Conversation

Notices

1. Embed this notice
   Philip Weiss (Phil in SF) (kingrat@sfba.social)'s status on Saturday, 26-Sep-2026 02:31:26 JST Philip Weiss (Phil in SF)

   in reply to

   • Tito Swineflu
   • edgeworth

   @edgeworth @tito_swineflu yes. it absolutely does. cheaper too. there are consultants who spend all their time "advising" clients to have their LLMs write deterministic tools which it then calls to answer questions.

   • Embed this notice
     edgeworth (edgeworth@sfba.social)'s status on Saturday, 26-Sep-2026 02:31:27 JST edgeworth

     in reply to

     • Tito Swineflu

     @tito_swineflu I was actually wondering if asking an LLM to write code to count letters and then execute that code against a string would have better results than just asking directly

   • Embed this notice
     Tito Swineflu (tito_swineflu@sfba.social)'s status on Saturday, 26-Sep-2026 02:31:28 JST Tito Swineflu

     I have made a new superintelligence that outperforms all of the frontier models!

     len([x for x in 'strawberry' if x == 'r'])