But robots do present great opportunities. They can probably be thoroughly sterilized in a large autoclave. Maybe some sort of ai can study videos of surgery as training.

Depending on the robot (like SRT-H), they do train them on recordings of surgeries, as well as give them verbal instructions during surgical procedures. Its also equally possible to equip them with feedback sensors and train them on cadavers or analogs, and once you have the data you can just copy / paste it across standardized hardware models. Of course, that doesn't necessarily account for ALL variables though.

Just to expand on humanoid robots potential utility:

1) Cost: Compared to units like the da Vinci, a humanoid robot could cost vastly less than conventional specialized robots.

2) Reduced Complexity: A humanoid robot (like Atlas), has far less mechanical complexity, allowing for ease of deployment using standardized off the shelf units / parts. Conventional surgical equipment could also potentially be used by it (gloves, gowns, scalpels, etc) to help with sterile conditions. The fact that the units are standardized and could be deployed in other fields makes bulk manufacturing more viable, and thus reduce costs. (see point #1)

3) Ease of Repair: A humanoid robot (like the Atlas), with less mechanical complexity and a standardized platform, can be easily repaired with off the shelf parts at low cost. This of course doesn't exclude the possibility of replacing its legs with a more sturdy platform, or altering its hands with specialized equipment, of course. (see point #1)

4) Portability: A humanoid robot has a reduced physical footprint, and can be dropped (walk?) quickly into an operating theater. In the event it becomes inoperable, it can quickly be swapped out or supported with a redundant unit on standby. With other, more dedicated surgical equipment (IE: da Vinci), the unit being inoperable would necessitate medical staff to perform the surgeries, and may require lengthly (and expensive) repairs. Or, as has been mentioned, in the event of a power outage requires a dedicated generator to stay online, whereas humanoid robots can have built in (and swappable, like Atlas) batteries for continuous function, or could even power critical medical equipment in an emergency (like ventilators) off their internal batteries.

For the most part, I don't think robots will replace human surgeons anytime soon. Augment them? Perhaps. Handle simple / repetitive surgeries / wound care? Sure, provided they can consistently provide safe results / value, but there will always be a place for humans in the mix, especially for complications / emergencies, or novel / complex tasks.

Even if they aren't in the OR, there have been other robots used to augment nursing staff by handling mundane tasks like ferrying samples to labs, delivering fresh linens to replace bed sheets and delivering them to the laundromat, providing meals, restocking shelves, etc. freeing staff to do more relevant tasks, like patient care.

But I also think, at least to a degree, that their deployment in surgical settings may be a foregone conclusion. Why? Because doctors are limited, wait times for some surgeries are long, and have strict selection processes. What if you have a serious medical condition, like requiring heart valve replacement, but are low on the list because of a history of smoking or some other issue, and may suffer / die long before you get the necessary surgery? Do you think such a person, with no other viable option, won't avail themselves of a willing robotic surgeon?

In other words, robots will if nothing else have an ample supply of the desperate and dispossessed, for better or worse...