I like this list. Some other nonexclusive possibilities:
General purpose robotics need very low failure rates (or at least graceful failure) without supervision. Every application which has taken off (ChatGPT, Copilot, Midjourney) has human supervision, so failure is ok. So it is an artifact of none of AI handling failure well, rather than something to do with robots. Predictions:
—Even non-robot apps intended to have zero human supervision will have problems, i.e., maybe why adept.ai hasn’t launched?
Most of this progress is in SF. There’s just more engineers good at HPC and ML than at robots, and engineers are the bottleneck anyhow.
—Predicts Shenzhen or somewhere might start to do better.
I like this list. Some other nonexclusive possibilities:
General purpose robotics need very low failure rates (or at least graceful failure) without supervision. Every application which has taken off (ChatGPT, Copilot, Midjourney) has human supervision, so failure is ok. So it is an artifact of none of AI handling failure well, rather than something to do with robots. Predictions: —Even non-robot apps intended to have zero human supervision will have problems, i.e., maybe why adept.ai hasn’t launched?
Most of this progress is in SF. There’s just more engineers good at HPC and ML than at robots, and engineers are the bottleneck anyhow. —Predicts Shenzhen or somewhere might start to do better.