Queiroz Ana Beatriz L, Sartori Letícia Regina Morello, Lima Giana da Silveira, Moraes Rafael R
Graduate Program in Dentistry, Universidade Federal de Pelotas, Brazil.
Graduate Program in Dentistry, Universidade Federal de Pelotas, Brazil.
J Dent. 2025 Jun 20:105923. doi: 10.1016/j.jdent.2025.105923.
This study mapped the editorial policies regarding the use and acknowledgment of artificial intelligence (AI) in dental journals.
Dental journals indexed in Web of Science were analyzed. Editorial guidelines and instructions for authors and reviewers were assessed. The outcome of interest was whether a journal reported an AI policy. Exposures of interest were Journal Impact Factor (JIF), Journal Citation Indicator (JCI), total number of citations, and percentage of open access content. Associations between journal metrics and the presence of AI policies were assessed using Poisson regression models with robust standard errors. Adjusted models were developed separately for JIF and JCI, including variables with p ≤ 0.2. Significance was set at p ≤ 0.05.
Among the 158 journals analyzed, 70.9% reported AI policies. Policies targeted authors (100%), reviewers (88.4%), and editors (46.4%). The most common areas addressed were authorship, language review, and writing assistance. Journals in the upper quartiles of JIF and JCI were about twice as likely to report an AI policy compared to those in the lowest quartile. In the adjusted models, no association was found between total citations and AI policy presence, whereas open access percentage showed a negative association. While AI tools are permitted to aid in language refinement, critical tasks were consistently required to remain human-led, with rare exceptions.
AI-related editorial policies were frequent in dental journals, and higher-impact journals were statistically more likely to report such policies.
Clearer and more standardized AI policies may promote transparency and ethical AI use.
本研究梳理了牙科期刊中关于人工智能（AI）使用及引用的编辑政策。
对科学网索引的牙科期刊进行分析。评估作者和审稿人的编辑指南及说明。关注的结果是期刊是否报道了人工智能政策。关注的暴露因素是期刊影响因子（JIF）、期刊引用指标（JCI）、总被引次数和开放获取内容百分比。使用具有稳健标准误的泊松回归模型评估期刊指标与人工智能政策存在之间的关联。针对JIF和JCI分别建立调整模型，纳入p≤0.2的变量。显著性设定为p≤0.05。
在分析的158种期刊中，70.9%报道了人工智能政策。政策针对作者（100%）、审稿人（88.4%）和编辑（46.4%）。涉及的最常见领域是作者身份、语言审核和写作协助。与最低四分位数的期刊相比，JIF和JCI上四分位数的期刊报道人工智能政策的可能性约为两倍。在调整模型中，未发现总被引次数与人工智能政策存在之间的关联，而开放获取百分比显示出负相关。虽然允许使用人工智能工具辅助语言润色，但关键任务始终要求由人类主导，仅有极少数例外。
牙科期刊中与人工智能相关的编辑政策很常见，影响较大的期刊在统计学上更有可能报道此类政策。
更清晰、更标准化的人工智能政策可能促进人工智能使用的透明度和道德性。
