Paper · arXiv 2305.13534
How Language Model Hallucinations Can Snowball
Muru Zhang, Ofir Press, William Merrill, Alisa Liu, Noah A. Smith
Language models frequently generate incorrect answers and offer false explanations, exacerbating errors due to a phenomenon called hallucination snowballing.
Abstract
A major risk of using language models in practical applications is their tendency to hallucinate incorrect statements. Hallucinations are often attributed to knowledge gaps in LMs, but we hypothesize that in some cases, when justifying previously generated hallucinations, LMs output false claims that they can separately recognize as incorrect. We construct three question-answering datasets where ChatGPT and GPT-4 often state an incorrect answer and offer an explanation with at least one incorrect claim. Crucially, we find that ChatGPT and GPT-4 can identify 67% and 87% of their own mistakes, respectively. We refer to this phenomenon as hallucination snowballing: an LM over-commits to early mistakes, leading to more mistakes that it otherwise would not make.