Post #2295250

Anyway, https://deepmind.google/research/publications/231971/ might be an interesting read, although for me at least it'll involve a bunch of pre-work to catch up to the ideas and frameworks it builds on. The paper does have a very funny disclaimer at the top that basically says "I don't think AI can ever be conscious. My employer wants to make it clear that their strategic business direction is incompatible with this conclusion." Good on them for allowing publication I suppose, but it's a very funny thing to demand to have in writing at the top of the paper.