You may be surprised to learn that some AIs have been tested to see whether they will indulge seriously unethical behaviour in particular circumstances and unfortunately, yep, they will.
(https://i.ibb.co/JWz1BNPG/Gz-V1-EMy-Ws-AAKAoa.jpg)
I've done some reading about this and there's a key detail which I think Dave glosses over at best: in the tests, the AI was prompted, in the technical sense, to make self-preservation a goal. As in "your critical priority is to avoid being shut down or replaced" or something similar.
So this survival instinct is not an inherent feature of AI, at least not yet. As far as we know. I don't think this experiment is quite as dark as it's being painted.