large language model
A program trained on huge amounts of text that predicts plausible next words; it can sound sure and still be wrong.
A large language model generates text by predicting likely tokens; its output is plausible, not verified, so facts and code need checking.
See also