An LLM is an artificial intelligence model trained on very large volumes of text data to predict and generate natural language sequences.

It works by calculating probabilities based on linguistic structures and has no semantic understanding of reality, no intention and no intrinsic capacity to verify the information it produces.