Generative AI use before medical visits: disclosure-item responses, trust, and care-seeking behaviors in a cross-sectional social-media survey in Poland

Key Findings

• — 84.3% of respondents (n=880; 95% CI 82.0 to 86.4) reported health-related generative AI use
• — Among the 519 respondents reporting AI use before a medical visit, 82.5% (n=428; 95% CI 79.0 to 85.5) selected 'No' on the item asking whether they had informed a physician about that use; 11.9% selected 'Yes'
• — Selection of 'No' was similar across strata of self-reported AI-related visit avoidance (79.2% to 86.8%); the most common stated reason was deference to physician authority (74.9%)
• — Male gender and greater trust in AI were associated with the behavioural outcomes in definite-response models; trust in AI was strongly negatively associated with age
• — Because consultation attendance and an index episode were not confirmed, the authors state the proportion cannot be read as a disclosure rate within completed clinical encounters

Methodology Notes

Anonymous Google Forms survey in Polish with programmed routing, N=1,044, convenience sample recruited through a multi-account social-media strategy; 24-item de novo questionnaire; chi-square, Mann-Whitney U, Kruskal-Wallis, Spearman and logistic regression restricted to definite Yes/No responses, with Benjamini-Hochberg adjustment as a sensitivity analysis. Not generalisable to the Polish population; the survey did not confirm that a consultation occurred after AI use. Received for review before online publication on 2026-08-13 and published 2026-08-27 (dates from the article metadata). Two authors are employed by LUX MED, a private healthcare provider. Open access; full text read on the Frontiers article page.

Sources