Transcription
But since I was immersed in technology, I have been immersed in technology for a long time. I could see it coming, and I do think overall that the potential is there for artificial intelligence (AI) to have most likely a positive effect, to create a future of abundance where there is no scarcity of goods and services.
However, it is somewhat the magic genie problem. If you have a magic genie that can grant all the wishes, usually those stories don’t end well. Be careful what you wish for. Including wishes, really anything that can be actuated by a computer is effectively a robot. So you can think of, frankly, Tesla cars as robots on wheels. Anything that’s connected to the internet is effectively an endpoint actuator for artificial intelligence.
I do think there is a safety concern, especially with humanoid robots because, at least the car can’t chase you into this building—not very easily, you know—or chase you up a tree. You can sort of run up a flight of stairs and get away from a Tesla. But if you have a humanoid robot, it can basically chase you anywhere.
So I think we should have some kind of hardwired local cutoff that you can’t update from the internet. Anything that can be software updated from the internet obviously can be overridden. But if you have a local sort of off switch where you have to say a keyword or something, and then that puts the robot into a safe state—some kind of localized safe state ability and off switch—you know, where you don’t have to get too close to the robot.
I don’t know. If you’ve got millions of these things going all over the place, you’re not selling it just, you know, like no. I know I’m saying this is something we should be quite concerned about. Because if a robot can follow you anywhere, then what if they just one day get a software update and they’re not so friendly anymore? Then we’ve got a James Cameron movie on our hands.
What would your kind of observation be on AI and the impact on labor markets and people’s jobs, and how they should feel about that as they think about this?
Well, I think we are seeing the most disruptive force in history here. You know, for the first time, we will have something that is smarter than the smartest human. I mean, it’s hard to say exactly what that moment is, but there will come a point where no job is needed. You can have a job if you want to have a job for sort of personal satisfaction, but the AI will be able to do everything.
So I don’t know if that makes people comfortable or uncomfortable. That’s why I say if you wish for a magic genie that gives you any wishes you want and there’s no limit—you don’t have those three wish limit nonsense—you just have many, many wishes as you want. It’s both good and bad.
One of the challenges in the future will be how do we find meaning in life if you have a magic genie that can do everything you want? It’s hard. You know, when there’s new technology, it tends to usually follow an S curve. In this case, we’re going to be on the exponential portion of the S curve for a long time. You’ll be able to ask for anything.
We won’t have Universal Basic Income; we’ll have Universal High Income. So in some sense, it’ll be somewhat of a leveler or an equalizer. You know, because really, I think everyone will have access to this magic genie.
Do you have a sense yet of where you think the big applications will be from AI? Is it going to be enabling self-driving? Is it going to be enabling robots? Is it transforming industries? I mean, it’s still, I think, early in terms of where the big business impact is going to be. Do you have a sense yet?
I think the spending on AI probably runs ahead of—I mean, does run ahead of the revenue right now. There’s no question about that. The rate of improvement of AI is faster than any technology I’ve ever seen, by far. I mean, for example, the Turing test used to be a thing. Now, you know, your basic open-source random LM writing on a freaking Raspberry Pi probably could beat the Turing test.
I think actually the good future of AI is one of immense prosperity, where there is an age of abundance—no shortage of goods and services. Everyone can have whatever they want, except for things we artificially define to be scarce, like some special artwork. But anything that is a manufactured good or provided service will, I think, with the advent of AI plus robotics, trend toward zero.
I’m not saying it will be actually zero, but everyone will be able to have anything they want. That’s the good future, and in my view, that’s probably 80% likely.
So look on the bright side. Is the 20% like what does that look like? I don’t know, man. I mean, frankly, I do have to engage in some degree of deliberate suspension of disbelief with respect to AI in order to sleep well. Because I think the actual issue—the most likely issue—is like, well, how do we find meaning in a world where AI can do everything we can do a bit better? That is perhaps the bigger challenge.
You need the sort of end factors; you need the autonomous cars, and you need the humanoid robots or general-purpose robots. But once you have general-purpose humanoid robots and autonomous vehicles, you can build anything. If you’ve got humanoid robots that can do, you know, where there’s no real limit on the number of humanoid robots, and they can operate very intelligently, then there’s no actual limit to the economy.
What’s the role of humanoid robotics, do you think? Is Tesla’s approach ridiculous? Do you think it’s super promising? Do you think it’s interesting, full of mystery? Nobody knows. What do you think about this thing?
Yeah, I think today humanoid form robotics is research. There are very few situations where you actually need a humanoid form to solve a problem. If you think about it, right, wheels are more efficient than legs. There are joints and degrees of freedom beyond a certain point that just add a lot of complexity and cost.
So if you’re doing a humanoid robot, oftentimes it’s in the pursuit of a humanoid robot, not in the pursuit of an application for the time being, especially when you have like kind of the gaps in interface and, you know, kind of AI that we talk about today.
So anything Elon does, I’m interested in following. There’s an element of that where no matter how crazy it is, I just like, you know, I’ll pay attention. I’m curious to see what comes out of it. So it’s like you can’t ever ignore it. But, you know, it’s definitely far afield from their kind of core business.
To me, the compelling aspect of the humanoid form and a lot of kind of robots, Cosmo for example, is the human-robot interaction part. Elon Musk’s perspective is that Tesla bot has nothing to do with the human; it’s a form that’s effective for the factory because the factory is designed for humans.
But to me, the reason you might want to argue for the humanoid form is because, you know, at a party, it’s a nice way to fit into the party. The humanoid form has a compelling notion to it in the same way that Cosmo is compelling.
I would argue if we were arguing about this that it’s cheaper to build a Cosmo-like form. But if you wanted to make an argument—which I have with Jim Keller—about, you know, you could actually make a humanoid robot for pretty cheap.
Then the question is, all right, if you are using an application where it can be flawed, it can have a personality and be flawed in the same way that Cosmo is, then maybe it’s interesting for integration into human society.
That’s to me is an interesting application of a humanoid form because humans are drawn, like I mentioned, to legged robots, drawn to legs and limbs and body language and all that kind of stuff. And even a face, even if you don’t have the facial features, which you might not want to have for not to reduce the creepiness factor, all that kind of stuff.
But yeah, that to me, the humanoid form is compelling. But in terms of that being the right form for the factory environment, I’m not so sure.
I think with Tesla and the real-time video coming from several million cars—ultimately tens of millions of cars—with Optimus, they might have hundreds of millions of Optimus robots, maybe billions, learning and training us not from the real world. That’s the biggest source of data.
I think ultimately, Optimus is going to be the biggest source of data because reality scales to the scale of reality. It’s actually humbling to see how little data humans have actually been able to accumulate.
Really, say how many trillions of usable tokens have humans generated? On a non-duplicative basis—like discarding spam and repetitive stuff—it’s not a huge number. You run out pretty quickly.
Optimus can go anywhere and go off-road. I mean, like Optimus can pick up a cup and see, did it pick up the cup in the right way? Did it, you know, pour water in the cup? Did the water go in the cup or not go in the cup? Did it spill water or not?
Yeah, simple stuff like that. I mean, but it can do that at scale times a billion, generating useful data from reality. So it’s cause and effect stuff.
What do you think it takes to get to mass production of humanoid robots like that?
It’s the same as cars, really. I mean, global capacity for vehicles is about 100 million a year. It could be higher; just that the demand is on the order of 100 million a year.
Then there are roughly two billion vehicles that are in use in some way. So which makes sense. The life of a vehicle is about 20 years. So at steady state, you can have 100 million vehicles produced a year with a 2 billion vehicle fleet, roughly.
Now for humanoid robots, the utility is much greater. So my guess is humanoid robots are more like at a billion plus per year. But, you know, until you came along and started building Optimus, it was thought to be an extremely difficult problem.
I mean, it’s still an extremely difficult problem—not a walk in the park. I mean, Optimus currently would struggle to walk in the park. I mean, it can walk in a not-too-difficult environment, but it will be able to walk over a wide range of terrain.
Yeah, there’s going to be an immense amount of engineering just going into the hand. The hand might be close to half of all the engineering in Optimus from an electromechanical standpoint.
The hand is probably roughly half of the engineering, but so much of the intelligence—so much of the intelligence of humans—goes into what we do with our hands. Yeah, like the manipulation of the world, manipulation of objects in the world, intelligent safe manipulation of objects in the world.
What do you think about Optimus, about Tesla bot? Do you think we’ll have robots in the factory and in the home in 10, 20, 30, 40, 50 years?
Yeah, I think it’s a very hard project. I think it’s going to take a while. But who else is going to build humanoid robots at scale?
I think it is a very good form factor to go after because, like I mentioned, the world is designed for the human form factor. These things would be able to operate our machines; they would be able to sit down in chairs, potentially even drive cars. Basically, the world is designed for humans. That’s the form factor you want to invest in and make work over time.
I think, you know, there’s another school of thought which is, okay, pick a problem and design a robot for it. But actually designing a robot and getting a whole data engine and everything behind it to work is actually an incredibly hard problem.
So it makes sense to go after general interfaces that, okay, they are not perfect for any one given task, but they actually have the generality of just with the prompt, with English, able to do something across.
So I think it makes a lot of sense to go after a general interface in the physical world. I think it’s a very difficult project; it’s going to take time. But I see no other company that can execute on that vision.
I think it’s going to be amazing. Like, basically, physical labor—if you think transportation is a large market, try physical labor. It’s like insane.
If you were to go with the goal of like, okay, let’s build a million humanoid robots and you’re not Tesla, that’s a lot to ask. If you’re Tesla, it’s actually not that crazy.
I think the ratio of humanoid robots to humans will probably be at least 2 to 1, something like that—1 to 1 for sure. So which means somewhere on the order of 10 billion humanoid robots.
I think the build rate will be probably something ultimately like a billion a year for humanoid robots. If Tesla just has a 10% share of that—and it might be a lot more than 10%—and we make like 100 million Optimus units a year, I think we could make one for a cost of maybe, at a really high scale, about $10,000.
It’s more—it’d be less expensive than a car. I think if you sell it for $20,000, something like that, at large scale volume, Tesla would basically make about a trillion dollars of profit a year from that.
Your prediction was that AI was increasing at a rate as fast as 100 times per year and that by 2029 or 2030, we might see AI as capable as 8 billion humans. Are you still seeing that pace?
Yeah, it’s been a difficult thing to quantify exactly, but I certainly feel comfortable saying that it’s getting 10 times better per year. Which is, you know, four years from now, that would mean 10,000 times better—maybe 100,000.
I think it will be able to do anything that any human can do, possibly within the next year or two. How much longer than that does it take to be able to do what all humans can do combined? I think not long.
So like 2029, 2028, something like that. Yeah, I mean, one can say it’s most likely going to be great, and there’s some chance, which could be 10 to 20%, that it goes bad. The chance is not zero that it goes bad, but overall, one could say the cup is 80% full.
That’s one positive way to look at it—maybe 90%.
Speaking of robots, everything we’ve developed for our cars—the batteries, power electronics, the advanced motors, gearboxes, the software, the AI inference computer—it all actually applies to a humanoid robot. It’s the same techniques; it’s just a robot with arms and legs instead of a robot with wheels.
We’ve made a lot of progress with Optimus, and as you can see, we started up with someone in a robot suit, sort of, and then we’ve progressed dramatically year after year.
So if you extrapolate this, you’re really going to have something spectacular—something that anyone could own. So you can have your own personal R2-D2, C-3PO.
I think at scale, this would cost something like, I don’t know, $20,000 to $30,000—probably less than a car is my prediction long term. You know, it’ll take us a minute to get to the long term, but fundamentally, at scale, the Optimus robot, you should be able to buy for I think probably $20,000 to $30,000 long term.
And what can it do? It can do anything you want. So it can be a teacher, babysit your kids, walk your dog, mow your lawn, get the groceries, just be your friend, serve drinks—whatever you can think of, it will do.
Yeah, it’s going to be awesome, and I think this will be the biggest product ever of any kind.