What remains when help ends?
Deciding before AI advice: Greater perceived ownership, no later advantage in health-claim verification
Trattner, C., Wessel, T. J., Lysova, Y., and Starke, A. (2026). Preprint. Revised version, 15 November 2026.
A two-session experiment on health-claim verification comparing answer-first AI, evidence-first AI and source practice. Participants who judged before seeing the AI answer reported more ownership of their judgement, but were not more accurate on new claims judged without AI about a week later.





