Ask HN: Does anyone else think that humanoid robots is a bubble?
9 comments
+100. The money burned on this will be in the trillions by the end of the decade. I still feel like humanity will benefit from general humanoid robots, but the timeline is underestimated. You are right that specialty robotics and general arms are the right choice for commerce and repeatable actions today and that these will be superior for speed/quality for the foreseeable future until software and power storage catches up to what is needed for untethered humanoid form robots.
Humanoids are much cheaper than a car or an ev to manufacture at scale - the economics for humanoids is potentially very scalable and efficient. The solid state batteries are remarkably dense too, and battery replacement via dockstations has already been implemented in some models.
The hardware is great and can definitely scale. That's why as a caveat I think teleoperation is a good general purpose application cluster for these.
But I really struggle to come up with any other economically viable short-term use cases, even with great hardware...
But I really struggle to come up with any other economically viable short-term use cases, even with great hardware...
It's a hard problem, but deep learning is very scalable and general and the pressure for general robotics to be solved is very strong in China and US, given the demographic shifts. I think the proliferation of humanoids is a near certainty over the next 8 years, ofc it won't be uniform and licensed labor won't be replaced.
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.
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.
Ok but can we get into the nuts and bolts of what we actually want these robots to do?
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.
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.
I think there is a lot of noise but the idea of a humanoid working "for free" is so appealing that I think it will go mainstream as robot vacuum cleaners did - but surely most of the current companies will fail. And yes, I think surgeries will probably be done by specialized robots. They just want you to believe how precise their humanoids can work.
Obviously we don't need 100 companies developing identical robots. 95% of them will fail.
Some companies invested in the young internet. That created "The Internet Bubble". We still had the internet after that bubble burst.
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.
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.
LLMs have already proven themselves to be economically valuable. At a bare minimum, they can help people develop most low-mid level software considerably faster, at a good enough quality.
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.
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.
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.