I’m missing from the article two things: - testing prompt (were llms instructed to progress in game, as opposed to just explore — the author said smarter llms

I’m missing from the article two things: - testing prompt (were llms instructed to progress in game, as opposed to just explore — the author said smarter llms were more likely to explore) - benchmark with humans