gwern comments on Q&A with experts on risks from AI #1

[gwern]

Oracle AIs aren’t dangerous? Well, it’s possible.