alexbleakley

Thanks for writing this! I found this valuable to read.

I think there's an important space in between "actionable reproducible recipe" and "not helpful". When I consume self-help-ish content (including autobiographical accounts like this one), it's only very occasionally that I think "I should try this at the next opportunity". A more common positive outcome is something more like "this feels like a useful frame to add to my collection of frames". For me, all three of the methods described here landed in that second category, and if only one out of three had met that bar, I would still have found the post worth reading.

I agree that if overhang is a relevant consideration for pauses, then it's also a relevant consideration for RSPs. My previous question was: Do you see the same people invoking overhang as an argument against pauses and also talking about RSPs as though they are not also impacted?

Maybe you're not saying that there are people taking that position, but rather that those who invoke overhang as an argument against pauses don't seem to be equally vocal against RSPs (if not necessarily in favor of them either). I can think of a couple of separate reasons this co... (read more)

Are they the same people advocating for RSPs and also using compute/algorithm overhang as a primary argument against a pause? My understanding of the main argument in favor of RSPs over an immediate pause is:

1. Sure, we could continue to make some progress on safety if we paused other AI progress.
2. But:
   1. we could make even more progress on safety if we could work with more advanced models; and 
   2. right now we have the necessary safety measures to create the next generation of models with low risk.
3. If AI progress continues without corresponding progress on safety