Resolves according to my opinion on whether Gwern is more correct here.
https://twitter.com/gwern/status/1741875125471846837
This has a fairly high bar- think of it like, will Chinese output be at least somewhat comparable to the US.
Definitely would've resolved no in 2023, I'm not super impressed by an open source model getting close to 3.5 level of performance.
I will not bet in this market.
Related questions
Let's say China releases a clone of today's GPT-4 with the same capabilities in Sep 2024, and something similar to today's dalle 3 in sep 2024. does this resolve yes or no?
edited to remove stupid thing
@Pykess Cool, but the bar is definitely higher than that. I'd like to see at least a couple advancements of similar level to the big breakthroughs in the US.
@SemioticRivalry I see, you're looking for significant advancement for the whole field of AI that comes out of China. I'd suggest changing the title to reflect this. Currently "significant Chinese AI progress" just means that the AI scene within China has progressed significantly, which DeepSeek would've satisfied. Something like "Will we find out in 2024 about breakthrough AI research/progress from China"