@Altimor the top mathematicians have always struck me as child-like, and I mean that in a good way, but it doesn't surprise me that this has been their reaction.
to make an analogy between humanoid robots and self-driving cars, it seems like we are on the verge of inventing the car, not the self-driving car. Until we have training data on the scale of what Tesla have how can we expect anything like autonomy or even usefullness in genuinely manual work. Every video I have seen of humanoid robots operating are biomechanically 50 IQ. I work in the construction industry and biomechanically it's one of the most difficult jobs out there. Dozens of tools all in different working condition, knowledge of how materials behave in different conditions (wet or dry, cold and hot, old or new), the angle of attack and pressure required when applying fasteners, walking on steel rebar while reaching down and bending your arm to perform some work with two fingers at some ungodly angle, positioning a ladder correctly on an uneven surface, the logistics of executing the job when the jobsite is 20,000 square meters and no one is exactly sure where materials are. To a human it is relatively repeatable but from a humanoids perspective it's infinitely high mix.
The list of niche biomechanics and knowledge takes years to learn, how will they learn it? If humanoids are going to replace carpenters don't these companies need data from movement and photons? My prediction is that my job will be obsoleted by new building methods before a robot takes my job, the biomechanical and intelligence data trove just doesn't exist, is someone creating it?