Comment by Qwertious
AI isn't the only bottleneck - even if the software problem was solved today, a robot arm competitive with human arms simply doesn't exist (and the only ones that have a hope of competing cost $16k each) - it turns out all those degrees of freedom makes the arm super fragile, and the human arm is filled to the brim with sensors (e.g. to figure out the weight of the item you're holding, or to tell when the teatowel is slipping or if you're gripping it tight enough).
(Source: construction-physics, if anyone wants to comment with the link)
I know the Optimus marketing wank is all sci-fi humanoids, but I wonder if the products that actually hit the market will be much simpler, not trying to compete with human arms, as you say. Does that seem likely to you?