de en es fr nl pl pt sv zh

verification

Pointers, Not Answers

Agam Brahma

tl;dr: treat llm output as a pointer to the answer, not the answerEveryone who leans on these tools has been burnt: an invented flag, a test that passes for the wrong reason. Abstaining isn't a serious option for me either ... the leverage is too large to leave on the table. So the question is how...