Gabriel Petersson, whose X bio notes prior research at OpenAI on Sora after earlier work at Midjourney, posted that AI will prove extremely important for national security of the Western world and allies, stressing the need to keep investing and maintain competing labs, then explicitly asked LessWrong users to link any well-developed scenarios they had already created.
maybe time to summon the @lesswrong people to link to scenarios, ya'll been thinking for so long, surely there must be a few well developed scenarios here
ai will be extremely important for national security, the western world and its allies
wonder what scenarios there are here, it's a little scary but also very important to keep investing in ai and have western competing labs
Petersson鈥檚 call for detailed future narratives did not include any attached examples or prior threads, and no replies supplying those specific links have been recorded in the available activity.
Andreas Kirsch replied by assuming familiarity with the scenario reports titled AI 2027 and Europe 2031, which he has described elsewhere as containing useful predictions, though full authorship, methodology, and publication dates for the documents remain unspecified in the thread.
Some users praised the Europe 2031 report as well-written and effective at explaining AI cut-off scenarios related to Western national security.
ai will be extremely important for national security, the western world and its allies
wonder what scenarios there are here, it's a little scary but also very important to keep investing in ai and have western competing labs
US revoking AI access to a developed country in a year or two might make it into a developing country
open source seem to fall increasingly behind, 6-12 months
most work will be done through ai, and a lot of companies will spend more money on tokens than employees
@gabriel1 @lesswrong Europe 2031 was released a few days ago and did a good job walking how a "cut-off" scenario could work. It's also *very* well-written. Strong rec: https://europe2031.ai/
@gabriel1 @lesswrong @KatjaGrace and @gwern probably would have the best opinions
@gabriel1 @lesswrong From 2013 has some relevant stuff (including a prescient sub title) https://intelligence.org/files/IEM.pdf
@gabriel1 I assume you've already read AI 2027 and now Europe 2031? 馃檭
ai will be extremely important for national security, the western world and its allies
wonder what scenarios there are here, it's a little scary but also very important to keep investing in ai and have western competing labs
@gabriel1 @lesswrong @DadaJudith and a group of researchers recently released there scenario on where this could go for europe, worth a read
https://europe2031.ai/
@gabriel1 It's crazy that token spend might overtake payroll this fast. Already rebuilding how I'd budget a team.
@gabriel1 @lesswrong More recent https://www.lesswrong.com/posts/BueeGgwJHt9D5bAsE And the OG from 2008: https://intelligence.org/files/AIPosNegFactor.pdf