I think a more accurate way to model them is “GiveWell recommends organizations that are [within the Overton Window]/[have very sound data to back impact estimates] that save as many current lives as possible.” If GiveWell wanted to recommend organizations that save as many human lives as possible, their portfolio would probably be entirely made up of AI safety orgs.
Sounds about right—this paper used an older AI Safety model to find $16 to $12,000 per life saved in the present generation. Though I think some other GCR interventions could also compete on that metric, such as neglected work on engineered pandemics, and resilience to food catastrophes.
Sounds about right—this paper used an older AI Safety model to find $16 to $12,000 per life saved in the present generation. Though I think some other GCR interventions could also compete on that metric, such as neglected work on engineered pandemics, and resilience to food catastrophes.