I think we'll reach "most work done by unattended robots" around 2070. But we'd be better off if started working on post-labor-scarcity economics ASAP--might as well start learning to swim before the ship sinks. It might even be fun.
It doesn't really matter if they can or can't do the work. If enough execs are convinced AI can replace the workers, they will replace the workers, and management will have moved onto other things before they suffer consequences.
There will not be much work that cannot be done by Figure, Optimus, Atlas, Claude, Grok or GPT by 2035.