Chris Stott's question on Mesa-Optimizers

From Stampy's Wiki

Is there a safety problem with making one-trick AIs? "Your objectives, in order of priority are: 1: Completely reset yourself 24 hours hence. 2: Seek a solution for optimizing Problem X, reporting your best result 23 hours from now."


Tags: None (add tags)
Question Info
Asked by: Chris Stott
OriginWhere was this question originally asked
YouTube (comment link)
On video: The OTHER AI Alignment Problem: Mesa-Optimizers and Inner Alignment
Date: 2021-02-20T20:55
Asked on Discord? Yes


Discussion