I think the strongest part of the argument is the emphasis on robustness rather than requiring grantmakers to predict AGI correctly. Even if the timelines are substantially longer, identifying projects that remain valuable across both worlds seems like a sensible portfolio strategy. I also like the suggestion that non-AI grantmakers should build enough AI literacy to recognize important inflection points without necessarily abandoning their existing causes. The difficult part is probably avoiding both complacency and overreaction when the evidence is still highly uncertain.
I think the strongest part of the argument is the emphasis on robustness rather than requiring grantmakers to predict AGI correctly. Even if the timelines are substantially longer, identifying projects that remain valuable across both worlds seems like a sensible portfolio strategy. I also like the suggestion that non-AI grantmakers should build enough AI literacy to recognize important inflection points without necessarily abandoning their existing causes. The difficult part is probably avoiding both complacency and overreaction when the evidence is still highly uncertain.