Which AI outputs should humans check for shenanigans, to avoid AI takeover? An overly simple abstract model
Published on March 27, 2023 11:36 PM GMTEpistemic status: incomplete early stage thinking. Thinking this through made me feel less confused about how labs might one day prioritise which AI outputs humans should check for sabotage. Short summary <ul> <li>As AI…
Epistemic status: incomplete early stage thinking. Thinking this through made me feel less confused about how labs might one day prioritise which AI outputs humans should check for sabotage. <ul><li… [+16176 chars]