# @davidad — 2022-11-02

♥16 ↻3 · https://x.com/davidad/status/1587825278126112770

Case: InstructGPT optimizing for answers that look impressively helpful (“use the inverse CDF method!…sqrt(-2*log(1-x))”) rather than answers that are aligned with the user’s goal (“np.random.normal()”).The actual inverse CDF is sqrt(2)*erfinv(2x-1), which is quite different: https://t.co/Hai8kqYzOg https://t.co/Qa51QgUWje

tags: author:davidad, kind:tweet, model:text-davinci-002, on:text-davinci-002, year:2022
cited on: _dossiers/text-davinci-002.md, text-davinci-002