Generative artificial intelligence (AI) is increasingly integrated into scientific research, particularly in tasks related to academic writing and knowledge synthesis. However, empirical data on how orthopaedic researchers perceive, use, and regulate these tools remain limited. The purpose of this study was to: 1) assess attitudes, perceptions, and patterns of generative AI use among corresponding authors in orthopaedic research; 2) evaluate whether academic career stage influences perceptions, training, familiarity, and intended future use of AI; 3) identify individual and professional factors associated with positive perceptions of future AI use; and 4) examine disclosure practices, transparency issues, and perceived benefits and limitations of AI across the research and scientific publishing process.

We hypothesized that generative IA tools are already widely adopted by orthopaedic researchers, but that this adoption is heterogeneous and associated with variable perceptions, training levels, and disclosure practices.

An international, anonymous web-based survey was conducted among corresponding authors of orthopaedic research articles published between January 2024 and June 2025. The questionnaire evaluated demographics, AI familiarity, patterns of past and intended AI use throughout the research process, perceived benefits and limitations, training needs, and disclosure practices. Associations with a positive perception of future AI use were analyzed according to career stage using univariate and multivariate logistic regression analyses.

Among 216 respondents, 175 (81.0%) had previously used an AI-based chatbot and 96 (44.4%) had used AI in at least one of their last five publications, mainly for manuscript writing or editing (71/216; 32.9%), translation (64/216; 29.6%), and literature searching (56/216; 25.9%). Career stage did not influence intention to use AI in future research (non-senior: 79/95; 83.2% vs senior: 73/91; 80.2%; p = 0.123), and no significant differences were observed in perceived benefits, including manuscript writing support (3.83 ± 1.20 vs 3.63 ± 1.22; p = 0.275), literature search support (3.64 ± 1.20 vs 3.61 ± 1.20; p = 0.852), and overcoming language barriers (4.08 ± 1.13 vs 4.07 ± 0.99; p = 0.952), although senior researchers reported slightly greater concern about AI-related bias (4.30 ± 0.81 vs 4.01 ± 0.94; p = 0.032). In univariate analysis, male respondents were more likely to report a positive perception of future AI use than female respondents (133/157; 84.7% vs 18/28; 64.3%; p = 0.032), and in multivariate analysis, male gender was the only independent predictor (aOR = 2.85; 95% CI, 1.12–7.29; p = 0.029). Regarding transparency, 93 respondents (43.1%) consistently disclosed AI use, 144 (66.7%) feared negative perception when doing so, and 163 (75.4%) considered AI important or very important for future scientific production.

Generative AI is widely used among orthopaedic researchers but remains underreported, likely due to concerns about negative journal perceptions. Despite this, most researchers consider AI important for future scientific production, highlighting the need for clearer and harmonized guidelines on its use and disclosure among corresponding authors in orthopaedics.

IV; cross-sectional survey study.