Designing Medical Chatbots where Accuracy and Acceptability Collide: An Exploratory, Vignette-based Study in Urban India

Why Read This from an HCI Perspective

This article clearly demonstrates from an HCI perspective that user acceptability is not guaranteed by accuracy alone in medical chatbots. In particular, it empirically examines how users’ experiential expectations, medical practices, and perceptions of authority in an urban Indian context shape interpretation—making it highly meaningful for UX practitioners and researchers. It also serves as an important reference point when thinking about designing fair medical AI in the Global South.

CIT's Commentary

From a CIT perspective, the core of this study is that when ‘the correct answer’ and ‘the answer that is accepted’ conflict, the interface functions not merely as a transmitter but as a mediator of meaning. A context-aware nudge can be read less as a device for persuading users and more as a design strategy that makes unfamiliar, normative recommendations interpretable by linking them to existing care experiences. However, while this approach may increase acceptability, it also carries the risk of reinforcing incorrect local practices. In HCI, this makes it crucial to design the boundary between the friendliness of information provision and clinical legitimacy. The fact that responses differ by education level suggests that refining explanation strategies and checking comprehensibility should be carried out in parallel.

Questions to Consider While Reading

• Q.How can context-aware nudges reduce users’ misunderstandings in real clinical situations while maintaining clinical accuracy—and what methods could verify this over the long term?
• Q.How can we set design criteria to balance framing that respects users’ existing medical practices with framing that does not reinforce harmful practices?
• Q.To reduce acceptability differences by education level, how should we adapt the explanation style or interaction structure of medical chatbots?