I'm completely open to the possibility of having good telepresence in reliable hardware be a valuable addition to our economies. The hardware really is getting good these days, at useful payloads.
But what I'm particularly pessimistic about is seeing academic-type research try to fold clothes and put screws into holes with humanoid-like systems, raise 10s of $Mils, and form a company out of it. Ditto for existing humanoids.
One of the major humanoid companies (you can google/guess who) recently claimed that their humanoids will be doing surgeries in hospitals within 5 years? This is egregious, there is zero chance that becomes the platform of choice for doing surgery, over a special purpose-built bed + multi-arm platform specialized for said tasks. This is perhaps even worse that Rosie from the Jetsons vacuuming our homes before the Roomba is invented.
Showing another example, I have to get into specifics - https://www.youtube.com/watch?v=WlUFoZstcWg But I want to know, how could this possibly be better than setting up arms for this task, assuming it is to be done thousands of times? Is the change to the factory layout the main issue that makes this worth it? Even if it is... a single arm on a mobile platform certainly does this more economically.
Anyways, I see all this stuff, there's tons of money and hype and optimism, and I feel crazy being pessimistic, as I'm usually the techno optimist. So, do others feel similarly? Are there things I'm missing that could fuel some optimism perhaps?
Looking forward to the discussion.
But I really struggle to come up with any other economically viable short-term use cases, even with great hardware...
Note that we are only starting to see the (much smaller compared to llms) DL data scaling in robotics - almost entire previous research has been achieved with very small robot fleets.
I think scaling data from industrial-sized robot fleets will lead to quick solution of various general robotics capabilities.
Because every time I think of something, either an existing industrial setup can or will do it better, or a special-purpose device will beat it.
So general intelligence + general form factor (humanoid) sounds great, if feasible. But what will it do exactly? And then let's do a reality check on said application.
Some companies are investing trillions in LLM AIs. It is probably a bubble. But we will still have LLM AIs if that bubble bursts.
Some companies are investing billions in humanoid robots. It may be a bubble. But we will still have humanoid robots if that bubble bursts.
They also have proven themselves in other white collar knowledge endeavors as well, as valuable tools that augment human economic output. Marketers can make more copy material, any office worker can improve the quality of their email communications, etc. Easy.
What are humanoids doing exactly? What can they do, that actually makes sense and provides positive economic impact over existing alternatives? Not clear to me.