I was 25 feet away from a well-known humanoid two weeks ago, and I was at a cocktail party, and no one was interested in it because there were roboticists at this cocktail party. They knew it was junk. Or junk. That's maybe an overstatement. But I said to someone, hey, hey, why don't we go and try robot tipping?

And she said, yeah, we weren't really going to do it. But we turned to look at the robot. was about 25 feet away. And at that moment, it fell flat on its face.

Brian Heater (00:32)

Hello and welcome to automated. My name is Brian heater. I am the managing editor at a three. I believe the first time that I met Rodney Brooks is when he spoke at TechCrunch's first robotics conference way back in 2017. I distinctly remember three things from that events. The first was an engaging fireside chat.

The second was a photo that we took of him and the other two iRobot co-founders. And the third was him actively engaging with someone else's panel from the crowd. Since then, he has become one of the first people that I go to when I need a pragmatic look at the latest robotics hype cycle. I was rereading your predictions in the lead up to this conversation because it had been several months. And one of the asides in there really jumped out at me given the current context of everything you wrote. I suspect many people will reason that I cannot have a valid opinion about this being humanoid precisely because I happen to have built more humanoid robots than anybody else on the planet. So read with caution. So what is the what what what's the root of your skepticism right now and how does your experience in humanoid inform that?

Rodney Brooks (01:58)

OK, I guess two things. One, I'm skeptical about the end-to-end learning for humanoids, because that can only possibly work if you're collecting the right data and trying to learn the right thing. And I don't think we've had any lab demos, even at small scale, collecting the right data or learning the right thing. it's sort of, you know, we'll try this. And maybe it'll work, but we're going to put lots and lots of money into it. I think that all the training regimes I've seen are really based around the human motions, either captured with cameras or wearing some sort of suit or holding some sort of joysticks. But the real thing about manipulating the world, and that's what humanoid is supposed to do, they're supposed to manipulate the world the same way humans do, the real thing about manipulating the world is when you make contact with something and you're both applying a force and sensing a force. And that's not being captured at the moment. So I'm skeptical there. The other place I'm skeptical is about the legs. The current humanoids do not walk like humans walk. We're very passive walkers. We don't need to use much muscle at all to walk. Our system is built to walk. Instead, the current walking algorithms, they're all based on the ZMP algorithm, zero moment point algorithm, which has been around for about 20 years, where they're not good at falling. And when they start to fall, they pump a lot of energy into the system to try to recover, which is good if they recover. But if they don't recover, then there's a lot of kinetic energy in all those joints and limbs. And it's got to go somewhere. And if a human happens to be in the way, it's just going to mess with them, which, by the way, you can see by the way some of the humanoid companies demo their robots.

When the humanoids are walking, they say keep away. Then you can come up to it and talk to it, but it's actually a person somewhere else with a microphone and camera. But they don't allow people next to the robot when it's moving its legs. Watch all the videos for that. It's a real tell.

Brian Heater (04:10)

Which is ironic, right? Because that's supposed to be one of the selling points is that you can keep them out of the cage. So when you say that they're not trying to learn the right thing, I'm curious what you mean by that and whether or not, like how much actually having these out in the field, you know, as I'm not telling you anything you don't know, but obviously you've got your robot out there now and it's constantly in the process of collecting data and it's constantly getting better. So through these pilot programs that they're doing right now is the right data coming in.

Rodney Brooks (04:40)

I don't believe so. There's the pilot programs and then there's the large manipulation models where they've got lots of people training the robot. Some of the companies have people training the robot, banks of people training the robots. there, humans are doing the motions and they're trying to capture those motions and put them into a model that the robot can then do the same things. But again, if you look at the...big tells in the, I call it humanoid theater. Humanoid theater are this set of videos you see coming out of the humanoid companies where they sort of shake, the robot arms shake as they go because they don't have good control algorithms underneath because it's coming from this reinforcement learning system. And then the grasp they do are very primitive grasps. Us humans can...

It's not staged, I just picked up three different things and I can manipulate them separately in my fingers. I can do all sorts of things. There's nothing like that. The best you see is sort of a, in humanoid theater, know, shake, move, just a pinch grasp. So I'm waiting to see. None of the videos have blown me away nor nor the in-person demos that I've seen. I was I was I was 25 feet away from a well-known humanoid two weeks ago.

Brian Heater (06:17)

There's only about two or three, so he could probably figure it out.

Rodney Brooks (06:20)

And it was at a cocktail party and no one was interested in it because there were roboticists at this cocktail party. They knew it was junk. Or junk. That's maybe an overstatement. But I said to someone, hey, hey, why don't we go and try robot tipping? And she said, yeah, we weren't really going to do it. But we turned to look at the robot. was about 25 feet away. And at that moment, it fell flat on its face.

Brian Heater (06:47)

I had a very similar, I will also not name the company, but I had a very similar experience wherein I was around a robot demo and they were telling me something it could do. So I said, well, let's see if it can do that on the fly and moved it towards the robot. And the guy was like, no, that's not part of the demo that we're doing right now. In terms of your skepticism, if I'm understanding it correctly, it's less that you don't believe in the form factor. Generally, it's more that you have doubts about claims where it is right now and how long it's going to take to really sort of roll out and be useful.

Rodney Brooks (07:24)

Yeah, and this gets back to my comment about maybe I'm not to be trusted because I sort of know how hard this is. And there's a whole bunch of people who haven't really done anything in robotics, but they see the human form and think, oh, that must be the right way. That must be easy. But, you know, I've been doing this for 50 years and it's hard. It's really hard to make progress. We really don't understand how to do physical manipulation in the world.

Brian Heater (07:53)

Yeah, I mean, obviously there are those people, but there are also a lot of very talented people with experience in the field who are currently working on these robots.

Rodney Brooks (08:04)

Mmm, yeah.

Brian Heater (08:06)

Okay.

Rodney Brooks (08:08)

Very few of them have ever deployed robots. And I know how hard it is to deploy robots and how hard it is to make something that the customer's going to pay for. And it's got to be very, very reliable. And it's got to work with a bunch of nines, 99.999 % of the time. If you get higher failure rates than that, it's really frustrating and perhaps dangerous. Yeah.

Brian Heater (08:30)

Okay, so that's a fair distinction. I was thinking, and maybe a little bit more theoretically, but obviously people who have been in research facilities and have been working around robots in that context, but the breakdown for you is that they haven't really scaled and they haven't really produced industrial robots in a meaningful way.

Rodney Brooks (08:49)

Yeah, the customers in general, in general customers don't care about the particular technology using. They want cost-effective results. They want it to somehow make them more efficient than they were without the technology, and that's what matters to them. It doesn't matter that it's using a large manipulation model or that it's using deep learning, which both can be good things, but...

They don't care about the particular technology. Whereas people come from a research background, they're built a particular technology. That's their identity. And they're going to use that and show the world that it's important. And it's something not so important.

Brian Heater (09:30)

Yeah, one of the things that really struck me last time we spoke, I was actually out there in Redwood City and actually seeing a demo of the robot in person. One of the things that's baked into it that I don't see baked into a lot of these robots that we're talking about right now is that collaborative element and is the ability for human workers working alongside these things to like literally physically pick it up and move it out of the way.

Rodney Brooks (09:57)

Yeah, and that's been something that I've been doing for a long time. You know, the Roomba had a handle on it. You could pick it up, you know, if we got stuck in a bunch of cords, it's pick it up and move it. The humanoid robots only waste up humanoids, Baxter and Sawyer. You could grab their arm at any time and push back on it, which you cannot do for most industrial robots. You know, the traditional industrial robots that are used in...auto factories, et cetera. We made a robot that you could grab the arm and move it out of the way. For the cart that we have in this company, it can get in the way and you just want to not where it is. And just having a handle that you can grab and move it is great. And then some of our customers are just really focused on that handle because there can be 100 kilograms on the cart and it feels like nothing. In fact, my both my three-year-old and six-year-old grandsons have come and pushed the robots around by that handle with a lot of weight on there. that makes the job easier for people because they've got power assist, but they can still have control if they want control.

Brian Heater (11:14)

This this necessary 99.999 % accurate rate is something certainly that I understand in the context of like self driving cars, for example, obviously, if one of those doesn't do the right thing, then that could result in death. But let's take the room for example, right? mean, people love their room is but every single person who owns a room has probably run into multiple issues. Is it just that the nature of industrial deployment is that different from the home?

Rodney Brooks (11:43)

Yeah, well, the nice thing about the Roomba is it doesn't weigh much. And so it can't do too much damage when it does fail. And we put a lot of effort into making it so it never falls downstairs. had triple redundancy there, because that's the only time you can get enough kinetic energy in a Roomba for it to be dangerous when it's falling from a great height. Other bigger robots, though, don't have to be moving too fast to have enough kinetic energy to squish you.

So there you really want that reliability. It's not a lot different from autonomous driving. 100 kilos of mass can do some damage even at a meter a second.

Brian Heater (12:28)

So we are talking about ultimately a safety issue there.

Rodney Brooks (12:31)

There's safety, but there's also frustration. So in the case of an industrial robot, an industrial robot that's sitting somewhere on the line that's supposed to pick something up and dip it in paint or put it on something else, if that's failing once an hour even, and maybe it's only doing 500 or 1,000 operations an hour, but if it fails once an hour, that's a pretty useless robot because that means a person has to be checking on it. It's cold all the time.

If it fails once a day, maybe that's okay, but if you've got 50 robots and they're all failing once a day, that's 50 failures you've got to deal with. as you scale up, there's got to be a lot more nines or you need more people to just take care of the failures.

Brian Heater (13:19)

The stakes of a Roomba accidentally sucking up one of your socks are probably not huge to you, but when your company is relying on these things and when you're paying probably 50 times the cost of the most expensive Roomba, then they really need to deliver almost all the time.

Rodney Brooks (13:37)

Yeah, otherwise the customer's not going to get return on investment because they need a babysitter for the robot. They need babysitter for every 10 robots or something like that, some calculation.

Brian Heater (13:44)

Yeah.

I know AGI is something that you've interested you for a long time and you've been thinking and writing and speaking about it. Is that something, how close do we have to get to that in order for these robots to be meaningful in a way that they're actually valuable to the businesses that use them?

Rodney Brooks (14:08)

we don't need AGI at all for those to be useful. And that's good because we're not going to get AGI for 300 years. That's an exact number. You know, everyone in the 50s and 60s in artificial intelligence, they didn't call it AGI, but that's what they meant by artificial intelligence. The same thing people meant. The reason we talk about AGI now is a group of researchers decided to differentiate themselves and claim they were doing general intelligence, but that's what everyone thought they were doing. So it's a...

It's a marketing ploy. And we're just a long, long way from it. Certainly, the LLMs are nothing like that. They can do things that are surprising, but that's different from being generally intelligent.

Brian Heater (14:47)

I wonder how much of this is a case of people, various people having different definitions of what AGI ultimately means. How would you briefly explain to somebody on the street what AGI is?

Rodney Brooks (15:00)

Yeah, so when AGI first became a term, it meant to be able to do any reasoning that people could do and be able to do anything in the world that people could do. But as with all these definitions, they change. Flying cars used to mean a car that you could drive and that you could fly. Now a flying car or a flying taxi means a quadcopter that can land at a particular location and then can't move.

So the definitions change over time. These are moving targets for what it is. And I think AGI has turned into it can take a test and pass the test maybe. That's not general intelligence. going to figure out what, you know, look at a situation like any human who's not even particularly trained can come into a room and see what stupid things a machine is doing and intercede in all sorts of situations, even a machine they've never seen before.

they can recognize, this is failing. This thing is clearly broken. Because they can build up a model in the head of what the machine must be trying to do and what it's actually doing and figure out a chain of things which involve geometric reasoning, involve perceiving things that they've never seen before. And that's not what any of these AGI tests are about.

Brian Heater (16:18)

And one of the things that I think you are really optimistic about is this long ongoing conversation about technology replacing jobs and that it's going to take a long time that the jobs that people, I guess we call them blue collar jobs that people have are going to continue to evolve. And it seems like an important piece of that in the meantime is to build technology that has humans in the loop as much as possible or as much as needed.

Rodney Brooks (16:49)

Well, if there's any grasping involved, we're going to need humans for a long time yet. Grasping has been worked on academically, and Amazon's put a lot of money into it. And we're still not anywhere close to having grasping that humans can do. So humans will have to grasp. But what we can do is automate the things around them that are not grasping. Amazon does that by having robots bring shelves full of stuff.

to a human who's in one place and the human picks from those shelves and the shelves drive away and another set of shelves come. In our case, we're going around existing warehouses without automation and the humans pick and put into the robot cart, but then, you know, the last 400, 500 feet to go to the pack station, they just say go and the robot goes and they don't have to do that what's called dead walking. So two things that we...we make better for the humans. The number of steps they have to do and pay, often in warehouses, typical sort of number of steps for a human can be over 30,000 steps a day, which is grueling when you're doing that every day for your job. So we reduce a lot of that. And then we make the carts appear to be lighter than they are when they are pushing them with the power assist and the handlebar. You don't have to be a big football player that you need to be sort of in some of these facilities just to move the carts around. You can be a small person and it's the same amount of effort as for a big person. So it increases the labor pool and it makes the jobs physically less taxing, but still has the people doing the grasping because we can't automate that. No one can.

Brian Heater (18:33)

Yeah, one thing as you're as you're speaking about Amazon reminded me of something I think that maybe doesn't get talked about enough or at least doesn't get talked about enough in broader circles, which is the degree to which given the current state of technology, it's not as much in a lot of cases is not as much about the robot complying to the existing environment is actually kind of building an environment that makes sense given those limitations.

Rodney Brooks (19:01)

Yeah, and that's what Amazon uses in about 80 % of their warehouses. They don't use it in all of their warehouses because big things are not amenable to that particular solution. need to move them and pick them up in different ways.

Brian Heater (19:17)

Yeah, I'm going to ask you a question and you know, it may or may not be relevant in the same way. Hopefully it won't be when we actually release this podcast out into the world. But given I'm really curious what kind of the macro state of politics and the economy in the US in particular has had on your company, both from the standpoint of orders and interest in the products, but also how much these things actually cost to produce.

Rodney Brooks (19:50)

Yeah, well, it's I say it's early days on that because it you know, I haven't looked at the news for two hours. So I don't know what the play to play. So it goes up, it goes down, you know, the so there's two aspects to that. There's the competitive aspect with other people trying to do the same task. And I'll get to that in a second. And then there's the question, will this these tariffs reduce the amount of goods that are around?

And will that have bad effects? Will it cause a depression so no one can afford to buy anything anymore? I think those are macro scale things that are unknowable at this time. And I would expect as we get closer to that being real, more people will step up to intercede and push back if we're going to destroy the country's economy. One hopes that doesn't happen.

You know, maybe it will, but I can't control that. So I just got to do what I can in the existing world. The good news for us is that for our solution, we've already done the cost reduction. So everyone's going to have the same problem with tariffs for building. It's going to change the cost by pretty much equivalent percentage. But we already did cost reduction, serious cost reduction to build low cost robots. So we don't think we're going to get we think we're to be in the same position even if all the prices move. That's our take on it. Now, it may mean that we have to do, you know, people say, well, why don't you just manufacture in the US? Well, you need a supply chain for all the components. And that means the components have to come from other places. you know, we are in an intertwined world. Where the particular components come from at different times has changed.

In my last company, when we wanted particular high-performance motors, the only place we could get them were in Eastern Europe. That was the place where you went to get high-performance motors. Since the influx of hoverboards and scooters, there's just some tremendous supply chain in China of building high-performance motors for that particular application. So the number of motors being built in China has gone way, way up. So it makes sense to buy motors in China because they're just so much cheaper. You can find manufacturers of equivalent motors in Japan. They're more expensive. Not so in the US. It's just not those motors. So how long is it going to take to replace all those supply chains? We'll see. Nvidia just said they're going to build their chips in the US, in Texas, to me says it's going to be built in a factory that is part of the Biden chips act. But we also hear, we're going to get rid of that socialist Biden chips. We're to manufacture in the US. It makes no sense. So I can't rationalize it. So I can't give you a rational answer because it makes no sense. If you ran your own business this way, you'd be fired by the board.

I'm sorry, but it's just stupid.

Brian Heater (23:13)

Again, ironic from the guy who is famous for firing everybody else.

Rodney Brooks (23:19)

I'm not going to go there. I'm just not going to there. But it's stupid and it's damaging. At the moment, we build our robots in the US and we export them. And now we're having counter tariffs put on us. So it's already hurting manufacturing in the US. It's hurting it.

Brian Heater (23:35)

There are all of these conversations that you have to have and all these things that you have to think about because you keep putting yourself in the position that you're in as a co-founder of a startup. Given the ups and downs that you've experienced with different companies along the way, what keeps bringing you back into that world instead of, I don't want to say retreating, but spending more time, say, in the university?

Rodney Brooks (24:04)

Yeah, well, you know, I've got a book, like Magnum Opus, so how you say it, that I'm trying to write, but it's been on hold since I started this company. And I want to get back to that. I also, I like being in the arena. I think Teddy Roosevelt used that phrase. So when I...say something's going to take a long time or something's going to be hard, it's not because I'm just, you know, read something or saw a picture. It's because I've lived it and continue to live it. And so I have pretty good estimation skills on how hard certain things are to do. And I get that from trying to do stuff.

Brian Heater (24:48)

Yeah, I mean, one one thing I like to ask people is as somebody I'm certainly. New to the space relative to somebody like you, who's been in here for decades is if this.

Rodney Brooks (24:59)

You didn't say it, Someone as old as me.

Brian Heater (25:02)

someone as old as Rodney, if this current hype cycle that we're going through, and I would include both, know, LLMs and human waste in that, is in any way different than ones you've been through in the past.

Rodney Brooks (25:15)

I don't think it's different in what people are saying or how over pumped up their expectations are. It's different in that it's leaked out into the general world. And then there's all these pundits who've never even seen a robot talking about what robots can do and how much per hour they're going to cost and all sorts of crap, to use a technical term, being talked about, all sorts of things.

So that's the difference. It has leaked out and has become commonplace. And so when I'm at a dinner party, people come to me and fearfully ask, is their job going to be replaced? Is this going to happen? Is that going to happen? And the questions are sort of ludicrous, because no, they're not going to happen the way people have been scared into thinking that they're going to happen. By the way, I think there's a real parallel between LLMs and humanoid robots in the following sense. The phenomenon.

Phenomena of them, know big tech companies are putting billions of dollars into them Academia is sort of getting priced out. That's not where the the action is happening. It's happening in the big companies But you know the promise then the investors get excited because LLM is gonna replace lots of white-collar workers and humanoids are gonna replace lots of blue-collar workers and then how do they get lured into that? Well, the LLMs use human language. Well, that's pretty damn good. And the humanoid robots have human form. Well, that must mean they're like humans. the LLMs can talk like humans. The humanoid robots can move like humans. And so there's a lot of similarity there between the two things. And I think that's why the expectations have gotten so large and it's burst out into the world. Once it starts, all the big tech companies have to say they're in.

or they'll be viewed as being left out. And then people who are trying to raise money, you know, come up with such inflated promises that will never be met in order to get big piles of money to pay for training data, et cetera.

Brian Heater (27:25)

Yeah, Apple's probably the best example of that right now when it comes to the AI side of things. Yeah. I'm surprised to hear that this is a relatively new phenomenon for you, people asking about their jobs being replaced because especially-

Rodney Brooks (27:42)

Yeah, I did get it back, you remember, all the way back in ancient time, 2011, when Watson won Jeopardy. And then IBM, the next day, John Markov had a story in the New York Times saying what IBM was saying, that they was going to do medicine with it. It was going to...how people choose objects online or to buy. They were going to put a voice on it. It was going to be out of the standard language. You would take the story that John, who's a dear friend, by the way, wrote in 2011, replace Watson with generative AI and IBM with Microsoft. And it's exactly what happened. The same story was the case. It's going to do health. It's going to do medicine. It's going to... And I got a lot of people...actually in my apartment building in Cambridge who worked in biotech and coming up to me, is this true? Is this going to completely change biotech? Is it going to change medicine? And no, wasn't true and no, it didn't. But they were really dumb back then, those people in 2011, to believe that. Not like now, we're smart now. So if we believe it now, it must be true.

Brian Heater (28:51)

Yeah.

Do you is it not one to one, but is it an extremely close parallel to what we just went through with autonomous cars?

Rodney Brooks (29:08)

Yeah, autonomous cars may or may not come through. I think the next 10 years will tell. 10 years. Back in the olden days, the mid teens, autonomous cars were going to be here and everywhere in three years, you might remember. But I think, you know, there are two players, Waymo, which is number one, you know, which is powered by Google, and Zooks, which is a distant number two, but nevertheless a number two, which is owned by Amazon.

They both have big resources in capital. If they can't in the next 10 years really make it big and widespread, think interest will get lost and we'll go back to other solutions, which maybe we should have been working on in the meantime. Where instead of hoping everything came in the car and we didn't have to change infrastructure, in fact, changing infrastructure may have been a better idea to put more sensing on the roads, more...communication with cars as they drive instead of it being you don't have to spend anything on the infrastructure I'm just going to deliver you this car that's going to do everything by itself again thinking that it can reach human level performance.

Brian Heater (30:21)

Maybe this is an impossible thing to predict, although that hasn't stopped you in the past from predicting things. Do you see these, do you see humanoid in particular continuing to play out a similar trajectory as autonomous cars in the sense that there's going to be one giant company keeping the flame alive and everybody's going to sort of shrink back or go away? Or is it something that's just kind of going to go away for a bit?

Rodney Brooks (30:46)

I think it's probably going to go away. If someone really comes up with a better walking mechanism soon, then that might make things safe enough and cheap enough to be useful. But I don't see the manipulation part getting fixed. So I think it's going to die

Brian Heater (31:08)

you group these kind of, guess, modular versions of human noise that we're seeing right now where it's like a torso on a wheel base in the same place and that, okay, it can move around, you know, the way any real robot can, but maybe the state of manipulation isn't as good as what's already out there.

Rodney Brooks (31:26)

Yeah, well, I built a humanoid robot with an arm and wheel base in 2004, published about it on page one of issue one of volume one of humanoid robots in 2004, where I made the argument that that tool form was going to be useful in built for human environments. And we had that robot going around and opening doors in a lab in the, you know, where the students had their offices etc. And manipulating that bill for human environment. It's called Cardea. Cardea is the Roman god of door handles.

Brian Heater (32:06)

They get very specific for an ancient realm, don't they?

Rodney Brooks (32:10)

Yeah, anyway. So, you know, the idea has been around a long time, but that was built on two wheels as the ones we see now are. It was actually a... I forget the name. The thing people rode around on two Segway. It was a Segway, an actual Segway base. But it was a bit dangerous when things got bad, and those wheeled bases are still dangerous when things get bad. So when you lose your balance.

So maybe, but maybe not. I have one other data point from 2011, March 2011, when the Great Tsunami in Japan happened. And there were meltdowns in three of the four reactors, the six reactors totaling Fukushima Daiichi. But they were forming the water line and three of the four ended up with meltdowns.

You know, the idea that a built for human environment is a perfect one for humanoids to go into was put to rest. Japan had more humanoid robots with legs than anyone, and they couldn't go in there. We were led to, you know, at iRobot, also had military robots. We had 6,500 in Afghanistan and Iraq. They were tracked. They could go over any terrain. And we were...given to understand that the Japanese government would not be insulted if we sent some. So we sent some and they got there seven days after the accident and they were the robots that went in to figure out what was going on, get cameras in place. The radiation was just so intense people couldn't go in there and it was a non-digital, it was a 40-year-old nuclear power plant. So the only thing you could do was go and look at physical dials, pressure and temperature with no data communications.

Tokyo Electric Power Company personnel who we trained remotely operated the robot, set up Wi-Fi hotspots, set up a network over those robots and had them go into the damaged reactors. there was the perfect situation for humanoids to into a built-for-human environment and it did not pay off.

Brian Heater (34:25)

In that particular scenario in search and rescue, do you think there is more efficacy potentially in drones and also quadrupeds like Spot?

Rodney Brooks (34:38)

Yeah, yeah, yeah, absolutely. drones were just not a thing. There were a few research ones, but they were nothing like the capabilities and communications we have today. So yes, drones would have been perfect for that. And spotlight robots, four legs, rather than two, because it's much harder to fall down with four legs and this with two legs. And dogs actually go in most places humans can go anyway.

Brian Heater (35:03)

Getting back to the humanoid that you built earlier, Baxter, one of the things that has really fascinated me about it over the years is how it became and continues to be an important part of the research world. They're still out there, and I still see them when I go to schools. And probably a lot of that is a product of other companies not really coming in and necessarily filling that void.

Rodney Brooks (35:31)

One of the things about them is they're safe. You can get sand in the way of it, and it senses when its forward model isn't matching reality, and it pumps the kinetic energy out of the system so you don't get hurt. So that makes it really good for research because you can feel confident having graduate students late at night working with it, and they're not going to get hurt. undergrads, you know.

Berkeley I think has six of them for their undergraduate program in robotics still that they teach.

Brian Heater (36:04)

Was that research, was that university aspect, was that almost accidental or was that part of the plan?

Rodney Brooks (36:10)

No, it was not part of the plan. It was accidental and I had to fight for it. Because, you know, I realized after we built them that it would be good for research. And I said, let's start selling them to universities. the industry people in the company said, no, no, no, it'd be so much support. It'd be so hard that it'd be horrible having to deal with graduate students and stuff. And so I got a few professors together and asked them some questions and I said, you know, one of the questions was, will you be comfortable maintaining the robot and fixing it, moldering it? They said, I've never had a robot that I didn't have to do that, of course. And that's what happened. So it was industry not understanding the mindset of research universities where that's what everyone does. They thought, oh, we have to build a perfect product that any student can use. Students take them apart and hack them and do all sorts of stuff.

Brian Heater (37:08)

Yeah, one of the things that I appreciate that that fetch was doing for a while and I think that this is kind of a direct product of them in a certain way being an outgrowth of Willow garage is that they were building a research robot and they were doing it because they had the very savvy understanding that getting students at universities familiar with your technology is ultimately going to serve you in the long run.

Rodney Brooks (37:32)

Yeah, and they used ROS, which students are already using. And by the way, here, when we started, we bought a bunch of fetch platforms and dressed them up as to simulate the real robots we wanted to build. it let us start running all the perception algorithms long before we built the mechanical robot that was our first series, which is now we no longer use. It's the second generation, which is what we're deploying at scale.

We simulated our robot with a fetch, which was great.

Brian Heater (38:05)

Yeah, I wanted to, you alluded a little bit to your magnum opus that you are unable to write because you don't have the time right now. What's the blurb on the back of that book? mean, what's sort of the big thing that you want to impart on people through all of your years of experience?

Rodney Brooks (38:22)

different than what we're talking about here. But it's true with AI. The main idea is that if you have a two by two matrix with science versus engineering and intelligence versus life. So there's four boxes. And one of them is neuroscience. One is AI. One is artificial life. One is abiogenesis. How did life come from matter?

And those four disciplines were all formed into their modern form in the period 1945 to 1965. any two of them, I can show you a person who worked in both of those two. And for any three of them, I can usually find a person who worked in three of them. And three out of the four have decided that computation is the primary metaphor. And it's questioning them. It's questioning whether computation is the right way to think about things. Let me use an exam.

If, you wanted to put something into orbit, you wouldn't just write a Python script and think that would get something into orbit. You know you have to burn rocket fuel or have something that releases tremendous amounts of energy in some way. Computation is not the right kind to get something into orbit. But we've decided it's the right kind to explain what goes on in our heads and our bodies. And that's the question, whether that's accurate.

Brian Heater (39:50)

Yeah, so this is my, listen, was a creative writing major at UC Santa Cruz, so maybe not as technical as I should be, but so my dumb liberal arts degree question is what's the alternative?

Rodney Brooks (40:04)

Yeah, and that's the question. But you don't say that about rockets. What's the alternative to writing code to make a rocket? You say there's something different. But we put that aside. I've gone, actually I start, I've written a whole bunch of this. I start back in 1614 with Latin manuscripts and trace the ideas from there through what computation is, what's the difference between simulation and actuality.

Um, so here's, here's an example. I'll give you a simple example of something, um, which looks intelligent in some way. If you look at it the right way, when you're on a freeway and there's an incoming, you know, an incoming, uh, single lane, there's a, normally a piece of space between them that gets narrower and narrower as the lanes come together. And if you look down there, you'll see them covered with little bits of junk and those bits of junk are not on the on the freeway and they're not on the on-ramp. But those bits of junk are on the blacktop and confined to that area. Well, how did they get there? Did they think about how to get there? No. What happened is every time there's a piece on one of the lanes or the on-ramp and a car hits it, it jumps up and randomly moves. And once it's in that area between the two white lines that are coming together, no car ever goes there, so it doesn't move again.

So the cars sort stuff into places where cars drive and cases where cars don't. You could build a robot that thought about that to sort, do the sorting and computed about it, or you can just have to let the cars drive. So there's a, there's an emergent thing and maybe we're just emergent in a big scale and what's that emergence look like? And evolution has built us in that way. And this hurts people's heads when I say this, they just.

They just can't conceive there could be something different. So I've got a chapter coming out in MIT Press Book with 12 questions that we assume about intelligence, but which we actually don't know. they're just hard for us to get over, thinking that we couldn't have it right. And some of the questions are, do you need a living body for intelligence?

We can't prove that you don't need a living body for intelligence because the only true intelligence we have is in living bodies. Well, maybe there's something there. Sorry, I'm trying to make people think different and hurt their heads.

Brian Heater (42:40)

I appreciate that. actually, on that note, so again, something I think that probably doesn't get talked enough about outside of computer science circles is how much of these LLMs or just like large model AIs are effectively like a black box from where we sit. And I saw in your piece that you wrote, in your predictions piece, you quoted that very famous the city

Rodney Brooks (43:15)

Interesting to watch from magic.

Brian Heater (43:18)

Yeah, which I mean, that has to be an exciting and one of the more interesting aspects of your job is not realizing that there's going to be rubber on the road until those cars are actually out there.

Rodney Brooks (43:30)

Yeah, some of them are pretty mild, but one thing that Roombas did was make people keep their houses more tidy so the Roombas wouldn't run into stuff that they couldn't handle. So the Roombas were to save them from cleaning their house, but they kept it tidier so the Roomba could do its work. So you get weird effects like that. I'm not sure that's really where you're going, but these things do change how we see.

Another one of the things I talk about is we tend to overestimate stuff in the short term. This is Roy Amara who said this. We tend to overestimate stuff in the short term, but underestimate in the long term. So, you my favorite thing is a Spencer Tracy, Catherine Hepburn movie from the 50s called Deskset, where a computer has been brought into this office building where there are women who aren't...you know, go and look up in the library questions the executives have and the computer's going to replace them. course, Catherine, know, Spencer Tracy brings the computer in, Catherine Hepburn's the head librarian, they pull in blah, blah, blah.

Brian Heater (44:41)

Unexpected unexpected results of technology.

Rodney Brooks (44:45)

But the point is that the computers of the 50s were not going to replace librarians anytime soon. And for 30 years, the number of librarians in the US continued to rise. And then it dropped suddenly in corporate wealth, because then we got to search engines and information systems. And then the idea that you can carry this little thing around with you and it's got all the world's information at your fingertips was not expected.

when we had punch card machines in the 50s. So it was overestimated, it's going to replace librarians, but underestimate the scale, which was going to replace librarians and replace information accessibility.

Brian Heater (45:29)

That's interesting. So there really is a parallel there between that example and the Roomba in that when the technology isn't quite where you want it to be or where you project it to be, actually takes more human interaction in order to get it there. And ultimately, like in the case of the Roomba, it means that you're actually working a little bit more when you hire this thing to relieve you of that problem.

Rodney Brooks (45:54)

Yeah, although, you know, it can, but it can also relieve you of a problem. have a house where, when I'm, you know, a whole vacation place, when I'm not there, I know that a lot of mice come to visit because there's mouse poop on the carpet when I come back. So I just let the Roomba pick it up and that's great, you know, it's just, so it saves time. But I want to use this example of information for...in the Singers from Magic, in the case of my mother, who's long departed. But I was sitting with her in Australia. I used to visit her twice a year after she was a widow. I was sitting with her and she was reminiscing about her grandfather and something about her grandfather's funeral in the 1930s. And I just went and I got the newspaper story up of her grandfather's funeral and showed it to her. And she said, you had that in this little box?

So it was interesting from magic to her. was magic that I had her grandfather's funeral newspaper report, which she probably never even read because she was too young when it happened.

Brian Heater (47:02)

Yeah, so as a very technically and logically minded person though, you don't have an issue using the word magic in that context.

Rodney Brooks (47:10)

No, because it's indistinguishable from magic. She had no way of explaining it. And so when we see things that we have no way of explaining, we don't know what its limitations are, because everything it does seems magic. So what is outside its range of possibilities?

Brian Heater (47:31)

So we're coming up on time and I want to leave you with this. For some reason, I've gotten in the habit of asking like big broad questions to end impossible questions to end interviews. But you very famously were one of the subjects in Errol Morris's Fast Chief and Out of Control. And I was, you know, it's been a very long time since I've seen that movie and I was reading up on some of the reactions to it online. And there's a very interesting Roger Ebert quote. And I don't know if you've seen this, but I'll read it out to you.

He basically describes the film as it's about people who are trying to control things to take upon themselves the mantle of God. Does that resonate with you?

Rodney Brooks (48:12)

Yeah, think that's the theme that was in Errol's head as he put the movie together. For people who haven't seen it, there were four people in there. One was a lion tamer. One was a topiary gardener, making animals out of shapes. One was a person who built displays for naked mole rats, which are little underground creatures. And there was me building robots in...human form. was building humanoid robots back in the 90s when he filmed that. So we were all in our own individual ways trying to make something bigger or different than ourselves in living-ish sort of systems and all ultimately failing. So he shot that movie in 1990. I he shot the footage of me. He shot the footage of the four people at different times in 91 or 92. So I've been failing for well over 30 years now.

Brian Heater (49:12)

Yeah, I mean the expectation of taking up the mantle of God is pretty high but I guess the parallel there and certainly in terms of AI and humanoids is that this is simplifying it and obviously this is just a gigantic analogy, but that you're recreating life essentially.

Rodney Brooks (49:32)

Yeah, and there's a belief that somehow it's going to be better than us. You know, I just saw Sam Altman in an interview say that he now has a child, but of course the child would never be as smart as the AGI that he's building. I bet on his child.

Brian Heater (49:50)

He clearly has a favorite child. And it's the AGI.

Rodney Brooks (49:53)

The human child,

Brian Heater (49:57)

It is always a pleasure speaking with Rodney Brooks. You can check out what he has been working on over at robust.ai. Head over to RodneyBrooks.com for his blog, including the annual prediction scorecard. If you've made this far, please like and subscribe and check out the automated newsletter over at automated.org. Thank you for tuning in. We will catch you again soon.