Impact of an AI workshop on knowledge and attitudes toward AI in scientific publishing among surgeons at an international abdominal wall surgery congress.

Abstract

Background: Generative artificial intelligence (AI) tools are increasingly used for scientific writing, literature synthesis, and peer-review, yet their responsible use requires awareness of hallucinations, fabricated references, confidentiality risks, authorship limitations, and disclosure requirements.

Objective: To evaluate the perceived impact of a focused educational workshop on knowledge and attitudes toward AI in scientific publishing among participants in an abdominal wall surgery workshop.

Methods: An anonymous pre-post survey study was conducted during the JAWS Workshop 2026 on Artificial Intelligence in abdominal wall surgery. Participants voluntarily completed a 10-item baseline survey before two educational presentations and a 10-item post-workshop survey immediately afterwards. Responses were analysed descriptively. Because the questionnaires were not identical, formal paired analysis was restricted to the common ordinal knowledge item using anonymous participant codes.

Results: Thirty-nine participants completed the baseline survey and 41 completed the post-workshop survey; 34 anonymous codes could be matched. Before the workshop, 48.7% reported basic knowledge, 35.9% intermediate knowledge, 5.1% advanced knowledge, and 10.3% no knowledge. AI tools were used occasionally, frequently, or systematically by 89.7% of respondents. The most frequent uses were language editing (74.4%), manuscript structuring (46.2%), literature summarization (46.2%), and data analysis or interpretation (41%). After the workshop, intermediate or advanced self-reported knowledge was observed in 68.3% of respondents. In matched analysis, self-reported knowledge scores improved in 18/34 participants (52.9%; Wilcoxon signed-rank, p = 0.0005). Post-workshop, 92.7% agreed or strongly agreed that they better understood AI limitations, 97.6% would almost always or always verify AI-generated references, 85.4% supported disclosure of AI use in manuscripts, and 87.8% considered society guidelines recommended or essential.

Conclusion: AI use for scientific activities was common among workshop participants, and a brief targeted educational intervention was associated with improved self-reported knowledge and support for responsible AI use. Scientific societies and journals may play an important role in promoting the responsible integration of AI into scientific publishing through focused educational initiatives.