Dr Vivek Nallur, Associate Professor, School of Computer Science, University College Dublin
Tuesday 1st December 2026, 13.15pm-14.15pm (GMT)
Chancellors Building, CB 4.10, University of Bath
You can attend this event in person or online.
This event is free
Dr Marina de Vos, [ART-AI Director of Training], invites you to join this AI seminar in person. If you would like to join online please e-mail us.
Abstract
With the advent of large language models, task-oriented conversational agents (aka chatbots) are increasingly used in various domains. They support decision-making by providing users with information, recommendations, and guidance through natural language interactions. We investigate if cognitive biases manifest in task-oriented conversational agent interactions and how cognitive load arising from dialogue complexity influences our susceptibility to cognitive biases.
We also investigate the use of LLMs as synthetic participants by evaluating whether LLMs can replicate human decision-making biases observed in human chatbot interactions.
Our speaker
Dr Vivek Nallur is Associate Professor, School of Computer Science, University College Dublin.
He primarily works with Multi-Agent Modelling and Social Simulation. He is very interested in computational modelling of autonomous decision-making in the presence of uncertainty, both at the individual level (such as cognitive biases, emotions, etc.) as well as at the social level (emergent, loosely coordinated). This typically requires the use of AI techniques both, symbolic as well as generative.
Another area of interest is how to implement and validate ethical behaviour in autonomous machines. To this end, he looks at modelling of social contexts, normative socio-technical systems, modelling of conflicting values, prioritization of goals and simulating how micro-behaviour influences macro norms.
