Let's say that a misaligned AGI or Artificial Superhuman Intelligence wanted to utilize humans as an energy source (something that I've read from one or more AI safety philosophers although I can't recall who). How might that occur within the laws of physics as we currently understand them?
This is not the central threat, but if you did want a mechanism, I recommend looking into the krebs cycle.
Interesting suggestion. Thank you!