Archive
About
Search
Log In
Home
All
Wiki
Shortform
Recent
Comments
oh54321 comments on
All AGI Safety questions welcome (especially basic ones) [~monthly thread]
oh54321
2 Nov 2022 22:17 UTC
1
point
0 ∶ 0
“RL agents with
coherent preference functions
will tend to be deceptively aligned by default.”—Why?
Back to top
“RL agents with coherent preference functions will tend to be deceptively aligned by default.”—Why?