Safety
49d agoCritique of AI Capability Measurement Strategies
The organization METR focuses on developing scientific methods to assess catastrophic risks from autonomous AI. This critique argues that measuring capabilities alone is insufficient to mitigate existential threats. Safety researchers must reweight their priorities toward preventative strategies rather than just observation.
The Signal
This shift targets the gap between detection and actual risk reduction.