Banal Deception Human-AI Ecosystems: A Study of People's Perceptions of LLM-generated Deceptive Behaviour
Xiao Zhan, Yifan Xu, Noura Abdi, Joe Collenette, Ruba Abu-Salma, Stefan Sarkadi
TL;DR
This paper investigates banal deception in LLMs by examining how people perceive ChatGPT generated deception and how those perceptions influence trust and usage. It employs a mixed-methods design with a large online survey (N=220) and in-depth semi-structured interviews (n=12) to identify deception types, contexts, and the factors shaping responsibility and behavioral responses. Key findings show oversimplified outputs as the most common deception, that perceived worthiness remains relatively high despite deception, and that responsibility is variably assigned to developers, hosting platforms, or users, with trust modulated by accuracy, transparency, and domain. The study argues for user-centric safeguards, verification mechanisms, and regulatory considerations to foster trustworthy human–AI ecosystems and mitigate deceptive AI harms. Overall, the work lays groundwork for Deceptive AI Ecosystems by outlining how perception, trust, and verification drive user interactions with deception-prone AI technologies and by outlining avenues for improving safety and accountability.
Abstract
Large language models (LLMs) can provide users with false, inaccurate, or misleading information, and we consider the output of this type of information as what Natale (2021) calls `banal' deceptive behaviour. Here, we investigate peoples' perceptions of ChatGPT-generated deceptive behaviour and how this affects peoples' own behaviour and trust. To do this, we use a mixed-methods approach comprising of (i) an online survey with 220 participants and (ii) semi-structured interviews with 12 participants. Our results show that (i) the most common types of deceptive information encountered were over-simplifications and outdated information; (ii) humans' perceptions of trust and `worthiness' of talking to ChatGPT are impacted by `banal' deceptive behaviour; (iii) the perceived responsibility for deception is influenced by education level and the frequency of deceptive information; and (iv) users become more cautious after encountering deceptive information, but they come to trust the technology more when they identify advantages of using it. Our findings contribute to the understanding of human-AI interaction dynamics in the context of \textit{Deceptive AI Ecosystems}, and highlight the importance of user-centric approaches to mitigating the potential harms of deceptive AI technologies.