Love the Way You Lie: Unmasking the Deceptions of LLMs

Yulia Kumar, Zachary Gordon, Patricia Morreale, J. Jenny Li, Brendan Hannon

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Scopus citations

Abstract

Within the dynamic realm of Artificial Intelligence (AI), models like ChatGPT, Bard, and Bing are renowned for replicating human language. However, their emergence sparks debate over biases and trustworthiness. This research delves into the predominant inaccuracies in chatbots that tend to mislead novices and explores the possibility of establishing an AI Reliability (AIR) framework to fortify trust in these entities. Errors are categorized as factual inaccuracies, misinformation, fabricated data, and deviations from topics, among others. The in-progress AIR Framework offers a meticulous approach to assess chatbot accuracy, leveraging the experiences of nearly 100 CS/IT students with mainly ChatGPT. Recognizing the limitations and hallucinations of these models is essential as they become integral to our lives, underscoring the imperative for responsible and reliable AI.

Original languageEnglish
Title of host publicationProceedings - 2023 IEEE 23rd International Conference on Software Quality, Reliability, and Security Companion, QRS-C 2023
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages875-876
Number of pages2
ISBN (Electronic)9798350359398
DOIs
StatePublished - 2023
Event23rd IEEE International Conference on Software Quality, Reliability, and Security Companion, QRS-C 2023 - Chiang Mai, Thailand
Duration: 22 Oct 202326 Oct 2023

Publication series

NameProceedings - 2023 IEEE 23rd International Conference on Software Quality, Reliability, and Security Companion, QRS-C 2023

Conference

Conference23rd IEEE International Conference on Software Quality, Reliability, and Security Companion, QRS-C 2023
Country/TerritoryThailand
CityChiang Mai
Period22/10/2326/10/23

Keywords

  • AI bias
  • AI Misinformation
  • AI Reliability (AIR) Framework
  • Ethics of AI
  • Large Language Models (LLMs)

Fingerprint

Dive into the research topics of 'Love the Way You Lie: Unmasking the Deceptions of LLMs'. Together they form a unique fingerprint.

Cite this