JOURNAL ARTICLE

Is your AI hallucinating? New approach can tell when chatbots make things up.

  • Published In: Sciencemag.org, 2024. P. N.PAG 1 of 3

  • Database: Applied Science & Technology Source Ultimate 2 of 3

  • Authored By: Amer, Pakinam 3 of 3

Abstract

A new study published in Nature proposes a method to detect and prevent "hallucinations" in large language models (LLMs), such as chatbots and answer engines. These hallucinations refer to confident but incorrect responses generated by AI systems. The method measures the semantic entropy of the responses, which indicates the randomness and reliability of the information. The researchers found that their method had a high degree of accuracy when compared to human raters. However, some experts suggest that existing methods, such as Google Gemini's "self-consistency," may already address this issue. The proposed method is relatively easy to integrate but comes with a computational cost and may not detect errors if the LLM sticks to its false narrative. [Extracted from the article]

Additional Information

  • Source:Sciencemag.org. 2024/06, pN.PAG
  • Document Type:Article
  • Subject Area:Computer Science
  • Publication Date:2024
  • Accession Number:178006276
  • Copyright Statement:Copyright of Sciencemag.org is the property of American Association for the Advancement of Science and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Looking to go deeper into this topic? Look for more articles on EBSCOhost.