Discussion about this post

User's avatar
Frank Tobe's avatar

What you wrote is partially true. It doesn't roll out like the optimists suggest but it WILL happen. Robots WILL be built to adapt to human form just because the world is predicated on human form. Door widths, elevators, heights, etc. They WILL assist us in various ways AND it will take time.

But... did you happen to go to the Humanoids Summit in Menlo Park this December 11th and 12th? It was an interesting show with 50-60 exhibitors. And it explained why there's a long wait before any really marketable vertical markets will employ humanoids: data. Data with which to build large models that not only provide and communicate knowledge but provide additional data with the physical intelligence for planning, controlling and utilizing the physics involved in actions a humanoid might take to perform a needed task from a library of tasks.

At the Summit the speakers acknowledged that until there were sufficient models for autonomy in actions growth would be small and incremental. Iterations would be plentiful. And data building would be costly with significant competition for the money it will take. Big American, European and Chinese players like Tesla will foot the bill for their companies but not be open sourced.

But it will happen. Humanoids will be in our lives if we live long enough. Just not like present-day LLM-driven instructions turning humans into robots to execute those actions.

There was lots of enthusiasm at the Summit and even more committment to the long haul. It seemed like the first of a series of summits in a long-growth-season of development. It's not hype but rather a combination of education, searching for dollars and engineering.

Expand full comment
Kevin Lawton's avatar

Great takes on the state of humanoids

Expand full comment

No posts