Speaking of rare materials like Cobalt, I doubt we could ever source enough of it to build batteries for 10 billion humanoids. We know the planet is in decline and natural resources are depleting, yet these techbros seem oblivious. Do you ever get the sense that they know their robotics and AI projects are bullshit but they're hoping to sell enough shovels and pipedreams to get away with it?
Great article! There are a lot of layers to this, from corporate greed to lack of innovative vision, but the technical issue that makes the sim-to-real gap extremely hard--regardless of the politics or the economics of it--, I think, is probably something similar to the issue of hallucinations. ML systems trained via backprop need to come up with differentiable approximations of the input space to be able to generalize at all. This means LLMs need to map sentences to vectors, for example, and compare vector distances to determine what a sentence "means".
In the same sense, these robots need to map the world to a vectorial representation to determine what is going on in the kitchen. But smooth, differentiable representations can only be an approximate model of macroscopic reality, and is in that tiny difference between the smooth proxy and the non-continuous reality you want to model that problems arise. For robots, this means tiny discrepancies between their prediction of the outcome of a decision and the actual outcome, that get amplified the longer the prediction chain.
This is why simple, reactive agents like Roombas work so well for they restricted domain, they are approximating a much smaller part of the real world (just flat floors with static obstacles) and they can self-correct fast because the model of the world they have is super simple, just a few parameters (plus they are not learning agents, they are hard-coded, I believe). Humanoid robots are supposed to deal with the whole chunk of reality humans have evolved over eons to represent in our brains accurately, and I think we do need symbolic world models with causal inference rules to model it accurately. I don't know if backprop can lead us all the way there.
(Sorry for the jargon, it's early and my brain is still too dumb to make this more intelligible.)
This is such a useful explaination, thanks. I tried to make the bridge with hallucinations but wasn’t able to figure out the right way to describe it (so I left it out).
It reminds me of a quote you might be familiar with: “All models are wrong but some are useful.”
The robot ‘thoughts’ we had for today remind me of the 1950’s thoughts of futuristic technologies that never came about, such as weather control. Over time, we’ve gained a better understanding that such ideas are likely impossible due to ever changing situations. Or, the technologies were eventually overshadowed by more practical innovations (smartphones replacing wristwatch TVs or email overtaking jetpack delivery.) When it comes to robots, you make a valid point: they’re not particularly useful to most consumers. As well, the technology may be cost prohibitive. I suspect since child labor and coal-mining is still performed by humans instead of robots, maybe human thinking is getting in the way – or, sadly, cheap labor is more important to some than robot technology.
We all watched the Jetson's and envisioned a life with Rosie the Robot. We watched Star Trek and saw how easy it was to use the Replicator and get Shrimp Scampi or what ever you craved in seconds. We thought that the future of robots would be to do the more menial, physically challenging and lower paid tasks for us all. These tasks are much more complex than we imagined. I keep hoping everyone has read Asimov though.
Every day I’m pushed closer to concluding AI & robots are being promoted primarily to attract investors and to fool people into thinking the elites are making a fabulous Jetsons future for us, also to send human peons the message that they’re highly inferior to robots and will be replaced by them. In reality, the elites are poisoning all life while telling us nothing about it—such as the Chernobyl-sized nuclear accident in Ukraine last year. Never even made the news cycle. And Fukushima’s radiation levels are increasing! They won’t tell us, because if we knew we’d say, “Why are you wasting all our time & money on robots and AI when the PLANET is dying? Why won’t you take care of THAT with your stolen trillions and massive Covid profits?”
Robots are another case where the promise far outweighs the performance. And anyways, even if they do get some form of robots into society – I always think 'I, Robot' – they'll be a tool for the rich and elite.
I think the "ChatGPT moment" analogy is very apt: ChatGPT was the moment for where corporate cynicism reached a historical peak. A non-product without purpose serving no benefit for the value proposed, marketed not at consumers, but investors and only investors - for trillions of dollars. "Look how excited it makes the little idiots - give us more money, they will lap it all up".
These robots aren't targeting any real problems and aren't aimed towards real people. Entirely a marketing stint for a pretense of a future for the benefit of investors. A simulation of the future being sold in exchange for a promise of endless profits.
There are no dirty jobs in that future, you see. In that world where their robots do the dishes for the ordinary plebs cobalt just pops out of the ground, coffee flows freely out of a tap, energy is free, and so is the drinking water. If we don't acknowledge child labour today - it won't be there tomorrow... At worst we can wave it off and graciously allow it to be someone else's problem.
We all watched the Jetson's and envisioned a life with Rosie the Robot. We watched Star Trek and saw how easy it was to use the Replicator and get Shrimp Scampi or what ever you craved in seconds. We thought that the future of robots would be to do the more menial, physically challenging and lower paid tasks for us all. These tasks are much more complex than we imagined. I keep hoping everyone has read Asimov though.
Speaking of rare materials like Cobalt, I doubt we could ever source enough of it to build batteries for 10 billion humanoids. We know the planet is in decline and natural resources are depleting, yet these techbros seem oblivious. Do you ever get the sense that they know their robotics and AI projects are bullshit but they're hoping to sell enough shovels and pipedreams to get away with it?
Great article! There are a lot of layers to this, from corporate greed to lack of innovative vision, but the technical issue that makes the sim-to-real gap extremely hard--regardless of the politics or the economics of it--, I think, is probably something similar to the issue of hallucinations. ML systems trained via backprop need to come up with differentiable approximations of the input space to be able to generalize at all. This means LLMs need to map sentences to vectors, for example, and compare vector distances to determine what a sentence "means".
In the same sense, these robots need to map the world to a vectorial representation to determine what is going on in the kitchen. But smooth, differentiable representations can only be an approximate model of macroscopic reality, and is in that tiny difference between the smooth proxy and the non-continuous reality you want to model that problems arise. For robots, this means tiny discrepancies between their prediction of the outcome of a decision and the actual outcome, that get amplified the longer the prediction chain.
This is why simple, reactive agents like Roombas work so well for they restricted domain, they are approximating a much smaller part of the real world (just flat floors with static obstacles) and they can self-correct fast because the model of the world they have is super simple, just a few parameters (plus they are not learning agents, they are hard-coded, I believe). Humanoid robots are supposed to deal with the whole chunk of reality humans have evolved over eons to represent in our brains accurately, and I think we do need symbolic world models with causal inference rules to model it accurately. I don't know if backprop can lead us all the way there.
(Sorry for the jargon, it's early and my brain is still too dumb to make this more intelligible.)
This is such a useful explaination, thanks. I tried to make the bridge with hallucinations but wasn’t able to figure out the right way to describe it (so I left it out).
It reminds me of a quote you might be familiar with: “All models are wrong but some are useful.”
The robot ‘thoughts’ we had for today remind me of the 1950’s thoughts of futuristic technologies that never came about, such as weather control. Over time, we’ve gained a better understanding that such ideas are likely impossible due to ever changing situations. Or, the technologies were eventually overshadowed by more practical innovations (smartphones replacing wristwatch TVs or email overtaking jetpack delivery.) When it comes to robots, you make a valid point: they’re not particularly useful to most consumers. As well, the technology may be cost prohibitive. I suspect since child labor and coal-mining is still performed by humans instead of robots, maybe human thinking is getting in the way – or, sadly, cheap labor is more important to some than robot technology.
We all watched the Jetson's and envisioned a life with Rosie the Robot. We watched Star Trek and saw how easy it was to use the Replicator and get Shrimp Scampi or what ever you craved in seconds. We thought that the future of robots would be to do the more menial, physically challenging and lower paid tasks for us all. These tasks are much more complex than we imagined. I keep hoping everyone has read Asimov though.
Asimov is my favorite sci-fi writer and a true visionary. One of my favorite quotes is: “Sizzling Saturn! We got a lunatic robot on our hands…”
Every day I’m pushed closer to concluding AI & robots are being promoted primarily to attract investors and to fool people into thinking the elites are making a fabulous Jetsons future for us, also to send human peons the message that they’re highly inferior to robots and will be replaced by them. In reality, the elites are poisoning all life while telling us nothing about it—such as the Chernobyl-sized nuclear accident in Ukraine last year. Never even made the news cycle. And Fukushima’s radiation levels are increasing! They won’t tell us, because if we knew we’d say, “Why are you wasting all our time & money on robots and AI when the PLANET is dying? Why won’t you take care of THAT with your stolen trillions and massive Covid profits?”
Robots are another case where the promise far outweighs the performance. And anyways, even if they do get some form of robots into society – I always think 'I, Robot' – they'll be a tool for the rich and elite.
I think the "ChatGPT moment" analogy is very apt: ChatGPT was the moment for where corporate cynicism reached a historical peak. A non-product without purpose serving no benefit for the value proposed, marketed not at consumers, but investors and only investors - for trillions of dollars. "Look how excited it makes the little idiots - give us more money, they will lap it all up".
These robots aren't targeting any real problems and aren't aimed towards real people. Entirely a marketing stint for a pretense of a future for the benefit of investors. A simulation of the future being sold in exchange for a promise of endless profits.
There are no dirty jobs in that future, you see. In that world where their robots do the dishes for the ordinary plebs cobalt just pops out of the ground, coffee flows freely out of a tap, energy is free, and so is the drinking water. If we don't acknowledge child labour today - it won't be there tomorrow... At worst we can wave it off and graciously allow it to be someone else's problem.
We all watched the Jetson's and envisioned a life with Rosie the Robot. We watched Star Trek and saw how easy it was to use the Replicator and get Shrimp Scampi or what ever you craved in seconds. We thought that the future of robots would be to do the more menial, physically challenging and lower paid tasks for us all. These tasks are much more complex than we imagined. I keep hoping everyone has read Asimov though.