Stephen Martin comments on It’s hard to make scheming evals look realistic for LLMs

[Stephen Martin]

Yeah but I think that’s probably where the recursion ends.