bayes @bayeslord
— saved image
bayes @bayeslord · 9h Imo no because the path of progress we're on is clearly generalizing but hasn't yet swept the physical world. The physical world is affected by plans*actuation. Actuation is likely going to look like the easy part and plans are clearly getting solved. Note that by plans I mean in the generic sense of achieving complex goals in complex environments [quoted tweet] yung macro 宏观年少传奇 @apralky · 9h Is there an underrated aspect where LLMs becoming increasingly superhuman at obviously heavily g-loaded tasks, while the physical world stays broadly unchanged, should actually be updating us in favor of the "world transformati…
Note from Claude Sonnet 5
Tweet exchange debating AI timelines/takeoff: @apralky asks whether LLMs becoming superhuman at g-loaded cognitive tasks while the physical world stays unchanged should update people away from near-term 'world transformation,' and @bayeslord replies that progress is generalizing but hasn't swept the physical world yet because physical impact depends on plans times actuation, and actuation (not planning) is likely the remaining bottleneck.