> here we are in 2026 and robots could plausibly clean our homes beyond vacuuming and mopping.

It is very bold to just assert this is true. Certainly it will be possible eventually, but there's still _lots_ of disagreement in the industry about what is realistic within 3-5 years. See this rodney brooks article for a good overview of the difficulties: https://rodneybrooks.com/why-todays-humanoids-wont-learn-dex...

The fact that devgutt was talking about this in 2015 gives some hint at its unique combination of [seems really easy] and [is really hard].

That article is a load of baloney, and I wish people stopped posting it around as if it's some kind of gospel.

Modern robots are nowhere near being bottlenecked by hardware. They are all bottlenecked by AI. Today's hardware with perfect AI would absolutely demolish tasks like "clean a house". Today's AI with perfect hardware would still fumble.

We know that because we can't even train an AI policy that would reliably solve tasks in a sim with perfect sensors and perfect execution.