Lists
Lists
Trivedi, Swati, Dey, Maitreyee and Patel, Preeti (2026) An analysis of key challenges in user modelling, uncertainty communication, and fairness in conversational AI systems. In: 14th International Conference on Frontiers of Intelligent Computing: Theory and Applications (FICTA-2026), 8-9 June 2026, London (UK) / Online. (In Press)
This paper examines critical challenges limiting the effectiveness and trustworthiness of conversational AI systems, particularly in domains such as healthcare, education, finance and energy. It focuses on three interconnected dimensions: user behaviour modelling, uncertainty communication, and fairness and transparency. Current systems predominantly model user preferences and actions but fail to capture how users communicate, restricting true personalisation. Additionally, conversational AI often exhibits poorly calibrated uncertainty, leading to overconfident and potentially misleading responses, while insufficient fairness and transparency mechanisms undermine user trust and accountability. To address these issues, the paper introduces a unified framework that organises these challenges into progressive capability levels and highlights their interdependencies. By analysing existing approaches across these dimensions, the study identifies critical gaps, particularly the absence of adaptive, user-aware, and accountable systems. The paper concludes by outlining key research directions toward developing more reliable, transparent, and user-centred conversational AI systems suitable for real-world deployment.
![[thumbnail of An Analysis of Key Challenges in User.pdf]](https://repository.londonmet.ac.uk/style/images/fileicons/text.png "An Analysis of Key Challenges in User.pdf")
Restricted to Repository staff only until 6 May 2027.
Download (1MB) | Request a copy
 |
View Item |