Imagine how dumb the California legislature would look today if Newsom had signed SB 1047 last fall instead of vetoing it.
Post
Conversation
You're saying that we should evaluate decisions in hindsight rather than based on what was known at the time?
![David Watson 🥑](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="96" height="96"><rect fill-opacity="0"/></svg>)
Post your reply
Would SB1047 have affected anything that has happened in the interim? My understanding is no. So the concern is still potential hampered US efforts in the future, even though powerful models with less compute are possible?
I'm not sure any model has been trained with 10^26 FLOPs yet so probably the practical impact of SB 1047 would be nil so far. But it seems bad to hold back US firms if they are competing with a Chinese firm releasing open-weight models.
China has stricter requirements and consequences re: making their models not say certain things. There is real threat of you disappearing if you fuck it up, compared to your company paying liabilities re: SB-1047 only if you cause big bucks of harm.