Humanoid robots may no longer need remote controls.
Over time, we can give commands to robots directly through voice.
The cost of training robots is lower than training humans.
This means that tens of millions or even hundreds of millions of blue-collar workers will simultaneously lose their comparative advantage.
It also represents a market worth tens of trillions or even hundreds of trillions.
With this prospect, venture capitalists will flock in.
So instead of focusing on the breakthrough in humanoid robots, we should focus on the end-to-end breakthrough in embodied AI.
In fact, looking at GPT-4 currently, our intelligence ceiling has already overflowed. Most embodied AI does not require a college-level standard.
What we lack is alignment technology and data. These all require money and time.
I feel that the GPT moment for embodied AI should be coming soon.
Optimistically speaking, a company like OpenAI for embodied AI has already emerged, but we just don't know which one it is. Perhaps by this time next year, we will welcome the GPT moment, and then begin exponential growth.
Indeed! I think you're right. Humanoid robots might be senseless. Maybe the best approach turns out to be a wheeled robot like Roomba, or a legged dog-type robot like those of BD. Data will be very necessary, and I think that main problem for embodied robots might be open-ended learning tasks, which might be a bit separate from this LLM world.
This approach reminds me of Brooks' famous paper "Elephants don't play chess" (https://www.sciencedirect.com/science/article/abs/pii/S0921889005800259) which contributed to changing the approach to AI, from GOFAI towards deep learning. It is probably a good path for robots to be exposed to all available sources of learning, not only interacting in the real world of laboratories but also "watching" real videos of humans moving through reality.
Yeap, that's a great article, and Brooks is a highly skeptical and brilliant researcher in robotics. Research papers might prove whatever you wish, and it ill seem that it's technically feasible to do anything with robots in a laboratory. But then, it's the economy stupid. Despite the hype of many papers and media, scalability and robustness of those systems to be industrially commercialized is not an easy step!
After GPT emerged, embodied AI became possible.
Humanoid robots may no longer need remote controls.
Over time, we can give commands to robots directly through voice.
The cost of training robots is lower than training humans.
This means that tens of millions or even hundreds of millions of blue-collar workers will simultaneously lose their comparative advantage.
It also represents a market worth tens of trillions or even hundreds of trillions.
With this prospect, venture capitalists will flock in.
So instead of focusing on the breakthrough in humanoid robots, we should focus on the end-to-end breakthrough in embodied AI.
In fact, looking at GPT-4 currently, our intelligence ceiling has already overflowed. Most embodied AI does not require a college-level standard.
What we lack is alignment technology and data. These all require money and time.
I feel that the GPT moment for embodied AI should be coming soon.
Optimistically speaking, a company like OpenAI for embodied AI has already emerged, but we just don't know which one it is. Perhaps by this time next year, we will welcome the GPT moment, and then begin exponential growth.
Indeed! I think you're right. Humanoid robots might be senseless. Maybe the best approach turns out to be a wheeled robot like Roomba, or a legged dog-type robot like those of BD. Data will be very necessary, and I think that main problem for embodied robots might be open-ended learning tasks, which might be a bit separate from this LLM world.
The economic numbers of the value and revenue of a single robots that Benjie Holson proposed in this post also got me thinking https://generalrobots.substack.com/p/humanoid-robots-dollars-and-gpts. It will be not easy.
Thanks a lot for posting!
This approach reminds me of Brooks' famous paper "Elephants don't play chess" (https://www.sciencedirect.com/science/article/abs/pii/S0921889005800259) which contributed to changing the approach to AI, from GOFAI towards deep learning. It is probably a good path for robots to be exposed to all available sources of learning, not only interacting in the real world of laboratories but also "watching" real videos of humans moving through reality.
Yeap, that's a great article, and Brooks is a highly skeptical and brilliant researcher in robotics. Research papers might prove whatever you wish, and it ill seem that it's technically feasible to do anything with robots in a laboratory. But then, it's the economy stupid. Despite the hype of many papers and media, scalability and robustness of those systems to be industrially commercialized is not an easy step!
Thanks a lot for commenting.