This research examines the use of large language models (LLMs), specifically GPT-4, for autonomous fact-checking as an answer to the rise of online misinformation. The study evaluates the effectiveness of these models in phrasing queries, retrieving contextual data, and making decisions. The researchers found that LLMs exhibit improved proficiency when equipped with contextual information, and GPT-4 showed better performance than GPT-3. However, the accuracy of these models fluctuates based on the language of the query and the truthfulness of the claim, necessitating caution in their application. The authors call for further research to understand the precise conditions under which these machine learning models succeed or fail.
Publication date: 23 Oct 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2310.13549
