Posts
Comments
Comment by
Sergei U (sergei-u) on
Catastrophic sabotage as a major threat model for human-level AI systems ·
2024-11-17T17:02:12.317Z ·
LW ·
GW
There is a greater chance that the catastrophic sabotage would get spotted and fail, upping our awareness level, compared to a chance that model will get sophisticated enough to execute the plan successfully on the first try