46/100

Emerging evidence

Cite
More
EmailLinkedInX

Urology applicants say AI gave wrong info, yet only 44% always checked it

Study title: Artificial Intelligence in the Urology Residency Match: Use, Perceptions, and Implications

Evidence type
Survey
Sample
n = 70
Sources
1 sourcePeer-reviewed
Published
2 min read

Study title: Artificial Intelligence in the Urology Residency Match: Use, Perceptions, and Implications

A survey of applicants to one academic urology residency program during the 2025 to 2026 match cycle found that 51% used AI while applying. ChatGPT was the most used tool, at 83% of users. Applicants also overestimated how much their peers used AI.

Read the study abstract

Objective: To evaluate artificial intelligence (AI) use and perceptions among urology applicants during the 2025 to 2026 residency match cycle. Methods: An anonymous survey was distributed to all applicants at a single academic urology residency program after rank list submission and before match day. A 29-item subset of the survey was analyzed, which assessed applicant demographics, application components, AI use patterns, and perceptions regarding various AI-related issues. Both descriptive and between-group c…

The problem

In a survey of 70 urology residency applicants, 86% had encountered inaccurate or misleading AI-generated information, yet only 44% always verified AI outputs.

In brief

Accuracy was a recurring issue: 86% had met inaccurate or misleading AI-generated information at least once, but only 44% always verified outputs. That gap matters in an application process where personal statements and interview answers shape selection.

Perceived benefits were modest. Among users, 61% reported at least slight improvement in application material quality, while 44% said the same about interview preparation. Opinions on rules were split: 23% backed restricting applicant AI use, and 54% backed prohibiting programs from using AI.

Main points

  1. Seventy complete responses were analysed, a 45% response rate, from applicants at a single academic urology program.
  2. The survey was distributed after rank list submission and before match day in the 2025 to 2026 cycle.
  3. 51% of applicants reported using AI in the application process, but they wrongly believed their peers used it more.
  4. ChatGPT was the most used AI tool, at 83%.
  5. AI was used most for personal statement writing, application content and interview preparation, which were also rated most helpful.
  6. 61% of AI users reported at least slight improvement in application material quality; 44% said so for interview preparation.
  7. Only 23% supported restricting applicant AI use, while 54% supported prohibiting AI use by residency programs.

The full story

The anonymous survey went to all applicants at one academic urology residency program after they submitted rank lists and before match day. A 29-item subset covered demographics, application components, AI use patterns and attitudes to AI-related issues, summarised with descriptive and between-group statistics.

Use was widespread but uneven. About half of respondents used AI, and those who did relied mainly on ChatGPT. They applied it chiefly to personal statements, application content and interview preparation, and rated those uses most helpful.

Reliability concerns were common. Most applicants had seen wrong or misleading AI output, yet fewer than half always checked it. Benefits were reported as modest, with interview preparation judged less improved than written materials.

Attitudes to governance differed by who would use AI. Few applicants wanted limits on their own use, but a majority wanted residency programs barred from using AI. The authors conclude that ethically integrated tools could improve efficiency, equity and outcomes, a view stated as potential rather than measured.

Sources

1 on file
  1. Journal of Surgical Educationdoi:10.1016/j.jsurg.2026.104254
    Peer-reviewed2026-11-16

Talk it over

Three questions to start on, two more if you have time
86% of the 70 applicants had met inaccurate or misleading AI information, yet only 44% always verified outputs. If you were one of those applicants, what would make you check every output?

Start with a question

Urology applicants say AI gave wrong info, yet only 44% always checked ithttps://truvace.com/item/artificial-intelligence-in-the-urology-residency-match-use-perceptions-and-impli

  1. 86% of the 70 applicants had met inaccurate or misleading AI information, yet only 44% always verified outputs. If you were one of those applicants, what would make you check every output?

    Answer in the thread
  2. The survey went to applicants at a single academic urology program, with a 45% response rate. How might the answers differ among applicants at other programs or among those who did not respond?

    Answer in the thread
  3. Only 23% backed restricting applicant AI use, while 54% backed prohibiting programs from using AI. Who should decide the rules for each side, and what should they weigh?

    Answer in the thread
If you have more time
  • Applicants thought their peers used AI more than the 51% who reported using it. What could that misjudgment change about how applicants behave in the match?

    Answer in the thread
  • Among users, 61% reported at least slight improvement in application material quality and 44% said so for interview preparation. What outcome data would show whether AI actually helped anyone in the match?

    Answer in the thread

What readers say