Authors: S Sabouri, JB Graber, J May, JK Kummerfeld, S Chattopadhyay Venue: VL/HCC 2026 (accepted July 2026)

Summary

A survey of 98 participants verifying factual claims with LLM advice presented in different rhetorical styles. Shows how the styling of advice — independent of its content — shifts user performance, confidence, and engagement. The study is about the effect of rhetorical patterns for AI advisors, and was accepted to VL/HCC 2026 in July 2026.

Paper

Thread: AI Aid for Decision Making — the research thread. Prompting Reflection: Encouraging Users to Critically Evaluate AI Suggestions — companion study on interrupting rather than styling.