@GregDance @EUCommission AI is a wayyy broader concept than just large language models.

But yes, LLMs are just advanced autocomplete. After being trained on many example texts, they predict which word is the most plausible continuation of a given text. Most of their limits stem from this.