James Vincent (2:30)

I lost count after the first few. I'll be honest. I met a bunch of very nice robots, and they were all very kind to me, and they treated me with, you know, great nobility and grace. No, I met a few from two of the leading companies in the US One is called Apptronic and another is called Agility Robotics. And they make two very different styles of robot. Well, I mean, they're both humanoids in that they, you know, they resemble a human, arms, legs, etc. But agility is very much focused on the warehouse and their robots look a little bit more inhuman. They have those backward facing knees. This may look like your average American

Sean Rainswerum (3:03)

warehouse, but Agility Robotics claims it's actually

Sponsor/Ad Voice (3:06)

the world's first factory for humanoid robots.

Sean Rainswerum (3:09)

Its weird grasshopper like legs allow it

Ken Goldberg (3:12)

to crouch and pick up items off the floor. And it can surprisingly handle stairs as well.

James Vincent (3:17)

Apptronic make a more general purpose robot that looks much more like a human in terms of, you know, normal sort of body proportions. Stands upright, you look it eye to eye, or eye to unblinking robot eye, whatever that might be.

Sponsor/Ad Voice (3:30)

Meet Apollo, your new robot coworker. He's designed to work alongside humans in a factory and help alleviate taxing and physical labor.

Sean Rainswerum (3:41)

The goal is that he's going to start doing work here on Earth, but then long term going to the space station. From the space station to the moon, Mars and beyond.

James Vincent (3:51)

I got to meet them, shake hands. I played Ikakok. Do you know this? No. Rock, paper, scissors is what it is.

Sean Rainswerum (3:58)

Sorry.

James Vincent (3:58)

Oh, it's called ick ac ock sometimes in the uk.

Sean Rainswerum (4:02)

I did not know that. I thought we were going to talk about robots, but that's now the most fun fact you've given us.

James Vincent (4:09)

Yeah, you go ick, ick ack.

Sean Rainswerum (4:10)

Ok. Yeah.

James Vincent (4:11)

And then you throw, whether it's rock, paper or scissors.

Sean Rainswerum (4:14)

Anyway, they come to us for one thing, but we give them something else, you know.

James Vincent (4:19)

Exactly. And I also got this was my heart's content, you know, I so wanted to do this. I wanted to kick a robot. Kicking robots happens to be an efficient method of testing a machine's balance. In recent years, as robots have become increasingly sophisticated, their makers have gone from kicking them to shoving them, tripping them, even hitting them with folding chairs. I had that burning urge inside me that I want to get my own back before they obviously take over the world.

Sean Rainswerum (4:48)

So the robots were nice to you, but you weren't that nice to them?

James Vincent (4:50)

Oh, I was horrible. I was terrible. They're going to be coming for me in the future. I have no doubt about that at all. They didn't actually let me kick a robot. I'm very sad to say that they said it might be a bit of a safety hazard, so I got to poke one very hard with a big stick instead. And that was the next Best thing.

Sean Rainswerum (5:05)

Did it tip over?

James Vincent (5:06)

No, it didn't. This was the creepy thing about it, okay? So they gave me this sort of, you know, very high tech stick which was, I think, you know, a broom handle with a bit of safety foam taped on the end of it. And they said, give it a shove, give it a punt, see how hard you can push it. And I was very nervous about this because they told me that this was, you know, one of the prototype humanoids. It was worth hundreds of thousands of dollars and all the rest of it and, you know, if I knock it down and it breaks, that's great copy. But it's also the end of my access to this company. They're not going to be pleased. So I gave it a shove and it sort of wobbled and they were like, no, you can do it harder than that. And I gave it as hard as I could and it staggered backwards and it did this. Amazing. I mean, like completely normal. Obviously, when, whenever you've pushed a human, you know what it's like, they stagger backwards, they throw their arms up in the air and they regained their equilibrium. And it was just such an uncanny moment to see a robot mimic so perfectly, to my eyes, the movements of a human. And I remember doing this and having it sort of stagger backwards and then trot back up to me, look me right in the face. And I was like, oh, gosh, you know, these things are real.

Sean Rainswerum (6:14)

And what are they meant to do, James?

James Vincent (6:17)

Well, humanoid robots, I mean, if you believe the pitch decks and the hype men, they're meant to do everything. They're meant to do anything that an able bodied human can do, do. They're meant to slot right into the workplace, they're meant to sort packages, they're meant to bolt on car doors, anything and everything. This is the pitch. This is why they are built like humans, because they want these creatures, these. Sorry, these machines, creatures. I know, I know, you know what

Sean Rainswerum (6:42)

I mean, you've been robot pilled.

James Vincent (6:44)

I've been robot pilled, yes. But they want them to do anything that a human laborer can do. And you know, that's a big ask.

Sean Rainswerum (6:51)

Who's trying, who's asking the robots to do it. All right, now a lot of companies

James Vincent (6:55)

in the US and in China mainly, these are the two leaders in the robotics space. It used to be mainly startups, I would say, but actually now we're seeing more of the big tech companies move into this space as well. Sometimes they're just doing little bits of research, you know, Meta recently Bought a robotics startup.

Ken Goldberg (7:12)

On May 1, Meta acquired a startup called Ari Assured Robot Intelligence. They build foundation models for humanoid robots designed to do physical labor in your home.

Sean Rainswerum (7:22)

Meta launched to be the Android, or Qualcomm, of the humanoid industry and power

Ken Goldberg (7:26)

the underpinnings, the software, the hardware, the artificial intelligence, the sensor stack, the compute

Sean Rainswerum (7:32)

for the whole industry.

James Vincent (7:33)

Google has been doing stuff with robots for ages. It's been, you know, testing its AI out on them.

Sean Rainswerum (7:38)

Google's DeepMind unit announcing a partnership with

Sponsor/Ad Voice (7:41)

Boston Dynamics as humanoid robotics steal the show at ces.

Sean Rainswerum (7:44)

They're really dexterous.

James Vincent (7:45)

They're doing some really intricate stuff like folding origami or like packing lunch. And Tesla, it's Elon Musk's obsession, or one of them. Alongside colonizing Mars.

Elon Musk (7:55)

My prediction is there'll be more robots than people.

James Vincent (7:58)

And he thinks that Optimus, which is the name of Tesla's robot, is going to be, you know, he makes all sorts of wild claims. It's going to be the most productive, the most profitable product ever invented.

Elon Musk (8:09)

I think everyone on Earth is going to have one and going to want one. Who wouldn't want a robot to, you know, assuming it's very safe, watch over your kids, take care of your pet.

James Vincent (8:20)

And I think this is, you know, typical Muskian hyperbole. But his interest is something that has moved the market hugely. And when he got involved, a lot of companies followed suit.

Sean Rainswerum (8:31)

Even though, as you write about in your cover story from Harper's from a few months ago, when he got involved, it was sort of anticlimactic and quite boring. But even that presentation in which I think a human dressed up as a robot, was like dancing to some Edmund. Yeah. It created perhaps the hype that we're seeing now.

James Vincent (8:52)

Yeah. So the event you're talking about was Tesla's AI Day. It does these sort of annual press jamborees. It shows off the new products and the roadmap to the public. This was in 2021, and prior to this, there had been increasing momentum and interest in humanoid robotics, helped very much by AI. But when Musk got involved, he tends to move companies or tries to move them very, very fast. And so he didn't have a fully working prototype to show the press, and instead, infamously, famously, he had an individual in a spandex robot suit who, as you say, came up and did a dance to EDM dubstep, dates it somewhat.

Sean Rainswerum (9:31)

Thank you.

James Vincent (9:33)

And he said, we're going to build these things.

Elon Musk (9:35)

We're setting it such that it is at a mechanical level, at a physical level, you can run away from it and most likely overpower it. So hopefully that doesn't ever happen, but you never know.

James Vincent (9:55)

And obviously he's been working on prototypes since that came out. They were very quick to produce a working machine prototype, but they're kind of running up against the limitations of the technology at the moment, I would say. And we're sort of waiting on the latest version of Optimus and that has been delayed several times. And I think we're beginning to see, okay, is this actually, is it pure or is there something coming out of it? And I think Musk and Tesla is very much being tested by this.

Sean Rainswerum (10:24)

But we are hearing a lot more about these robots now than we were back in 2021 when Musk held that symposium or whatever it was, right? I mean, why is it that we're seeing more of this stuff? Is it just because there are more robots now?

James Vincent (10:38)

The big reason for why we're having this moment for humanoids at the moment is AI, the chatgpt boom. And from deep learning from these technologies that have enable, like language models or chatbots, a lot of people have thought that this is a transferable technology that we can plug into humanoid machines and other machines and it can learn in the same way that chatbots have been able to learn to reproduce human speech. Now, the big thing that they're depending on is essentially robots. In the past, you had to program them manually. You had to say, move your, you know, your arm here down this many degrees across like this, and apply this much pressure. What you have with the new form of AI is that it learns these lessons by itself. Essentially, you plug in a lot of data, you give it an output that you want, and it learns how to connect those pieces together. The hope is from these companies that if we get enough data, we will, quote, unquote, solve the problem of physical robotics, and we will have these machines that are multi dextrous, capable of all these different tasks. And the big criticism of that is that robots are not in the same world as chatbots. Chatbots are dealing with text. If they make mistakes, you know, you talk to a chatbot even today, and it will still make mistakes every now and again. When those mistakes are transferred to the physical world, they suddenly become a lot more potentially dangerous. You know, a big thing that a lot of companies are doing at the moment is they're saying, we're going to put these robots in the home. They are going to be the perfect robot butler and they will take care of your dishes and Your laundry and all the rest of it. If a chatbot gets something wrong when you're asking it to do some research, then it's not the biggest deal in the world. You may spot the error and correct it. If a robot gets something wrong when it is cleaning away your plates and dishes, if it breaks one in every 10 cups, are you going to be happy with that sort of, that quality? No, I don't think so.

Sean Rainswerum (12:29)

Is the way China's developing these machines different than the way we are? Is the US Going for the more practical application, and is China going for maybe the more industrial?

James Vincent (12:42)

I would say that the main difference is that China's doing it faster and better. I think there is more of a focus in the US on home products as a sort of marketing to the rich and saying, look, we're going to take care of all these chores for you. In China, you have what is one of the fastest aging populations in the world. I think it's going to be over 60s are going to be predicted to be 30% of the population by 2040. So you have a loss of manufacturing labor and you have an increased burden on social care. And I think for Chinese state planners, humanoid robotics could very much plug in both of those gaps at the same time. So there is a slightly different focus, but it is one that is sort of organic in terms of emerging from the advantages of the Chinese economy. So the big thing that the Chinese economy has, the US doesn't, is scale. It has a massive ability to manufacture these units. It can make thousands at a time. This is why China is pulling ahead.

Sean Rainswerum (13:44)

I feel like you spent a lot of time in your piece trying to suss out the hype versus the reality. Where do you land? Is this going to be our reality within, what, a few years? Or is this still like, you know, flying cars or something of that sort?

James Vincent (14:02)

I think it's nearer to flying cars than it is to say the chatbot side of things, where we've seen really rapid advances. You know, there has been a legitimate leap forward in terms of capabilities. However, however, however, that does not mean that we are matching the hype that is being pushed out by people like Elon Musk, by other leading companies who are saying, we're going to have one of these robots in your house next year and it's going to be doing all the chores you need and will never make a mistake, and it certainly won't fall over and kill your cat or something like this. I think those promises are just. They're not true. They're Simply not true. I can see humanoid robots becoming a more common presence within both the work and the home over the next 10 plus years. Certainly, absolutely I can. But in the next five years, in the next three years, I really doubt it.

Sean Rainswerum (14:54)

So we don't have to yet worry that, let's say that robot you pushed over in Austin is gonna come back and kill you in London anytime soon.

James Vincent (15:03)

No, no, I'm planning on moving to the country where there's gonna be many muddy paths and sort of rivers that it has to cross over. English countryside is like, it's full of, you know, forests and things like this. The robots, they will never find me.

Sean Rainswerum (15:16)

You could just be playing ick ack ock with your friends quietly and be undisturbed.

James Vincent (15:21)

That's all we do.

Sean Rainswerum (15:28)

Saint James wrote Kicking Robots for Harpers. When we're back on Today Explained, we're gonna ask a guy who's been tinkering in robotics for decades. The essential question why? Robot. Support for today explained comes from 1-800-FLowers and who doesn't love flowers? If you're planning to send your mom flowers this Mother's Day, you might want a little help choosing the perfect bouquet to show her how much she means to you. 1-800-Flowers says that for more than 50 years they've been helping people send beautiful arrangements by perfecting the details that matter most. And if you're Prone to procrastination, 1-800-Flowers says that they've been saving their customers with same day delivery for years. Even if May 9th snuck up on you, they can still make May 10th count. That's a nice thought. Claire White did you try flowers?

James Vincent (16:25)

I took advantage of 1-800-flowers by a

Sponsor/Ad Voice (16:28)

dozen roses and get a dozen free

James Vincent (16:31)

deal for this Mother's Day.

Sponsor/Ad Voice (16:32)

I sent some over to my mom.

James Vincent (16:34)

They're beautiful, really well packaged and fresh, bright spring roses. And now I've got a bouquet for myself, which is just great to have.

Sean Rainswerum (16:44)

Mother's day is Sunday, May 10, just FYI. And bouquets are selling out fast. They say you don't have to wait to claim your double roses offer before they're gone. You can visit 1-800-flowers.com explained. That's 1,800-flowers.com explained 1,800flowers.com explained 1, 800-Flowers. A website. Support for the show today comes from npr. Love their work. Curious about the economic forces that shape your daily life. Then you can turn to the Planet Money podcast from npr. Friends of the show. It makes sense of the economy in ways you'll actually understand and enjoy Each story on Planet Money starts with a simple question. Recent episodes asked why Pokemon cards are growing faster than your retirement account. How Russia's economy hasn't collapsed after four years of war and sanctions. And what a 750 pound walk robot means for the future of restaurant work. Wow, I want to listen to that one. Maybe all of them. You should too. The hosts of Planet Money go to unusual lengths to explain the economy. From publishing their own book. I have a copy. To Tracking the Global Supply Chain. I would like someone to do that for me.

James Vincent (17:52)

They did it.

Sean Rainswerum (17:53)

Oh, I love that. I know. They're talking. The T shirt. You guys remember the T shirt thing? The T shirt thing's really good. If you haven't heard it. It's the kind of show where you learn something, probably laugh and walk away seeing the world a little differently. Simply put, its econ down to earth. Love shows that do that. Follow NPR's Planet Money podcast. Understand how money shapes the world.

Sponsor/Ad Voice (18:16)

Ready to soundtrack your summer with Red Bull Summer All Day Play. You choose a playlist that fits your summer vibe the best. Are you a festival fanatic, a deep end dj, a road dog, or a trail mixer? Just add a song to your chosen playlist and put your summer on track. Red Bull Summer All Day Play. Red Bull gives you wings. Visit RedBull.com BrightSummerAhead to learn more. See ya this summer.

Sean Rainswerum (18:42)

You're listening to Today Explained.

Ken Goldberg (18:51)

My name is Ken Goldberg, professor of engineering at UC Berkeley. Or professor, actually. Let's do this. I'd actually rather be called a professor of robotics at UC Berkeley.

Sean Rainswerum (19:00)

That's a better.

Ken Goldberg (19:01)

Yeah, that sums it up a little better. I'm also an artist, if that matters.

Sean Rainswerum (19:04)

But yeah, it does. It does.

Ken Goldberg (19:06)

Okay.

Sean Rainswerum (19:06)

And I'm glad you made the clarification on the robotics side because as far as I can tell, you've been into robots since before they were. I don't want to say cool. Cause something about robots has always been cool to us as humans. But at least since before they were so trendy as they are now.

Ken Goldberg (19:22)

Yeah, well, I've been interested in robots since I was a kid, but that was back in the days of the Jetsons in the 60s.

Sponsor/Ad Voice (19:30)

Here I am, sir. Yes, sir.

Ken Goldberg (19:34)

When, you know, I, like most kids, were just fascinated by the science fiction.

Sponsor/Ad Voice (19:38)

What are you?

Ken Goldberg (19:39)

I am a robot of the class M3. Programmed to provide information and support to all Jupiter personnel. My father was a tinkerer and an engineer also. He and I both, I built a robot for. It never worked, but it was ahead of its time back in the 70s, and I got interested in this. And when I was in college, there was a robot lab that I discovered. It used pen, and I got very excited. So I built a tactile sensor, a hand that could feel, to basically have it feel around for objects and try and determine what their shape was. I did some work on running machines, and then I started working on grasping in particular. I've been really, like, I like to say I've been studying the same problem for 45 years, which is how to pick up a glass. It turns out that's incredibly difficult to do reliably.

Sean Rainswerum (20:28)

Why is it so difficult?

Ken Goldberg (20:30)

All right. Well, it all boils down to one word, uncertainty. Robots are uncertain. They are uncertain in their perception, so they can't see the glass very well, and they can't position where it is exactly in space. They have uncertainty and control, so they can't move their gripper exactly into a particular location in space, even if they could see it. And then there's a third uncertainty, which I call uncertainty in physics, which is that we don't know the frictional properties, the mass properties of the object, which actually turns out to be very important if you're going to grasp it. And all those conspire together, so they add up at the fingertips. So that's why robots today are better, but they still often are very clumsy.

Sean Rainswerum (21:09)

And maybe you can help us understand, as someone who's been at this since you were tinkering in a garage, it sounds like, with your dad in the 70s, why is it so important that a humanoid robot learn how to pick up a glass? Because here we are, people who can pick up a glass, for the most part.

Ken Goldberg (21:31)

Yeah. No, we do it very well, very naturally. We do it so effortlessly that it doesn't seem hard to us at all. Playing go, on the other hand, seems extremely difficult. Right.

Sean Rainswerum (21:41)

There was an actual discussion whether computers could ever beat a grandmaster at chess. They have that same discussion about. About go.

Elon Musk (21:49)

Right.

Sean Rainswerum (21:49)

Well, GO is much, much more complicated than chess.

James Vincent (21:54)

So chess was solved, was cracked sort of 20 years ago, and then, since then, go has been one of the holy grails for AI research.

Ken Goldberg (22:03)

This is Moravec's paradox, that the things that are easy for us are hard for robots, and that AI is very good at playing go. That's hard for us. Right? So it is a paradox. And yes, we do it, but we've evolved over 300 million years.

Sean Rainswerum (22:22)

And why do you think it's so important that we help these humanoid robots figure out how to pick up a glass just for the sake of doing it, or is there some greater intention here? I mean, this is your life's work.

Ken Goldberg (22:32)

Yeah, yeah. So there's two schools of thought. There's a sort of scientific approach which is like, how do we understand this in the same sense as why do we understand the distant nebula? It's a scientific question. But then there's the engineering question, which I'm more of an engineer, and it's like, let's get something to work. We want these things to help us do work that we don't want to do. And in particular, things that are drudgery. For example, delivering packages faster so that we can get the stuff we want from our favorite vendor as soon as possible. So there's a shortage of workers. And this is another misconception. People are afraid that robots are going to put all the humans out of work. I don't believe that. I actually think there's a shortage of humans. Especially humans are willing to do a lot of the drudgery work in warehouses and factories and farms and cleaning hotel rooms, et cetera. And so robots could be very helpful there. And to do almost any of those jobs, the first thing they want is a robot to be able to grasp things. And so that's a very fundamental task.

Sean Rainswerum (23:30)

I mean, devil's advocate, though. I think a lot of people like their jobs at Costco, stocking shelves that pay well and afford people health care. I mean, what is that person gonna do when a robot replaces them in, I don't know, 10 years?

Ken Goldberg (23:45)

Let me just say I'm very sympathetic to that. I really do have a lot of concerns about the humans in this world and workers in particular. And I don't think robots are out to steal their jobs, by the way. I think that's an old fear that people have about immigrants, and they now attach that to robots. I think that the bigger question is that we want machines to be able to do things to increase productivity. There's a lot of reasons where that's really desirable, and many, many companies want it. Actually, what seems to happen historically is that as new technologies come in, like the car, we thought it would wipe out all the blacksmiths, for example, but it turns out that actually it produced many more jobs, Just new jobs, like jobs like building roads and fixing cars and selling cars, making cars. The unemployment rate has stayed relatively constant over the past century through all these inventions. The airplane, the tv, the computer, the Internet. So I'm not an economist, but I'll just say, coming back to why we want robots. It is something that we would love to have more capabilities. And there's been some changes in the last few years that make people more excited than they were before.

Sean Rainswerum (24:55)

Help us understand why it's going to take so long. I mean, why do we have maybe self driving cars already but not humanoid robots that can competently pick up a glass?

Ken Goldberg (25:09)

Okay, good. So let's talk about that one thing is that self driving cars, remember it's taken 20 years to get to the point of feasibility and it still may not be cost effective yet.

Sean Rainswerum (25:18)

Right.

Ken Goldberg (25:18)

The reason it's a lot easier is that there you just essentially want to avoid hitting anything. That's the goal. And you have pretty good maps, you have a lot of other good information and you can simulate that pretty well. Turns out that locomotion, which is walk ins, is similar there again, you want to avoid stuff, you don't want to bump into things, but you want to do a backflip that turns out to be pretty easy to simulate. And everybody sees that and they project ahead. It's understandable. It's human nature. Oh, robots can do this better than me. So therefore they can do everything better than me. And that is a fallacy because they can't stack a little group of blocks, sugar cubes, on your table very reliably at all and they can't pick up that glass. You see a lot of hands that look very human like, they have very fast moving joints etc and they look pretty human like. But do they do anything interesting? I mean, how many times have you seen a robot hand tying up a sneaker, you know?

Sean Rainswerum (26:10)

Zero.

Ken Goldberg (26:10)

You haven't. That's right. And then sometimes you'll see them doing some tasks, maybe folding some laundry. But if you look closely, as my wife says, this is not acceptable laundry folding, you know. So that's what I'm getting at, which is these are very subtle skills. And we know that robots could do them with very simple grippers. And the existence proof for that, Sean, is robot surgery. And I really want to make it clear. People say, well, a robot took out my daughter's appendix. Well, that was a robot operated by a human. So there's a human driving that robot right. Now the robot tools are very simple, just two grippers, but they can do incredible things when there's a human behind the wheel. But if you try and take the human out of the loop, as we've tried to do in our lab, then everything becomes much more difficult.

Sean Rainswerum (26:56)

Do you see people in this field of robotics asking often if something should be done, or is it mostly just questions of if something could be done?

Ken Goldberg (27:06)

That's a very good, very, very deep question. And I think most of the engineers, the world I live in, people are asking if it can be done. I do have some concerns, certainly about robots being used as weapons. You know, that that's changed the landscape of warfare all over the world right now. And there's some very deep questions. But in the labs, in research labs, people are just trying to figure out how to get the robots to actually perform measurably better. But at the same time, it's not going to be possible. I do not believe you can say, okay, well, we should have a moratorium on research or stop the research or anything like that, because someone else will do it. But I think the question is, how can we be thoughtful? And there are people working on robot safety, which, by the way, turns out to be super important if you want to start a company, because you have to work with OSHA standards, et cetera. So that is a big issue. Let's see what happens. I'm an optimist, Sean. I really am. I think this technology is really interesting. It's moving in interesting directions, but I don't think it's going to wipe us out. And I don't think we have to worry about, you know, the robo apocalypse.

Sean Rainswerum (28:23)

Ken Goldberg is a robotics professor at UC Berkeley and an artist. Avis Hay Artsy made our show today. Julie Meyers edited. David Tadashiori mixed. Gabriel Dunatov checked the facts, and Sean Rainswerum was the host for this episode of Today explained.