Why do LLMs make stuff up? New research peers under the hood.
One of the most frustrating things about using a large language model is dealing with its tendency to confabulate information, hallucinating answers that are not supported by its training data.…