Curated. The question beneath feels really quite interesting. As the OP have said, even if it’s the case that a vastly superhuman intelligent AI could defeat even at extreme disadvantage, this doesn’t mean there isn’t some advantage that would let humans defeat a more nascently powerful AGI, and it’s pretty interesting to understand the how that works out. I’m excited to see more work on this, especially in domains resembling more and more real life* (e.g. Habryka suggests Starcraft).
*Something about chess is it feels quite “tight” in terms of not admitting exploits or hacks the way I could imagine other games have hidden exploitable bugs that can be mined – like reality.
Curated. The question beneath feels really quite interesting. As the OP have said, even if it’s the case that a vastly superhuman intelligent AI could defeat even at extreme disadvantage, this doesn’t mean there isn’t some advantage that would let humans defeat a more nascently powerful AGI, and it’s pretty interesting to understand the how that works out. I’m excited to see more work on this, especially in domains resembling more and more real life* (e.g. Habryka suggests Starcraft).
*Something about chess is it feels quite “tight” in terms of not admitting exploits or hacks the way I could imagine other games have hidden exploitable bugs that can be mined – like reality.