One of the component technologies of AI, and a cornerstone of the current AI hype train. The hype train version of LLMs is basically: process everything ever at great cost, then host a service that responds to user prompts with something that looks like the most probable response to those prompts.
Not to be confused with “the correct answer.”