If we a seriously believe AI is an extinction level risk, and getting there more slowly might improve our chances, then it seems like we should try anything wit

If we a seriously believe AI is an extinction level risk, and getting there more slowly might improve our chances, then it seems like we should try anything with a good chance of working. I think this has a better chance of working than regulating AI research itself, since chip manufacturing is centralized and highly visible, whereas AI research could be advanced in secret by a few people in a room. Much easier to successfully regulate the former than the latter.