I think it’s plausible that NIST will play an important role in the US government’s response to AI in the future, but it definitely did not end up producing high-quality alignment research, or really high-quality public writing or thinking of any kind, and the whole situation really doesn’t feel like it’s playing to Paul’s strength.
I don’t have a super strong take on whether Paul taking on this role was a mistake ex-ante, but ex-post it seems like there hasn’t been that much traction (but also, stuff might still happen, the policy world always moves exceptionally slowly).
I think it’s plausible that NIST will play an important role in the US government’s response to AI in the future
This might be why he sticks around, I certainly can’t think of any other reason. He must also think this chance outweighs the opportunity cost of working with ARC or similar.
(unless it’s health problems or some other personal trouble)
I think it’s plausible that NIST will play an important role in the US government’s response to AI in the future, but it definitely did not end up producing high-quality alignment research, or really high-quality public writing or thinking of any kind, and the whole situation really doesn’t feel like it’s playing to Paul’s strength.
I don’t have a super strong take on whether Paul taking on this role was a mistake ex-ante, but ex-post it seems like there hasn’t been that much traction (but also, stuff might still happen, the policy world always moves exceptionally slowly).
This might be why he sticks around, I certainly can’t think of any other reason. He must also think this chance outweighs the opportunity cost of working with ARC or similar.
(unless it’s health problems or some other personal trouble)