab at MIT, who was it that was doing that?

Rodney Brooks (29:27.760)

It was a bunch of really smart kids who got into MIT and they were intelligent.

Lex Fridman (29:32.480)

So what's intelligence about?

Rodney Brooks (29:34.160)

Well, the stuff they were good at, playing chess, doing integrals, that was hard stuff.

Rodney Brooks (29:40.480)

But, you know, a baby could see stuff, that wasn't intelligent, anyone could do that,

Rodney Brooks (29:45.680)

that's not intelligence.

Lex Fridman (29:47.280)

And so, you know, there was this intuition that the hard stuff is the things they were

Rodney Brooks (29:52.480)

good at and the easy stuff was the stuff that everyone could do.

Lex Fridman (29:57.440)

Yeah.

Lex Fridman (29:57.760)

And maybe I'm overplaying it a little bit, but I think there's an element of that.

Lex Fridman (2:00:03.640)

But on the Turing test side, do you think, so forget about love for a second, let's talk

Rodney Brooks (2:00:09.720)

about just like the Alexa Prize.

Lex Fridman (2:00:12.280)

Actually, I was invited to be a part of the Alexa Prize.

Rodney Brooks (2:00:16.200)

Actually, I was invited to be a, what is the interviewer for the Alexa Prize or whatever

Lex Fridman (2:00:23.080)

that's in two days.

Rodney Brooks (2:00:25.320)

Their idea is success looks like a person wanting to talk to an AI system for a prolonged

Lex Fridman (2:00:32.440)

period of time, like 20 minutes.

Lex Fridman (2:00:35.080)

How far away are we and why is it difficult to build an AI system with which you'd want

Lex Fridman (2:00:41.400)

to have a beer and talk for an hour or two hours?

Rodney Brooks (2:00:45.720)

Like not for to check the weather or to check music, but just like to talk as friends.

Rodney Brooks (2:00:53.160)

Yeah, well, you know, we saw Weizenbaum back in the 60s with his programmer, Elisa, being

Rodney Brooks (2:01:00.840)

shocked at how much people would talk to Elisa.

Lex Fridman (2:01:03.080)

And I remember, you know, in the 70s typing, you know, stuff to Elisa to see what it would

Rodney Brooks (2:01:08.360)

come back with.

Rodney Brooks (2:01:09.000)

You know, I think right now, and this is a thing that Amazon's been trying to improve

Rodney Brooks (2:01:17.960)

with Alexa, there is no continuity of topic.

Lex Fridman (2:01:22.760)

There's not, you can't refer to what we talked about yesterday.

Rodney Brooks (2:01:27.880)

It's not the same as talking to a person where there seems to be an ongoing existence, which

Lex Fridman (2:01:32.360)

changes.

Rodney Brooks (2:01:33.800)

We share moments together and they last in our memory together.

Lex Fridman (2:01:37.080)

Yeah, there's none of that.

Lex Fridman (2:01:39.000)

And there's no sort of intention of these systems that they have any goal in life, even

Lex Fridman (2:01:46.840)

if it's to be happy, you know, they don't even have a semblance of that.

Rodney Brooks (2:01:51.880)

Now, I'm not saying this can't be done.

Lex Fridman (2:01:53.720)

I'm just saying, I think this is why we don't feel that way about them.

Rodney Brooks (2:01:57.960)

That's a sort of a minimal requirement.

Rodney Brooks (2:02:01.560)

If you want the sort of interaction you're talking about, it's a minimal requirement.

Rodney Brooks (2:02:06.840)

Whether it's going to be sufficient, I don't know.

Lex Fridman (2:02:10.360)

We haven't seen it yet.

Rodney Brooks (2:02:11.560)

We don't know what it feels like.

Rodney Brooks (2:02:14.120)

I tend to think it's not as difficult as solving intelligence, for example, and I think it's

Rodney Brooks (2:02:23.160)

achievable in the near term.

Lex Fridman (2:02:26.680)

But on the Turing test, why don't you think the Turing test is a good test of intelligence?

Rodney Brooks (2:02:32.200)

Oh, because, you know, again, the Turing, if you read the paper, Turing wasn't saying

Lex Fridman (2:02:39.080)

this is a good test.

Rodney Brooks (2:02:40.440)

He was using it as a rhetorical device to argue that if you can't tell the difference

Rodney Brooks (2:02:46.520)

between a computer and a person, you must say that the computer's thinking because you

Rodney Brooks (2:02:52.920)

can't tell the difference, you know, when it's thinking.

Lex Fridman (2:02:56.600)

You can't say something different.

Lex Fridman (2:02:58.280)

What it has become as this sort of weird game of fooling people, so back at the AI Lab in

Rodney Brooks (2:03:08.920)

the late 80s, we had this thing that still goes on called the AI Olympics, and one of

Rodney Brooks (2:03:14.280)

the events we had one year was the original imitation game, as Turing talked about, because

Lex Fridman (2:03:21.320)

he starts by saying, can you tell whether it's a man or a woman?

Lex Fridman (2:03:25.160)

So we did that at the Lab.

Rodney Brooks (2:03:28.680)

You'd go and type, and the thing would come back, and you had to tell whether it was a

Rodney Brooks (2:03:33.720)

man or a woman, and one man came up with a question that he could ask, which was always

Lex Fridman (2:03:50.920)

a dead giveaway of whether the other person was really a man or a woman.

Lex Fridman (2:03:56.520)

He would ask them, did you have green plastic toy soldiers as a kid?

Lex Fridman (2:04:01.400)

Yeah.

Lex Fridman (2:04:01.880)

What did you do with them?

Lex Fridman (2:04:03.240)

And a woman trying to be a man would say, oh, I lined them up.

Rodney Brooks (2:04:07.160)

We had wars.

Lex Fridman (2:04:07.800)

We had battles.

Lex Fridman (2:04:08.760)

And the man, just being a man, would say, I stomped on them.

Lex Fridman (2:04:11.240)

I burned them.

Lex Fridman (2:04:11.960)

So that's what the Turing test with computers has become.

Lex Fridman (2:04:21.480)

What's the trick question?

Rodney Brooks (2:04:23.560)

That's why I say it's sort of devolved into this weirdness.

Rodney Brooks (2:04:29.480)

Nevertheless, conversation not formulated as a test is a fascinatingly challenging dance.

Rodney Brooks (2:04:36.680)

That's a really hard problem.

Rodney Brooks (2:04:38.200)

To me, conversation, when non poses a test, is a more intuitive illustration how far away

Rodney Brooks (2:04:45.720)

we are from solving intelligence than computer vision.

Lex Fridman (2:04:48.760)

It's hard.

Rodney Brooks (2:04:49.960)

Computer vision is harder for me to pull apart.

Lex Fridman (2:04:53.000)

But with language, with conversation, you could see.

Rodney Brooks (2:04:55.400)

Because language is so human.

Lex Fridman (2:04:56.840)

It's so human.

Rodney Brooks (2:04:58.680)

We can so clearly see it.

Lex Fridman (2:05:04.280)

Shit, you mentioned something I was going to go off on.

Rodney Brooks (2:05:06.920)

OK.

Rodney Brooks (2:05:08.920)

I mean, I have to ask you, because you were the head of CSAIL, AI Lab, for a long time.

Rodney Brooks (2:05:17.560)

I don't know.

Lex Fridman (2:05:18.840)

To me, when I came to MIT, you were one of the greats at MIT.

Lex Fridman (2:05:22.840)

So what was that time like?

Lex Fridman (2:05:25.960)

And plus, you're friends with, but you knew Minsky and all the folks there, all the legendary

Rodney Brooks (2:05:34.760)

AI people of which you're one.

Lex Fridman (2:05:37.960)

So what was that time like?

Lex Fridman (2:05:39.560)

What are memories that stand out to you from that time, from your time at MIT, from the

Lex Fridman (2:05:46.760)

AI Lab, from the dreams that the AI Lab represented, to the actual revolutionary work?

Rodney Brooks (2:05:53.000)

Well, let me tell you first the disappointment in myself.

Rodney Brooks (2:05:56.760)

As I've been researching this book, and so many of the players were active in the 50s

Lex Fridman (2:06:03.960)

and 60s, I knew many of them when they were older, and I didn't ask them all the questions

Lex Fridman (2:06:08.600)

now I wish I had asked.

Rodney Brooks (2:06:11.320)

I'd sit with them at our Thursday lunches, which we had a faculty lunch, and I didn't

Lex Fridman (2:06:16.760)

ask them so many questions that now I wish I had.

Lex Fridman (2:06:19.720)

Can I ask you that question?

Lex Fridman (2:06:20.840)

Because you wrote that.

Rodney Brooks (2:06:22.440)

You wrote that you were fortunate to know and rub shoulders with many of the greats,

Lex Fridman (2:06:26.600)

those who founded AI, robotics, and computer science, and the World Wide Web.

Lex Fridman (2:06:30.680)

And you wrote that your big regret nowadays is that often I have questions for those who

Rodney Brooks (2:06:34.760)

have passed on, and I didn't think to ask them any of these questions, even as I saw

Rodney Brooks (2:06:41.560)

them and said hello to them on a daily basis.

Lex Fridman (2:06:44.120)

So maybe also another question I want to ask, if you could talk to them today, what question

Lex Fridman (2:06:51.160)

would you ask?

Lex Fridman (2:06:51.960)

What questions would you ask?

Rodney Brooks (2:06:53.240)

Well, Licklider, I would ask him.

Rodney Brooks (2:06:56.440)

You know, he had the vision for humans and computers working together, and he really

Rodney Brooks (2:07:02.600)

founded that at DARPA, and he gave the money to MIT, which started Project MAC in 1963.

Lex Fridman (2:07:12.360)

And I would have talked to him about what the successes were, what the failures were,

Lex Fridman (2:07:16.200)

what he saw as progress, etc.

Rodney Brooks (2:07:18.680)

I would have asked him more questions about that, because now I could use it in my book,

Rodney Brooks (2:07:24.680)

you know, but I think it's lost.

Lex Fridman (2:07:25.880)

It's lost forever.

Rodney Brooks (2:07:26.920)

A lot of the motivations are lost.

Rodney Brooks (2:07:33.240)

I should have asked Marvin why he and Seymour Pappert came down so hard on neural networks

Rodney Brooks (2:07:40.840)

in 1968 in their book Perceptrons, because Marvin's PhD thesis was all about neural networks.

Lex Fridman (2:07:48.440)

And how do you make sense of that?

Rodney Brooks (2:07:50.280)

That book destroyed the field.

Lex Fridman (2:07:52.040)

He probably, do you think he knew the effect that book would have?

Rodney Brooks (2:07:59.480)

All the theorems are negative theorems.

Lex Fridman (2:08:02.280)

Yeah.

Rodney Brooks (2:08:03.880)

Yeah.

Lex Fridman (2:08:04.920)

So, yeah.

Rodney Brooks (2:08:05.960)

That's just the way of, that's the way of life.

Lex Fridman (2:08:10.920)

But still, it's kind of tragic that he was both the proponent and the destroyer of neural

Rodney Brooks (2:08:15.800)

networks.

Lex Fridman (2:08:16.360)

Yeah.

Lex Fridman (2:08:19.160)

Is there other memories stand out from the robotics and the AI work at MIT?

Lex Fridman (2:08:28.120)

Well, yeah, but you gotta be more specific.

Rodney Brooks (2:08:31.320)

Well, I mean, like, it's such a magical place.

Rodney Brooks (2:08:33.160)

I mean, to me, it's a little bit also heartbreaking that, you know, with Google and Facebook,

Rodney Brooks (2:08:40.520)

like DeepMind and so on, so much of the talent, you know, it doesn't stay necessarily

Lex Fridman (2:08:46.280)

for prolonged periods of time in these universities.

Rodney Brooks (2:08:50.440)

Oh, yeah.

Rodney Brooks (2:08:50.940)

I mean, some of the companies are more guilty than others of paying fabulous salaries to

Rodney Brooks (2:08:57.800)

some of the highest, you know, producers.

Lex Fridman (2:09:00.120)

And then just, you never hear from them again.

Rodney Brooks (2:09:02.840)

They're not allowed to give public talks.

Lex Fridman (2:09:04.600)

They're sort of locked away.

Lex Fridman (2:09:06.600)

And it's sort of like collecting, you know, Hollywood stars or something.

Lex Fridman (2:09:12.280)

And they're not allowed to make movies anymore.

Rodney Brooks (2:09:13.960)

I own them.

Lex Fridman (2:09:14.460)

Yeah.

Rodney Brooks (2:09:15.660)

That's tragic because, I mean, there's an openness to the university setting where you

Rodney Brooks (2:09:20.700)

do research to both in the space of ideas and like publication, all those kinds of things.

Rodney Brooks (2:09:25.580)

Yeah, you know, and, you know, there's the publication and all that.

Lex Fridman (2:09:28.940)

And often, you know, although these places say they publish.

Rodney Brooks (2:09:32.940)

There's pressure.

Lex Fridman (2:09:33.660)

But I think, for instance, you know, on net net, I think Google buying those eight or

Rodney Brooks (2:09:41.260)

nine robotics company was bad for the field because it locked those people away.

Rodney Brooks (2:09:46.620)

They didn't have to make the company succeed anymore, locked them away for years, and then

Rodney Brooks (2:09:53.660)

sort of all frid it away.

Lex Fridman (2:09:55.660)

Yeah.

Lex Fridman (2:09:56.160)

So do you have hope for MIT, for MIT?

Lex Fridman (2:10:02.960)

Yeah.

Lex Fridman (2:10:03.460)

Why shouldn't I?

Rodney Brooks (2:10:04.560)

Well, I could be harsh and say that I'm not sure I would say MIT is leading the world

Rodney Brooks (2:10:11.200)

in AI or even Stanford or Berkeley.

Lex Fridman (2:10:15.440)

I would say, I would say DeepMind, Google AI, Facebook AI, all of those things.

Rodney Brooks (2:10:23.680)

I would take a slightly different approach, a different answer.

Lex Fridman (2:10:30.560)

I'll come back to Facebook in a minute.

Lex Fridman (2:10:32.880)

But I think those other places are following a dream of one of the founders.

Lex Fridman (2:10:42.880)

And I'm not sure that it's well founded, the dream.

Lex Fridman (2:10:46.560)

And I'm not sure that it's going to have the impact that he believes it is.

Lex Fridman (2:10:54.720)

You're talking about Facebook and Google and so on.

Rodney Brooks (2:10:56.560)

I'm talking about Google.

Lex Fridman (2:10:57.600)

Google.

Lex Fridman (2:10:58.320)

But the thing is, those research labs aren't, there's the big dream.

Lex Fridman (2:11:03.920)

And I'm usually a fan of no matter what the dream is, a big dream is a unifier.

Rodney Brooks (2:11:08.480)

Because what happens is you have a lot of bright minds working together on a dream.

Lex Fridman (2:11:15.200)

What results is a lot of adjacent ideas and how so much progress is made.

Rodney Brooks (2:11:20.000)

Yeah.

Lex Fridman (2:11:21.040)

So I'm not saying they're actually leading.

Rodney Brooks (2:11:22.560)

I'm not saying that the universities are leading.

Lex Fridman (2:11:25.280)

Yeah.

Lex Fridman (2:11:25.780)

But I don't think those companies are leading in general because they're,

Lex Fridman (2:11:28.960)

we saw this incredible spike in attendees at NeurIPS.

Lex Fridman (2:11:36.160)

And as I said in my January 1st review this year for 2020, 2020 will not be

Lex Fridman (2:11:44.800)

remembered as a watershed year for machine learning or AI.

Rodney Brooks (2:11:48.560)

There was nothing surprising happened anyway.

Lex Fridman (2:11:52.720)

Unlike when deep learning hit ImageNet.

Rodney Brooks (2:11:57.440)

That was a shake.

Lex Fridman (2:12:02.080)

And there's a lot more people writing papers, but the papers are fundamentally

Rodney Brooks (2:12:06.640)

boring and uninteresting.

Lex Fridman (2:12:08.800)

Incremental work.

Rodney Brooks (2:12:09.760)

Yeah.

Lex Fridman (2:12:10.260)

Is there a particular memories you have with Minsky or somebody else at

Lex Fridman (2:12:13.140)

MIT that stand out, funny stories?

Lex Fridman (2:12:16.340)

I mean, unfortunately, he's another one that's passed away.

Rodney Brooks (2:12:21.940)

You've known some of the biggest minds in AI.

Lex Fridman (2:12:24.580)

Yeah.

Lex Fridman (2:12:25.080)

And you know, they, they did amazing things and sometimes they were grumpy.

Lex Fridman (2:12:31.460)

Well, he was, uh, he was interesting cause he was very grumpy, but that,

Rodney Brooks (2:12:35.380)

that was his, uh, I remember him saying in an interview that the key to success

Rodney Brooks (2:12:41.780)

or being to keep being productive is to hate everything you've ever done in the past.

Rodney Brooks (2:12:45.940)

Maybe that, maybe that explains the Perceptron book.

Lex Fridman (2:12:49.940)

There it was.

Rodney Brooks (2:12:50.440)

He told you exactly.

Lex Fridman (2:12:53.540)

But he, meaning like, just like, I mean, maybe that's the way to not

Rodney Brooks (2:12:58.100)

treat yourself too seriously.

Lex Fridman (2:12:59.380)

Just, uh, you know, you're not, you're not, you're not, you're not, you're not,

Rodney Brooks (2:13:03.940)

you're not treating yourself too seriously.

Lex Fridman (2:13:05.620)

Just, uh, always be moving forward.

Rodney Brooks (2:13:09.220)

Uh, that was the idea.

Lex Fridman (2:13:10.100)

I mean, that, that crankiness, I mean, there's a, uh, that's the scary.

Lex Fridman (2:13:14.980)

So let me, let me, let me tell you, uh, you know, what really, um, you know,

Rodney Brooks (2:13:21.060)

the joy memories are about having access to technology before anyone else has seen

Rodney Brooks (2:13:27.460)

it.

Lex Fridman (2:13:27.960)

You know, I got to Stanford in 1977 and we had, um, you know, we had terminals

Rodney Brooks (2:13:34.860)

that could show live video on them.

Lex Fridman (2:13:37.260)

Um, digital, digital sound system.

Rodney Brooks (2:13:40.620)

We had a Xerox graphics printer.

Lex Fridman (2:13:45.020)

We could print, um, uh, it wasn't, you know, it wasn't like a typewriter

Rodney Brooks (2:13:50.140)

ball hitting in characters.

Lex Fridman (2:13:51.980)

It could print arbitrary things.

Rodney Brooks (2:13:53.580)

I mean, you know, one bit, you know, black or white, but you get arbitrary pictures.

Lex Fridman (2:13:58.300)

This was science fiction sort of stuff.

Rodney Brooks (2:14:00.380)

Um, um, at, at MIT, the, uh, the list machines, which, you know, they were the

Lex Fridman (2:14:07.260)

first personal computers and, you know, cost a hundred thousand dollars each.

Lex Fridman (2:14:12.060)

And I could, you know, I got there early enough in the day.

Lex Fridman (2:14:14.620)

I got one for the day.

Rodney Brooks (2:14:15.980)

Couldn't, couldn't stand up.

Lex Fridman (2:14:17.420)

I had to keep working.

Rodney Brooks (2:14:18.380)

Um, um, so they're having that like direct glimpse into the future.

Lex Fridman (2:14:25.340)

Yeah.

Rodney Brooks (2:14:25.580)

And, and, you know, I've had email every day since 1977.

Rodney Brooks (2:14:29.100)

Um, and, uh, you know, the, the host field was only eight bits, you know, that many

Rodney Brooks (2:14:36.060)

places, but I could send the email to other people at a few places.

Lex Fridman (2:14:39.980)

So that was, that was pretty exciting to be in that world so different from what

Rodney Brooks (2:14:45.340)

the rest of the world knew.

Lex Fridman (2:14:46.780)

Um, uh, uh, let me ask you probably edit this out, but just in case you have a

Rodney Brooks (2:14:53.420)

story, uh, I'm hanging out with Don Knuth, uh, for a while tomorrow.

Lex Fridman (2:15:00.060)

Did you ever get a chance to such a different world than yours?

Rodney Brooks (2:15:03.340)

He's a very kind of theoretical computer science, the puzzle of, uh, of, uh, computer

Lex Fridman (2:15:08.300)

science and mathematics.

Lex Fridman (2:15:09.500)

And you're so much about the magic of robotics, like the practice of it.

Lex Fridman (2:15:13.740)

You mentioned him earlier for like, not, you know, about computation.

Lex Fridman (2:15:17.820)

Did your worlds cross?

Lex Fridman (2:15:19.580)

They did enough.

Rodney Brooks (2:15:20.540)

You know, I, I know him now we talk, you know, but let me tell you my, my Donald

Lex Fridman (2:15:25.100)

Knuth story.

Rodney Brooks (2:15:26.700)

So, um, you know, besides, you know, analysis of algorithms, he's well known for

Lex Fridman (2:15:32.140)

writing tech, which is in LaTeX, which is the academic publishing system.

Lex Fridman (2:15:37.580)

So he did that at the AI lab and he would do it.

Lex Fridman (2:15:41.740)

He would work overnight at the AI lab.

Lex Fridman (2:15:45.020)

And one, one day, one night, the, uh, the mainframe computer went down and, um, uh,

Lex Fridman (2:15:55.660)

a guy named Robert Pore was there.

Rodney Brooks (2:15:57.340)

He did his PhD at the Media Lab at MIT and he was, um, you know, an engineer.

Lex Fridman (2:16:04.300)

And so I, he and I, you know, tracked down what were the problem was.

Rodney Brooks (2:16:08.780)

It was one of this big refrigerator size or washing machine size disk drives had

Lex Fridman (2:16:13.100)

failed.

Lex Fridman (2:16:13.500)

And that's what brought the whole system down.

Lex Fridman (2:16:15.500)

So we've got panels pulled off and we're pulling, you know, circuit cards out.

Lex Fridman (2:16:20.300)

And Donald Knuth, who's a really tall guy walks in and he's looking down and says,

Lex Fridman (2:16:25.340)

when will it be fixed?

Rodney Brooks (2:16:26.540)

You know, cause he wanted to get back to writing his tech system.

Lex Fridman (2:16:31.340)

And so we, we figured out, you know, it was a particular chip, 7,400 series chip,

Rodney Brooks (2:16:37.420)

which was socketed.

Lex Fridman (2:16:38.700)

We popped it out.

Rodney Brooks (2:16:40.780)

We put a replacement in, put it back in.

Lex Fridman (2:16:43.340)

Smoke comes out cause we put it in backwards.

Rodney Brooks (2:16:45.740)

Cause we were so nervous that Donald Knuth was standing over us.

Lex Fridman (2:16:49.500)

Anyway, we eventually got it fixed and got the mainframe running again.

Lex Fridman (2:16:53.660)

So that was your little, when was that again?

Lex Fridman (2:16:56.220)

Well, that must have been before October 79.

Rodney Brooks (2:16:58.860)

Cause we moved out of that building then.

Lex Fridman (2:17:00.300)

So sometime probably 78 sometime early 79.

Rodney Brooks (2:17:03.740)

Yeah, those, all those figures is just fascinating.

Lex Fridman (2:17:06.140)

All the people with pass, pass through MIT is really fascinating.

Rodney Brooks (2:17:10.220)

Is there, let me ask you to put on your big wise man hat.

Lex Fridman (2:17:18.140)

Is there advice that you can give to young people today,

Rodney Brooks (2:17:20.860)

whether in high school or college who are thinking about their career

Lex Fridman (2:17:24.700)

or thinking about life, how to live a life they're proud of, a successful life?

Rodney Brooks (2:17:32.060)

Yeah. So, so many people ask me for advice and have asked for,

Lex Fridman (2:17:36.140)

and I give, I talk to a lot of people all the time and there is no one way.

Rodney Brooks (2:17:44.060)

You know, there's a lot of pressure to produce papers

Lex Fridman (2:17:51.900)

that will be acceptable and be published.

Rodney Brooks (2:17:56.460)

Maybe I was, maybe I can't do it.

Lex Fridman (2:17:58.620)

Maybe I was, maybe I come from an age where I would,

Rodney Brooks (2:18:03.340)

I could be a rebel against that and still succeed.

Lex Fridman (2:18:07.100)

Maybe it's harder today, but I think it's important not to get too caught up

Rodney Brooks (2:18:14.860)

with what everyone else is doing.

Lex Fridman (2:18:18.380)

And if you, if, well, it depends on what you want of life.

Rodney Brooks (2:18:22.940)

If you want to have real impact, you have to be ready to fail a lot of times.

Lex Fridman (2:18:31.100)

So you have to make a lot of unsafe decisions.

Lex Fridman (2:18:34.220)

And the only way to make that work is to make, keep doing it for a long time.

Lex Fridman (2:18:38.700)

And then one of them will be work out.

Lex Fridman (2:18:40.220)

And so that, that, that will make something successful.

Lex Fridman (2:18:43.740)

Or not.

Rodney Brooks (2:18:45.500)

Or yeah, or you may, or you just may, you know, end up, you know,

Lex Fridman (2:18:48.780)

not having a, you know, having a lousy career.

Rodney Brooks (2:18:50.780)

I mean, it's certainly possible.

Lex Fridman (2:18:52.140)

Taking the risk is the thing.

Rodney Brooks (2:18:53.420)

Yeah.

Lex Fridman (2:18:56.220)

But there's no way to, to make all safe decisions and actually really contribute.

Lex Fridman (2:19:06.620)

Do you think about your death, about your mortality?

Lex Fridman (2:19:12.300)

I got to say when COVID hit, I did.

Rodney Brooks (2:19:15.660)

Because we did, you know, in the early days, we didn't know how bad it was going to be.

Lex Fridman (2:19:18.860)

And I, that, that made me work on my book harder for a while,

Lex Fridman (2:19:22.780)

but then I'd started this company and now I'm doing full time,

Lex Fridman (2:19:25.900)

more than full time of the company.

Lex Fridman (2:19:27.100)

So the book's on hold, but I do want to finish this book.

Lex Fridman (2:19:30.300)

When you think about it, are you afraid of it?

Rodney Brooks (2:19:35.820)

I'm afraid of dribbling, you know, of losing it.

Lex Fridman (2:19:42.220)

The details of, okay.

Rodney Brooks (2:19:43.980)

Yeah.

Lex Fridman (2:19:45.180)

Yeah.

Lex Fridman (2:19:45.580)

But the fact that the ride ends, I've known that for a long time.

Lex Fridman (2:19:51.260)

So it's, yeah, but there's knowing and knowing.

Rodney Brooks (2:19:55.420)

It's such a, yeah.

Lex Fridman (2:19:57.580)

And it really sucks.

Rodney Brooks (2:19:58.780)

It feels, it feels a lot closer.

Lex Fridman (2:20:01.820)

So my, in, in my, my blog with my predictions, my sort of push back against that was that I said,

Rodney Brooks (2:20:08.940)

I'm going to review these every year for 32 years and that puts me into my mid nineties.

Lex Fridman (2:20:14.860)

So, you know, it's my whole every, every time you write the blog posts,

Rodney Brooks (2:20:18.780)

you're getting closer and closer to your own prediction of your death.

Lex Fridman (2:20:23.660)

Yeah.

Lex Fridman (2:20:24.940)

What do you hope your legacy is?

Lex Fridman (2:20:28.140)

You're one of the greatest roboticist AI researchers of all time.

Lex Fridman (2:20:34.700)

What I hope is that I actually finished writing this book

Lex Fridman (2:20:38.140)

and that there's one person who reads it and see something about changing the way they're thinking.

Lex Fridman (2:20:48.940)

And that leads to the next big.

Lex Fridman (2:20:54.860)

And then there'll be on a podcast a hundred years from now saying I once read that book

Lex Fridman (2:21:01.580)

and that changed everything.

Lex Fridman (2:21:04.460)

What do you think is the meaning of life?

Rodney Brooks (2:21:06.140)

This whole thing, the existence, the, the, the, all the hurried things we do

Lex Fridman (2:21:10.140)

on this planet, what do you think is the meaning of it all?

Rodney Brooks (2:21:13.260)

Yeah. Well, you know, I think we're all really bad at it.

Lex Fridman (2:21:17.180)

Life or finding meaning or both.

Rodney Brooks (2:21:19.020)

Yeah. We get caught up in, in, in the, it's easy to get easier to do the stuff that's immediate

Lex Fridman (2:21:24.940)

and not through the stuff. It's not immediate.

Lex Fridman (2:21:27.820)

So the big picture we're bad at.

Lex Fridman (2:21:29.980)

Yeah. Yeah.

Lex Fridman (2:21:31.020)

Do you have a sense of what that big picture is?

Lex Fridman (2:21:33.900)

Like why you ever look up to the stars and ask, why the hell are we here?

Rodney Brooks (2:21:41.580)

You know, my, my, my, my atheism tells me it's just random, but you know, I want to understand the,

Rodney Brooks (2:21:50.380)

the way random in the, in the, that's what I talk about in this book, how order comes from disorder.

Rodney Brooks (2:21:55.660)

Yeah.

Lex Fridman (2:21:58.220)

But it kind of sprung up like most of the whole thing is random, but this, this, this,

Rodney Brooks (2:22:02.460)

the whole thing is random, but this little pocket of complexity they will call earth

Lex Fridman (2:22:07.660)

that like, why the hell does that happen?

Rodney Brooks (2:22:10.300)

And, and what we don't know is how common that those pockets of complexity are or how often,

Lex Fridman (2:22:18.060)

um, cause they may not last forever.

Rodney Brooks (2:22:22.780)

Which is, uh, more exciting slash sad to you if we're alone or if there's infinite number of.

Lex Fridman (2:22:30.460)

Oh, I think, I think it's impossible for me to believe that we're alone.

Rodney Brooks (2:22:36.300)

Um, that would just be too horrible, too cruel.

Lex Fridman (2:22:41.500)

It could be like the sad thing.

Rodney Brooks (2:22:43.180)

It could be like a graveyard of intelligent civilizations.

Lex Fridman (2:22:46.300)

Oh, everywhere.

Rodney Brooks (2:22:46.940)

Yeah.

Lex Fridman (2:22:47.980)

That might be the most likely outcome.

Lex Fridman (2:22:50.620)

And for us too.

Lex Fridman (2:22:51.660)

Yeah, exactly.

Rodney Brooks (2:22:52.540)

Yeah.

Lex Fridman (2:22:52.860)

And all of this will be forgotten.

Rodney Brooks (2:22:54.700)

Yeah.

Lex Fridman (2:22:54.940)

Yeah, including all the robots you build, everything forgotten.

Rodney Brooks (2:23:01.500)

Well, on average, everyone has been forgotten in history.

Lex Fridman (2:23:05.740)

Yeah.

Rodney Brooks (2:23:06.220)

Right.

Lex Fridman (2:23:06.940)

Yeah.

Rodney Brooks (2:23:07.500)

Most people are not remembered beyond the generation or two.

Lex Fridman (2:23:11.100)

Um, I mean, yeah.

Rodney Brooks (2:23:12.780)

Well, not just on average, basically very close to a hundred percent of people who've ever lived

Lex Fridman (2:23:17.900)

are forgotten.

Rodney Brooks (2:23:18.780)

Yeah.

Rodney Brooks (2:23:19.020)

I mean, you know, long arc of, I don't know anyone alive who remembers my great grandparents

Rodney Brooks (2:23:24.140)

because we didn't meet them.

Lex Fridman (2:23:26.300)

So still this fun, this, uh, this, uh, life is pretty fun somehow.

Rodney Brooks (2:23:32.460)

Yeah.

Lex Fridman (2:23:33.660)

Even the immense absurdity and, and, uh, at times, meaninglessness of it all.

Rodney Brooks (2:23:39.180)

It's pretty fun.

Lex Fridman (2:23:40.220)

And one of the, for me, one of the most fun things is robots.

Lex Fridman (2:23:43.740)

And I've looked up to your work.

Lex Fridman (2:23:45.180)

I've looked up to you for a long time.

Rodney Brooks (2:23:46.780)

That's right.

Lex Fridman (2:23:47.180)

God.

Rodney Brooks (2:23:47.740)

Rod, it's, it's an honor that, uh, you would spend your valuable time with me today talking.

Lex Fridman (2:23:53.580)

It was an amazing conversation.

Rodney Brooks (2:23:54.780)

Thank you so much for being here.

Lex Fridman (2:23:55.980)

Well, thanks for, thanks for talking with me.

Rodney Brooks (2:23:57.820)

I've enjoyed it.

Lex Fridman (2:24:00.060)

Thanks for listening to this conversation with Rodney Brooks.

Rodney Brooks (2:24:02.700)

To support this podcast, please check out our sponsors in the description.

Lex Fridman (2:24:06.860)

And now let me leave you with the three laws of robotics from Isaac Asimov.

Rodney Brooks (2:24:12.620)

One, a robot may not injure a human being or through inaction, allow human being to come to

Rodney Brooks (2:24:19.020)

harm. Two, a robot must obey the orders given to it by human beings, except when such orders

Rodney Brooks (2:24:25.580)

would conflict with the first law. And three, a robot must protect its own existence as long

Lex Fridman (2:24:32.860)

as such protection does not conflict with the first or the second laws.

Rodney Brooks (2:24:38.620)

Thank you for listening.

Lex Fridman (2:24:39.740)

I hope to see you next time.

Rodney Brooks (30:00.880)

Yeah, I mean, I don't know how much truth there is to, like chess, for example, was

Lex Fridman (30:08.480)

for the longest time seen as the highest level of intellect, right?

Rodney Brooks (30:14.720)

Until we got computers that were better at it than people.

Lex Fridman (30:17.200)

And then we realized, you know, if you go back to the 90s, you'll see, you know, the

Rodney Brooks (30:21.120)

stories in the press around when Kasparov was beaten by Deep Blue.

Lex Fridman (30:26.320)

Oh, this is the end of all sorts of things.

Rodney Brooks (30:28.320)

Computers are going to be able to do anything from now on.

Lex Fridman (30:30.640)

And we saw exactly the same stories with Alpha Zero, the Go Playing program.

Rodney Brooks (30:36.160)

Yeah.

Lex Fridman (30:37.280)

But still, to me, reasoning is a special thing.

Lex Fridman (30:41.200)

And perhaps...

Lex Fridman (30:41.920)

No, actually, we're really bad at reasoning.

Rodney Brooks (30:44.640)

We just use these analogies based on our hunter gatherer intuitions.

Lex Fridman (30:48.400)

But why is that not, don't you think the ability to construct metaphor is a really powerful

Lex Fridman (30:53.520)

thing?

Lex Fridman (30:53.920)

Oh, yeah, it is.

Rodney Brooks (30:54.400)

Tell stories.

Lex Fridman (30:55.200)

It is.

Rodney Brooks (30:55.520)

It's the constructing the metaphor and registering that something constant in our brains.

Lex Fridman (31:00.960)

Like, isn't that what we're doing with vision too?

Lex Fridman (31:04.000)

And we're telling our stories.

Lex Fridman (31:06.080)

We're constructing good models of the world.

Rodney Brooks (31:08.560)

Yeah, yeah.

Lex Fridman (31:09.760)

But I think we jumped between what we're capable of and how we're doing it right there.

Rodney Brooks (31:16.400)

It was a little confusion that went on as we were telling each other stories.

Lex Fridman (31:21.680)

Yes, exactly.

Rodney Brooks (31:23.440)

Trying to delude each other.

Lex Fridman (31:24.800)

No, I just think I'm not exactly so.

Rodney Brooks (31:27.280)

I'm trying to pull apart this Moravec's paradox.

Lex Fridman (31:30.160)

I don't view it as a paradox.

Lex Fridman (31:33.280)

What did evolution spend its time on?

Lex Fridman (31:36.000)

Yes.

Rodney Brooks (31:36.320)

It spent its time on getting us to perceive and move in the world.

Lex Fridman (31:39.360)

That was 600 million years as multi cell creatures doing that.

Lex Fridman (31:43.600)

And then it was relatively recent that we were able to hunt or gather or even animals hunting.

Lex Fridman (31:53.120)

That's much more recent.

Lex Fridman (31:54.960)

And then anything that we, speech, language, those things are a couple of hundred thousand

Lex Fridman (32:02.960)

years probably, if that long.

Lex Fridman (32:05.760)

And then agriculture, 10,000 years.

Rodney Brooks (32:09.520)

All that stuff was built on top of those earlier things, which took a long time to develop.

Lex Fridman (32:14.320)

So if you then look at the engineering of these things, so building it into robots,

Lex Fridman (32:20.160)

what's the hardest part of robotics?

Lex Fridman (32:22.000)

Do you think as the decades that you worked on robots in the context of what we're talking

Rodney Brooks (32:29.920)

about, vision, perception, the actual sort of the biomechanics of movement, I'm kind

Rodney Brooks (32:37.520)

of drawing parallels here between humans and machines always.

Lex Fridman (32:40.800)

Like what do you think is the hardest part of robotics?

Rodney Brooks (32:44.320)

I just want to think all of them.

Lex Fridman (32:45.920)

I just want to think all of them.

Rodney Brooks (32:49.360)

There are no easy parts to do well.

Lex Fridman (32:53.040)

We sort of go reductionist and we reduce it.

Rodney Brooks (32:55.600)

If only we had all the location of all the points in 3D, things would be great.

Lex Fridman (33:02.400)

If only we had labels on the images, things would be great.

Lex Fridman (33:07.440)

But as we see, that's not good enough.

Lex Fridman (33:10.640)

Some deeper understanding.

Lex Fridman (33:13.040)

But if I came to you and I could solve one category of problems in robotics instantly,

Lex Fridman (33:21.680)

what would give you the greatest pleasure?

Lex Fridman (33:28.160)

I mean, you look at robots that manipulate objects, what's hard about that?

Rodney Brooks (33:36.400)

You know, is it the perception, is it the reasoning about the world, that common sense

Lex Fridman (33:43.040)

reasoning, is it the actual building a robot that's able to interact with the world?

Rodney Brooks (33:49.680)

Is it like human aspects of a robot that's interacting with humans in that game theory

Lex Fridman (33:54.960)

of how they work well together?

Rodney Brooks (33:56.080)

Well, let's talk about manipulation for a second because I had this really blinding

Rodney Brooks (34:00.000)

moment, you know, I'm a grandfather, so grandfathers have blinding moments.

Rodney Brooks (34:05.360)

Just three or four miles from here, last year, my 16 month old grandson was in his new house

Lex Fridman (34:16.240)

for the first time, right?

Lex Fridman (34:18.240)

First time in this house.

Lex Fridman (34:19.760)

And he'd never been able to get to a window before, but this had some low windows.

Lex Fridman (34:25.040)

And he goes up to this window with a handle on it that he's never seen before.

Lex Fridman (34:29.360)

And he's got one hand pushing the window and the other hand turning the handle to open

Lex Fridman (34:34.800)

the window.

Rodney Brooks (34:36.640)

He knew two different hands, two different things he knew how to put together.

Lex Fridman (34:44.080)

And he's 16 months old.

Lex Fridman (34:45.520)

And there you are watching in awe.

Lex Fridman (34:51.840)

In an environment he'd never seen before, a mechanism he'd never seen.

Lex Fridman (34:55.200)

How did he do that?

Lex Fridman (34:56.320)

Yes, that's a good question.

Lex Fridman (34:57.600)

How did he do that?

Lex Fridman (34:58.880)

That's why.

Rodney Brooks (34:59.380)

It's like, okay, like you could see the leap of genius from using one hand to perform a

Rodney Brooks (35:05.700)

task to combining, doing, I mean, first of all, in manipulation, that's really difficult.

Rodney Brooks (35:11.460)

It's like two hands, both necessary to complete the action.

Lex Fridman (35:15.940)

And completely different.

Lex Fridman (35:16.820)

And he'd never seen a window open before, but he inferred somehow handle open something.

Rodney Brooks (35:25.140)

Yeah, there may have been a lot of slightly different failure cases that you didn't see.

Rodney Brooks (35:32.180)

Not with a window, but with other objects of turning and twisting and handles.

Lex Fridman (35:37.540)

There's a great counter to reinforcement learning.

Rodney Brooks (35:42.900)

We'll just give the robot plenty of time to try everything.

Lex Fridman (35:50.260)

Can I tell a little side story here?

Rodney Brooks (35:52.260)

Yeah, so I'm in DeepMind in London, this is three, four years ago, where there's a big

Rodney Brooks (36:01.940)

Google building, and then you go inside and you go through this more security, and then

Rodney Brooks (36:06.020)

you get to DeepMind where the other Google employees can't go.

Lex Fridman (36:09.060)

And I'm in a conference room, a conference room with some of the people, and they tell

Rodney Brooks (36:15.540)

me about their reinforcement learning experiment with robots, which are just trying stuff out.

Lex Fridman (36:23.940)

And they're my robots.

Rodney Brooks (36:25.380)

They're Sawyer's.

Lex Fridman (36:26.900)

We sold them.

Lex Fridman (36:29.060)

And they really like them because Sawyer's are compliant and can sense forces, so they

Lex Fridman (36:33.300)

don't break when they're bashing into walls.

Rodney Brooks (36:36.180)

They stop and they do all this stuff.

Lex Fridman (36:38.980)

So you just let the robot do stuff, and eventually it figures stuff out.

Rodney Brooks (36:42.580)

By the way, Sawyer, we're talking about robot manipulation, so robot arms and so on.

Lex Fridman (36:47.380)

Yeah, Sawyer's a robot.

Lex Fridman (36:50.180)

What's Sawyer?

Lex Fridman (36:51.220)

Sawyer's a robot arm that my company Rethink Robotics built.

Rodney Brooks (36:55.140)

Thank you for the context.

Lex Fridman (36:56.580)

Sorry.

Rodney Brooks (36:57.060)

Okay, cool.

Lex Fridman (36:57.540)

So we're in DeepMind.

Lex Fridman (36:59.380)

And it's in the next room, these robots are just bashing around to try and use reinforcement

Lex Fridman (37:04.100)

learning to learn how to act.

Lex Fridman (37:05.940)

Can I go see them?

Lex Fridman (37:06.740)

Oh no, they're secret.

Rodney Brooks (37:08.340)

They were my robots.

Lex Fridman (37:09.300)

They were secret.

Rodney Brooks (37:10.820)

That's hilarious.

Lex Fridman (37:11.700)

Okay.

Rodney Brooks (37:12.100)

Anyway, the point is, you know, this idea that you just let reinforcement learning figure

Lex Fridman (37:17.860)

everything out is so counter to how a kid does stuff.

Lex Fridman (37:21.780)

So again, story about my grandson.

Lex Fridman (37:24.740)

I gave him this box that had lots of different lock mechanisms.

Rodney Brooks (37:29.780)

He didn't randomly, you know, and he was 18 months old, he didn't randomly try to touch

Lex Fridman (37:34.260)

every surface or push everything.

Rodney Brooks (37:35.940)

He found he could see where the mechanism was, and he started exploring the mechanism

Lex Fridman (37:42.020)

for each of these different lock mechanisms.

Lex Fridman (37:44.580)

And there was reinforcement, no doubt, of some sort going on there.

Lex Fridman (37:48.660)

But he applied a pre filter, which cut down the search space dramatically.

Rodney Brooks (37:55.540)

I wonder to what level we're able to introspect what's going on.

Rodney Brooks (37:59.700)

Because what's also possible is you have something like reinforcement learning going

Rodney Brooks (38:03.780)

on in the mind in the space of imagination.

Lex Fridman (38:05.860)

So like you have a good model of the world you're predicting and you may be running those

Rodney Brooks (38:10.900)

tens of thousands of like loops, but you're like, as a human, you're just looking at yourself

Lex Fridman (38:16.820)

trying to tell a story of what happened.

Lex Fridman (38:18.740)

And it might seem simple, but maybe there's a lot of computation going on.

Lex Fridman (38:24.500)

Whatever it is, but there's also a mechanism that's being built up.

Rodney Brooks (38:28.020)

It's not just random search.

Lex Fridman (38:30.420)

Yeah, that mechanism prunes it dramatically.

Rodney Brooks (38:33.780)

Yeah, that pruning, that pruning stuff, but it doesn't, it's possible that that's, so

Rodney Brooks (38:40.980)

you don't think that's akin to a neural network inside a reinforcement learning algorithm.

Lex Fridman (38:46.740)

Is it possible?

Lex Fridman (38:49.140)

It's, yeah, until it's possible.

Rodney Brooks (38:52.340)

It's possible, but I'll be incredibly surprised if that happens.

Rodney Brooks (39:01.380)

I'll also be incredibly surprised that after all the decades that I've been doing this,

Rodney Brooks (39:06.020)

where every few years someone thinks, now we've got it.

Lex Fridman (39:10.100)

Now we've got it.

Rodney Brooks (39:12.580)

Four or five years ago, I was saying, I don't think we've got it yet.

Lex Fridman (39:15.620)

And everyone was saying, you don't understand how powerful AI is.

Rodney Brooks (39:18.820)

I had people tell me, you don't understand how powerful it is.

Rodney Brooks (39:22.420)

I sort of had a track record of what the world had done to think, well, this is no different

Rodney Brooks (39:30.420)

from before.

Lex Fridman (39:31.460)

Or we have bigger computers.

Rodney Brooks (39:33.060)

We had bigger computers in the 90s and we could do more stuff.

Lex Fridman (39:37.940)

But okay, so let me push back because I'm generally sort of optimistic and try to find

Rodney Brooks (39:43.380)

the beauty in things.

Rodney Brooks (39:44.260)

I think there's a lot of surprising and beautiful things that neural networks, this new generation

Rodney Brooks (39:51.860)

of deep learning revolution has revealed to me, has continually been very surprising

Lex Fridman (39:57.460)

the kind of things it's able to do.

Rodney Brooks (39:59.300)

Now, generalizing that over saying like this, we've solved intelligence.

Lex Fridman (40:03.140)

That's another big leap.

Lex Fridman (40:05.220)

But is there something surprising and beautiful to you about neural networks that were actually

Lex Fridman (40:10.500)

you said back and said, I did not expect this?

Rodney Brooks (40:16.100)

Oh, I think their performance on ImageNet was shocking.

Lex Fridman (40:22.260)

The computer vision in those early days was just very like, wow, okay.

Rodney Brooks (40:26.340)

That doesn't mean that they're solving everything in computer vision we need to solve or in

Lex Fridman (40:32.500)

vision for robots.

Lex Fridman (40:33.700)

What about AlphaZero and self play mechanisms and reinforcement learning?

Lex Fridman (40:37.220)

Yeah, that was all in the 90s.

Rodney Brooks (40:39.300)

Yeah, that was all in Donald Mickey's 1961 paper.

Lex Fridman (40:44.020)

Everything that was there, which introduced reinforcement learning.

Rodney Brooks (40:48.340)

No, but come on.

Lex Fridman (40:49.300)

So no, you're talking about the actual techniques.

Lex Fridman (40:52.020)

But isn't it surprising to you the level it's able to achieve with no human supervision

Lex Fridman (40:58.740)

of chess play?

Rodney Brooks (40:59.700)

Like, to me, there's a big, big difference between Deep Blue and...

Lex Fridman (41:05.860)

Maybe what that's saying is how overblown our view of ourselves is.

Rodney Brooks (41:13.140)

You know, the chess is easy.

Rodney Brooks (41:16.740)

Yeah, I mean, I came across this 1946 report that, and I'd seen this as a kid in one of

Rodney Brooks (41:28.340)

those books that my mother had given me actually.

Rodney Brooks (41:30.340)

The 1946 report, which pitted someone with an abacus against an electronic calculator,

Lex Fridman (41:39.620)

and he beat the electronic calculator.

Rodney Brooks (41:42.500)

You know, so there at that point was, well, humans are still better than machines at calculating.

Rodney Brooks (41:48.980)

Are you surprised today that a machine can, you know, do a billion floating point operations

Lex Fridman (41:54.420)

a second and, you know, you're puzzling for minutes through one?

Rodney Brooks (41:58.500)

I mean, I don't know, but I am certainly surprised there's something, to me, different about

Lex Fridman (42:07.460)

learning, so a system that's able to learn.

Rodney Brooks (42:10.420)

Learning.

Lex Fridman (42:10.980)

See, now you're getting into one of the deadly sins.

Rodney Brooks (42:15.300)

Because of using terms overly broadly.

Lex Fridman (42:19.220)

Yeah, I mean, there's so many different forms of learning.

Rodney Brooks (42:21.700)

Yeah.

Lex Fridman (42:22.260)

So many different forms.

Rodney Brooks (42:23.300)

You know, I learned my way around the city.

Lex Fridman (42:24.980)

I learned to play chess.

Rodney Brooks (42:26.500)

I learned Latin.

Lex Fridman (42:28.580)

I learned to ride a bicycle.

Rodney Brooks (42:30.100)

All of those are, you know, very different capabilities.

Lex Fridman (42:33.700)

Yeah.

Lex Fridman (42:34.180)

And if someone, you know, has a, you know, in the old days, people would write a paper

Lex Fridman (42:41.860)

about learning something.

Rodney Brooks (42:43.220)

Now the corporate press office puts out a press release about how Company X is leading

Lex Fridman (42:52.580)

the world because they have a system that can...

Rodney Brooks (42:56.820)

Yeah, but here's the thing.

Lex Fridman (42:58.180)

Okay.

Lex Fridman (42:58.500)

So what is learning?

Lex Fridman (43:00.100)

When I refer to...

Rodney Brooks (43:00.820)

Learning is many things.

Lex Fridman (43:02.420)

But...

Rodney Brooks (43:02.580)

It's a suitcase word.

Rodney Brooks (43:04.660)

It's a suitcase word, but loosely, there's a dumb system, and over time, it becomes smart.

Rodney Brooks (43:13.700)

Well, it becomes less dumb at the thing that it's doing.

Lex Fridman (43:16.340)

Smart is a loaded word.

Rodney Brooks (43:19.140)

Yes, less dumb at the thing it's doing.

Rodney Brooks (43:21.220)

It gets better performance under some measure, under some set of conditions at that thing.

Lex Fridman (43:27.060)

And most of these learning algorithms, learning systems, fail when you change the conditions

Lex Fridman (43:35.780)

just a little bit in a way that humans don't.

Lex Fridman (43:37.940)

So I was at DeepMind, the AlphaGo had just come out, and I said, what would have happened

Lex Fridman (43:45.940)

if you'd given it a 21 by 21 board instead of a 19 by 19 board?

Rodney Brooks (43:49.940)

They said, fail totally.

Lex Fridman (43:51.620)

But a human player would actually be able to play.

Lex Fridman (43:55.540)

And actually, funny enough, if you look at DeepMind's work since then, they're presenting

Lex Fridman (44:02.980)

a lot of algorithms that would do well at the bigger board.

Lex Fridman (44:07.620)

So they're slowly expanding this generalization.

Lex Fridman (44:10.340)

I mean, to me, there's a core element there.

Rodney Brooks (44:12.580)

I think it is very surprising to me that even in a constrained game of chess or Go, that

Rodney Brooks (44:20.100)

through self play, by a system playing itself, that it can achieve superhuman level performance

Rodney Brooks (44:28.580)

through learning alone.

Lex Fridman (44:29.940)

Okay, so you didn't like it when I referred to Donald Mickey's 1961 paper.

Rodney Brooks (44:38.980)

There, in the second part of it, which came a year later, they had self play on an electronic

Rodney Brooks (44:46.020)

computer at tic tac toe, okay, but it learned to play tic tac toe through self play.

Lex Fridman (44:52.180)

And it learned to play optimally.

Lex Fridman (44:54.580)

What I'm saying is, okay, I have a little bit of a bias, but I find ideas beautiful,

Lex Fridman (45:02.740)

but only when they actually realize the promise.

Lex Fridman (45:06.660)

That's another level of beauty.

Rodney Brooks (45:08.420)

For example, what Bezos and Elon Musk are doing with rockets.

Rodney Brooks (45:13.540)

We had rockets for a long time, but doing reusable cheap rockets, it's very impressive.

Rodney Brooks (45:18.900)

In the same way, I would have not predicted.

Rodney Brooks (45:22.980)

First of all, when I started and fell in love with AI, the game of Go was seen to be impossible

Rodney Brooks (45:30.820)

to solve.

Rodney Brooks (45:31.300)

Okay, so I thought maybe, you know, maybe it'd be possible to maybe have big leaps in

Rodney Brooks (45:38.500)

a Moore's law style of way, in computation, I'll be able to solve it.

Lex Fridman (45:42.020)

But I would never have guessed that you can learn your way, however, I mean, in the narrow

Rodney Brooks (45:50.500)

sense of learning, learn your way to beat the best people in the world at the game of

Lex Fridman (45:55.620)

Go without human supervision, not studying the game of experts.

Rodney Brooks (45:59.300)

Okay, so using a different learning technique, Arthur Samuel in the early 60s, and he was

Rodney Brooks (46:08.900)

the first person to use machine learning, had a program that could beat the world champion

Rodney Brooks (46:14.900)

at checkers.

Lex Fridman (46:16.100)

And that at the time was considered amazing.

Rodney Brooks (46:19.860)

By the way, Arthur Samuel had some fantastic advantages.

Lex Fridman (46:23.460)

Do you want to hear Arthur Samuel's advantages?

Rodney Brooks (46:25.700)

Two things.

Lex Fridman (46:26.660)

One, he was at the 1956 AI conference.

Rodney Brooks (46:30.500)

I knew Arthur later in life.

Lex Fridman (46:32.420)

He was at Stanford when I was a graduate student there.

Rodney Brooks (46:34.500)

He wore a tie and a jacket every day, the rest of us didn't.

Lex Fridman (46:38.900)

Delightful man, delightful man.

Rodney Brooks (46:42.980)

It turns out Claude Shannon, in a 1950 Scientific American article, on chess playing, outlined

Lex Fridman (46:51.620)

the learning mechanism that Arthur Samuel used, and they had met in 1956.

Rodney Brooks (46:57.140)

I assume there was some communication, but I don't know that for sure.

Lex Fridman (47:00.580)

But Arthur Samuel had been a vacuum tube engineer, getting reliability of vacuum tubes, and then

Rodney Brooks (47:07.060)

had overseen the first transistorized computers at IBM.

Lex Fridman (47:11.860)

And in those days, before you shipped a computer, you ran it for a week to get early failures.

Lex Fridman (47:18.180)

So he had this whole farm of computers running random code for hours and hours for each computer.

Lex Fridman (47:28.580)

He had a whole bunch of them.

Lex Fridman (47:29.940)

So he ran his chess learning program with self play on IBM's production line.

Rodney Brooks (47:38.820)

He had more computation available to him than anyone else in the world, and then he was

Rodney Brooks (47:43.700)

able to produce a chess playing program, I mean a checkers playing program, that could

Lex Fridman (47:48.260)

beat the world champion.

Lex Fridman (47:49.940)

So that's amazing.

Rodney Brooks (47:51.540)

The question is, what I mean surprised, I don't just mean it's nice to have that accomplishment,

Rodney Brooks (47:58.020)

is there is a stepping towards something that feels more intelligent than before.

Lex Fridman (48:06.180)

Yeah, but that's in your view of the world.

Rodney Brooks (48:08.740)

Okay, well let me then, it doesn't mean I'm wrong.

Lex Fridman (48:11.380)

No, no it doesn't.

Lex Fridman (48:13.540)

So the question is, if we keep taking steps like that, how far that takes us?

Lex Fridman (48:18.740)

Are we going to build a better recommender systems?

Lex Fridman (48:21.780)

Are we going to build a better robot?

Lex Fridman (48:23.860)

Or will we solve intelligence?

Rodney Brooks (48:25.940)

So, you know, I'm putting my bet on, but still missing a whole lot.

Lex Fridman (48:33.300)

A lot.

Lex Fridman (48:34.500)

And why would I say that?

Rodney Brooks (48:36.020)

Well, in these games, they're all, you know, 100% information games, but again, but each

Rodney Brooks (48:43.060)

of these systems is a very short description of the current state, which is different from

Lex Fridman (48:50.420)

registering and perception in the world, which gets back to Marovec's paradox.

Rodney Brooks (48:55.620)

I'm definitely not saying that chess is somehow harder than perception or any kind of, even

Rodney Brooks (49:05.780)

any kind of robotics in the physical world, I definitely think is way harder than the

Rodney Brooks (49:10.180)

game of chess.

Lex Fridman (49:10.820)

So I was always much more impressed by the workings of the human mind.

Rodney Brooks (49:15.300)

It's incredible.

Lex Fridman (49:15.940)

The human mind is incredible.

Rodney Brooks (49:17.700)

I believe that from the very beginning, I wanted to be a psychiatrist for the longest

Lex Fridman (49:20.340)

time.

Rodney Brooks (49:20.740)

I always thought that's way more incredible in the game of chess.

Lex Fridman (49:23.140)

I think the game of chess is, I love the Olympics.

Rodney Brooks (49:26.740)

It's just another example of us humans picking a task and then agreeing that a million humans

Lex Fridman (49:31.860)

will dedicate their whole life to that task.

Lex Fridman (49:33.860)

And that's the cool thing that the human mind is able to focus on one task and then compete

Lex Fridman (49:39.860)

against each other and achieve like weirdly incredible levels of performance.

Rodney Brooks (49:44.500)

That's the aspect of chess that's super cool.

Lex Fridman (49:46.740)

Not that chess in itself is really difficult.

Rodney Brooks (49:49.700)

It's like the Fermat's last theorem is not in itself to me that interesting.

Rodney Brooks (49:53.460)

The fact that thousands of people have been struggling to solve that particular problem

Rodney Brooks (49:57.780)

is fascinating.

Lex Fridman (49:58.500)

So can I tell you my disease in this way?

Rodney Brooks (50:00.500)

Sure.

Lex Fridman (50:01.460)

Which actually is closer to what you're saying.

Lex Fridman (50:03.380)

So as a child, I was building various, I called them computers.

Lex Fridman (50:07.620)

They weren't general purpose computers.

Rodney Brooks (50:09.380)

Ice cube tray.

Lex Fridman (50:10.180)

The ice cube tray was one.

Lex Fridman (50:11.380)

But I built other machines.

Lex Fridman (50:12.660)

And what I liked to build was machines that could beat adults at a game and the adults

Rodney Brooks (50:18.100)

couldn't beat my machine.

Lex Fridman (50:19.700)

Yeah.

Lex Fridman (50:19.940)

So you were like, that's powerful.

Lex Fridman (50:22.660)

That's a way to rebel.

Lex Fridman (50:24.820)

Oh, by the way, when was the first time you built something that outperformed you?

Lex Fridman (50:33.220)

Do you remember?

Rodney Brooks (50:34.660)

Well, I knew how it worked.

Rodney Brooks (50:36.340)

I was probably nine years old and I built a thing that was a game where you take turns

Rodney Brooks (50:42.020)

in taking matches from a pile and either the one who takes the last one or the one who

Lex Fridman (50:47.460)

doesn't take the last one wins.

Rodney Brooks (50:48.660)

I forget.

Lex Fridman (50:49.460)

And so it was pretty easy to build that out of wires and nails and little coils that were

Rodney Brooks (50:54.500)

like plugging in the number and a few light bulbs.

Rodney Brooks (50:59.060)

The one I was proud of, I was 12 when I built a thing out of old telephone switchboard switches

Rodney Brooks (51:07.220)

that could always win at tic tac toe.

Lex Fridman (51:11.380)

And that was a much harder circuit to design.

Lex Fridman (51:14.500)

But again, it was no active components.

Lex Fridman (51:17.620)

It was just three position switches, empty, X, zero, O.

Lex Fridman (51:23.300)

And nine of them and a light bulb on which move it wanted next.

Lex Fridman (51:29.460)

And then the human would go and move that.

Rodney Brooks (51:31.540)

See, there's magic in that creation.

Lex Fridman (51:33.060)

There was.

Rodney Brooks (51:33.860)

Yeah, yeah.

Rodney Brooks (51:34.580)

I tend to see magic in robots that like I also think that intelligence is a little bit

Rodney Brooks (51:43.700)

overrated.

Lex Fridman (51:44.740)

I think we can have deep connections with robots very soon.

Lex Fridman (51:49.140)

And well, we'll come back to connections for sure.

Lex Fridman (51:52.500)

But I do want to say, I think too many people make the mistake of seeing that magic and

Rodney Brooks (52:00.100)

thinking, well, we'll just continue.

Lex Fridman (52:02.820)

But each one of those is a hard fought battle for the next step, the next step.

Rodney Brooks (52:07.300)

Yes.

Rodney Brooks (52:08.180)

The open question here is, and this is why I'm playing devil's advocate, but I often

Rodney Brooks (52:11.940)

do when I read your blog post in my mind because I have like this eternal optimism, is it's

Lex Fridman (52:18.420)

not clear to me.

Lex Fridman (52:19.380)

So I don't do what obviously the journalists do or they give into the hype, but it's not

Rodney Brooks (52:23.940)

obvious to me how many steps away we are from a truly transformational understanding of

Lex Fridman (52:34.740)

what it means to build intelligent systems or how to build intelligent systems.

Rodney Brooks (52:40.580)

I'm also aware of the whole history of artificial intelligence, which is where your deep grounding

Rodney Brooks (52:45.140)

of this is, is there has been an optimism for decades and that optimism, just like reading

Rodney Brooks (52:51.860)

old optimism is absurd because people were like, this is, they were saying things are

Rodney Brooks (52:57.300)

trivial for decades since the sixties, they're saying everything is true.

Rodney Brooks (53:00.740)

Computer vision is trivial, but I think my mind is working crisply enough to where, I

Rodney Brooks (53:07.700)

mean, we can dig into if you want.

Lex Fridman (53:09.700)

I'm really surprised by the things DeepMind has done.

Rodney Brooks (53:12.900)

I don't think they're so, they're yet close to solving intelligence, but I'm not sure

Lex Fridman (53:19.300)

it's not 10 to 10 years away.

Lex Fridman (53:22.500)

What I'm referring to is interesting to see when the engineering, it takes that idea to

Lex Fridman (53:30.100)

scale and the idea works.

Lex Fridman (53:32.660)

And no, it fools people.

Lex Fridman (53:34.900)

Okay.

Rodney Brooks (53:35.300)

Honestly, Rodney, if it was you, me and Demis inside a room, forget the press, forget all

Rodney Brooks (53:40.420)

those things, just as a scientist, as a roboticist, that wasn't surprising to you that at scale.

Lex Fridman (53:47.060)

So we're talking about very large now, okay, let's pick one.

Lex Fridman (53:50.180)

That's the most surprising to you.

Rodney Brooks (53:52.340)

Okay.

Lex Fridman (53:52.820)

Please don't yell at me.

Rodney Brooks (53:53.940)

GPT three, okay.

Rodney Brooks (53:56.180)

Hold on, hold on, I was going to say, okay, alpha zero, alpha go, alpha go, zero, alpha

Rodney Brooks (54:03.300)

zero, and then alpha fold one and two.

Lex Fridman (54:06.340)

So do any of these kind of have this core of, forget usefulness or application and so

Rodney Brooks (54:13.460)

on, which you could argue for alpha fold, like, as a scientist, was those surprising

Lex Fridman (54:19.220)

to you that it worked as well as it did?

Rodney Brooks (54:23.140)

Okay, so if we're going to make the distinction between surprise and usefulness, and I have

Rodney Brooks (54:30.820)

to explain this, I would say alpha fold, and one of the problems at the moment with alpha

Rodney Brooks (54:40.580)

fold is, you know, it gets a lot of them right, which is a surprise to me, because they're

Rodney Brooks (54:44.820)

a really complex thing, but you don't know which ones it gets right, which then is a

Rodney Brooks (54:51.940)

bit of a problem.

Lex Fridman (54:52.500)

Now they've come out with a recent...

Rodney Brooks (54:53.620)

You mean the structure of the proteins, it gets a lot of those right.

Lex Fridman (54:56.180)

Yeah, it's a surprising number of them right, it's been a really hard problem.

Lex Fridman (55:00.180)

So that was a surprise how many it gets right.

Lex Fridman (55:03.220)

So far, the usefulness is limited, because you don't know which ones are right or not,

Lex Fridman (55:07.460)

and now they've come out with a thing in the last few weeks, which is trying to get a useful

Lex Fridman (55:12.900)

tool out of it, and they may well do it.

Rodney Brooks (55:15.620)

In that sense, at least alpha fold is different, because your alpha fold tool is different,

Rodney Brooks (55:21.940)

because now it's producing data sets that are actually, you know, potentially revolutionizing

Rodney Brooks (55:27.460)

competition biology, like they will actually help a lot of people, but...

Lex Fridman (55:31.620)

You would say potentially revolutionizing, we don't know yet, but yeah.

Rodney Brooks (55:36.020)

That's true, yeah.

Lex Fridman (55:36.820)

But they're, you know, but I got you.

Rodney Brooks (55:39.220)

I mean, this is...

Lex Fridman (55:40.020)

Okay, so you know what, this is gonna be so fun, so let's go right into it.

Rodney Brooks (55:45.860)

Speaking of robots that operate in the real world, let's talk about self driving cars.

Lex Fridman (55:52.740)

Oh, okay.

Rodney Brooks (55:54.740)

Okay, because you have built robotics companies, you're one of the greatest roboticists in

Rodney Brooks (56:00.500)

history, and that's not just in the space of ideas, we'll also probably talk about that,

Lex Fridman (56:06.500)

but in the actual building and execution of businesses that make robots that are useful

Lex Fridman (56:13.220)

for people and that actually work in the real world and make money.

Rodney Brooks (56:18.660)

You also sometimes are critical of Mr. Elon Musk, or let's more specifically focus on

Lex Fridman (56:24.420)

this particular technology, which is autopilot inside Teslas.

Lex Fridman (56:29.380)

What are your thoughts about Tesla autopilot, or more generally vision based machine learning

Lex Fridman (56:33.780)

approach to semi autonomous driving?

Rodney Brooks (56:38.580)

These are robots, they're being used in the real world by hundreds of thousands of people,

Lex Fridman (56:43.700)

and if you want to go there, I can go there, but that's not too much, which they're...

Rodney Brooks (56:49.940)

Let's say they're on par safety wise as humans currently, meaning human alone versus human

Lex Fridman (56:57.220)

plus robot.

Rodney Brooks (56:58.500)

Okay, so first let me say I really like the car I came in here today.

Lex Fridman (57:03.860)

Which is?

Rodney Brooks (57:06.260)

2021 model, Mercedes E450.

Lex Fridman (57:12.740)

I am impressed by the machine vision, sonar, other things.

Rodney Brooks (57:19.620)

I'm impressed by what it can do.

Lex Fridman (57:21.700)

I'm really impressed with many aspects of it.

Lex Fridman (57:29.540)

It's able to stay in lane, is it?

Lex Fridman (57:31.380)

Oh yeah, it does the lane stuff.

Rodney Brooks (57:35.940)

It's looking on either side of me, it's telling me about nearby cars.

Lex Fridman (57:40.260)

For blind spots and so on.

Rodney Brooks (57:41.540)

Yeah, when I'm going in close to something in the park, I get this beautiful, gorgeous,

Lex Fridman (57:48.100)

top down view of the world.

Rodney Brooks (57:49.780)

I am impressed up the wazoo of how registered and metrical that is.

Lex Fridman (57:56.900)

So it's like multiple cameras and it's all ready to go to produce the 360 view kind of

Lex Fridman (58:00.900)

thing?

Lex Fridman (58:00.900)

360 view, it's synthesized so it's above the car, and it is unbelievable.

Rodney Brooks (58:06.580)

I got this car in January, it's the longest I've ever owned a car without digging it.

Lex Fridman (58:11.460)

So it's better than me.

Rodney Brooks (58:13.540)

Me and it together are better.

Lex Fridman (58:15.940)

So I'm not saying technology's bad or not useful, but here's my point.

Rodney Brooks (58:24.980)

Yes, it's a replay of the same movie.

Lex Fridman (58:31.380)

Okay, so maybe you've seen me ask this question before.

Lex Fridman (58:34.900)

But when did the first car go over 55 miles an hour for over 10 miles on a public freeway

Lex Fridman (58:54.100)

with other traffic around driving completely autonomously?

Lex Fridman (58:56.660)

When did that happen?

Lex Fridman (58:59.060)

Was it CMU in the 80s or something?

Rodney Brooks (59:01.380)

It was a long time ago.

Lex Fridman (59:02.340)

It was actually in 1987 in Munich at the Bundeswehr.

Lex Fridman (59:09.380)

So they had it running in 1987.

Rodney Brooks (59:12.500)

When do you think, and Elon has said he's going to do this, when do you think we'll

Rodney Brooks (59:16.660)

have the first car drive coast to coast in the US, hands off the wheel, feet off the

Lex Fridman (59:23.780)

pedals, coast to coast?

Rodney Brooks (59:25.940)

As far as I know, a few people have claimed to do it.

Lex Fridman (59:28.340)

1995, that was Carnegie Mellon.

Lex Fridman (59:30.660)

I didn't know, but oh, that was the, they didn't claim, did they claim 100%?

Lex Fridman (59:35.700)

Not 100%, not 100%.

Lex Fridman (59:37.540)

And then there's a few marketing people who have claimed 100% since then.

Rodney Brooks (59:41.940)

My point is that, you know, what I see happening again is someone sees a demo and they overgeneralize

Lex Fridman (59:50.740)

and say, we must be almost there.

Lex Fridman (59:52.340)

But we've been working on it for 35 years.

Lex Fridman (59:54.900)

So that's demos.

Lex Fridman (59:56.180)

But this is going to take us back to the same conversation with AlphaZero.

Rodney Brooks (59:59.540)

Are you not, okay, I'll just say what I am because I thought, okay, when I first started