I don’t have a background in AI, but I recently read a book which seems like a pretty good metaphor for AI and thought I’d share!
The book, A Beautifully Foolish Endeavor by Hank Green, focuses on [spoilers ahead] two alien intelligences battling, primarily via digital manipulation of humans on the internet. The alien intelligences aren’t AI, but they seem like good analogues for aligned and non-aligned AI: one of the intelligences has programming which prohibits it from doing things that would contravene human values or destroy fundamental aspects of humanity, while the other has no such programming and is trying to save humans from destroying ourselves by locking us into a dystopian virtual reality simulation where we are controlled by the alien intelligence.
A Beautifully Foolish Endeavor is actually a sequel; the first book, An Absolutely Remarkable Thing, is less about AI, but still related (as well as being necessary to understand the sequel). Both books are well-reviewed, and they’re also both NYT bestsellers, so they’re quite popular.
I’m not sure how, if at all, the EA community can use the success of A Beautifully Foolish Endeavor to increase awareness of AI risk, but I figured I’d let folks know about it.