When Should I Trust This? Designing and Evaluating Factuality Indicators for LLM-based Chatbots
4.60T1 sourcearXiv cs.HC
Source record
Published by arXiv cs.HC (T1 source). The original is at https://arxiv.org/abs/2610.32883.
Pipeline notes
The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.
SummaryResearchers studied how people judge AI-generated answers and whether confidence and source indicators affect trust. Interviews, a survey, and an experiment covering gardening, mathematical, and medical queries found indicators shape initial perceptions, effects shrink after users learn correctness, and medical answers receive greater skepticism.
Why it mattersFor chatbot and interface designers, the study links trust behavior to user verification, source quality, and task risk, offering evidence for where confidence and citations should appear.
Cited by
No citations on record.