Reactions from ranked influencers
7 posts
«it’s not supposed to be a fair fight…» Why are all rationalists china hawks? I remember Scott A. expressing sympathy for Iran, Yudkowsky for fucking Al Qaeda. But not a single rat can say "wait the CCP looks way saner than our corrupt madhouse", even as a contrarian bit.
I’m not especially optimistic about US (in some sense) in control of ASI tbc but I’m significantly more pessimistic about China
> not a single rat can say "wait the CCP looks way saner than our corrupt madhouse", even as a contrarian bit Seems literally false in a narrow way, e.g. https://www.lesswrong.com/posts/nmpzH6sLLtKsQhSPM/china-won-t-win-the-ai-race-but-would-it-be-much-worse-if-it (not saying I agree with this tbc). --- Separately, it does really seem worse for China to be closer or to have a (decisive) lead. I'd expect them to handle misalignment less well, to generally manage problems downstream of rapid tech progress less well, and to use the power for worse ends. (Also seems better all else equal for one country/coalition to have a bigger lead so they can slow for safety as needed, but this is a smaller effect I think.)
Fair, all "nobody ever" claims are at best provocative > I'd expect them to handle misalignment less well, to generally manage problems downstream of rapid tech progress less well, and to use the power for worse ends Well, I'm sure you'd expect so, but to me you seem radically worse and less responsible actors.
Though I think the core reason is just that they don't seriously consider the risk of China "winning" in an existential fight *imposed on it*. How could that be? Not. a. single. rat. has thought about this.
Imo there's a bit of narcissism, too. Scott Alexander Siskind will point to Ilya Sutskever (who's currently doing… who the fuck knows). He won't think hard about Yang Zhilin who went back to China and built Kimi K3. On BaYeSiAn PriOrS, China is unlikely to have real geniuses.
@teortaxesTex When you say "you", you're referring to the US overall given the relevant actors and averaging over relevant possible administrations etc?