If they are physically capable of performing any minor surgery with slightly better precision, accuracy and speed than any human surgeon in 4 years, that’s probably an easy box to check. Yet, there will also be specialized non humanoid robots purpose built for surgery. For humanoid robots to be worth it, they will have to fit into the existing medical profession in a very seamless way and offer far more versatility than the specialized surgeon robots. If a surgeon has to stand next to the humanoid robot and babysit it in case it malfunctions during a surgery, then the surgeon will think they are better off doing it themselves. I think that slow adoption phase where the robots build trust and surgeons in the operating room adapt to working with them will take much longer.