From galleries to generators : Applying museum empathetic strategies to reduce confirmation bias in chatbots

Tools

Kist, Cassandra (2026) From galleries to generators : Applying museum empathetic strategies to reduce confirmation bias in chatbots. In: EmpathiCH’26 Workshop Co-located with CHI’26 Conference on Human Factors in Computing Systems, 2026-04-14 - 2026-04-14, Barcelona, Spain.

[thumbnail of Kist_Empathy_and_confirmation_bias_ACM_2026]
Preview
 Text. Filename: Kist_Empathy_and_confirmation_bias_ACM_2026.pdf
Accepted Author Manuscript
License: Creative Commons Attribution 4.0 logo
Download (391kB)| Preview

Abstract

Museums have long positioned themselves as socially responsible institutions, designing heritage interpretation to foster inclusion, and cultivate cross cultural understanding. The ‘affective-turn’ in museum practice has expanded heritage interpretation beyond text toward multimodal, embodied, sensory, and emotive experiences that support visitors’ reflective meaning making. In this provocation, I argue that general-use Generative AI chatbots (such as ChatGPT and Co-pilot) should take inspiration from museum interpretive techniques to address growing risks of confirmation bias arising from biased queries, belief consistent personalisation, and overly agreeable chatbot response styles. Drawing on museum strategies for cultivating empathy – encompassing perspectivity, reflective action, and sensory, and affective connection – I outline how these interpretive techniques could inform chatbot interaction design. I conclude with design directions for perspective expanding chatbots that mitigate bias not only through cognition, but also by privileging users’ meaning making as an emotive process.

ORCID iDs

Kist, Cassandra ORCID logoORCID: https://orcid.org/0000-0001-9960-2236;
CORE (COnnecting REpositories)