LLM produces bad pizza topping suggestions: funny
LLM produces bad "vibe coding" results: possibly ouch, but probably reversible
LLM produces unreliable medical evidence synthesis: there is literally no way to know or correct this, this was supposed to be the method for correcting medical bias itself and it's really REALLY fucking bad if you don't know if a systematic review was run slightly differently 1000 times and then cherry-picked
https://blog.bgcarlisle.com/2025/05/16/a-plausible-scalable-and-slightly-wrong-black-box-why-large-language-models-are-a-fascist-technology-that-cannot-be-redeemed/
