ProgramCrafter

User-in­cli­na­tion-guess­ing al­gorithms: reg­is­ter­ing a goal

Pro­gramCrafter’s Shortform

LLM mis­al­ign­ment can prob­a­bly be found with­out man­ual prompt engineering

[Question] Does ob­ject per­ma­nence of simu­lacrum af­fect LLMs’ rea­son­ing?

The frozen neutrality

Pro­posal on AI eval­u­a­tion: false-proving

How AI could workaround goals if rated by people