[geysersam]

> In this case the statistically most likely explanation the LLM replied with just happened to correspond with the correct one.

I claim that case is not so uncommon as people in this thread seem to think