Lying Large Language Models: Understanding Untruth in Artificial Intelligence Through Synthetic Ethnography

Abstract

Large Language Models (LLMs) have many practical uses in areas like journalism, search, coding and more. However, a growing concern is that they are also prone to presenting incorrect information. This is sometimes called “hallucinations”, but a more correct term would be “bullshit”, i.e. text produced without concern for its truth. In this study, we are not interested in what specific untruths LLMs presents, but how they do it. We used synthetic ethnography, a method for the qualitative study of generative models, to study two LLMs with different size and capability. We collected 3 cases where LLMs presented incorrect information and observed the strategies they used to justify this. From these observations we can start to form an understanding of what happens when an LLM reaches the edge of its knowledge-base and takes corrective action. Our conclusion is that the interfaces should be better designed to reveal this tendency of LLMs to “fill in” information they are missing, but that this ability may also be one of their strengths.

Bio
Lars Erik Holmquist is Professor of Design and Innovation at Nottingham Trent University. He is an internationally leading researcher in human-computer interaction (HCI), interaction design and ubiquitous computing (ubicomp). He has published over 100 articles in fields such as HCI, design methods and mobile applications. His work has been presented at major scientific conferences including ACM CHI, SIGGRAPH, UIST, UbiComp, Mobile HCI, InfoVis and ECSCW. He is the author of 6 granted US patents and 1 granted UK patent. When working in industry, he developed features that were included in Yahoo, Inc’s flagship Yahoo! Search app, reaching millions of users. His book, Grounded Innovation: Strategies for Creating Digital Products, was published by Morgan Kaufman in 2012, and provides a practical guide to the design-driven innovation process, with many examples drawn from his own research and elsewhere.

Before joining NTU he was most recently Professor of Innovation at Northumbria University, School of Design, where he was the recipient of a Royal Society Wolfson Research Merit Award and led the Connected Experiences Lab. Previous appointments include Full Professor in Media Technology at Södertörn University; Lab Leader at the Swedish Institute of Computer Science; Group Leader at the Viktoria Institute; Studio Director at the Interactive Institute; and Guest Researcher at Tokyo University in the Human Augmentation Group.