AI辅助写论文，科技期刊怎么看——专访《柳叶刀》副主编萨宾娜·克莱纳特

Core Viewpoint - The integration of artificial intelligence (AI) in research activities has led to new forms of academic misconduct, such as paper writing, data fabrication, and implicit plagiarism, posing significant challenges to academic integrity. Group 1: AI Usage in Research - The use of generative AI in research is increasing, with only 7% of authors disclosing its use in submissions, contrasting sharply with surveys indicating over 50% usage among users [2][3] - Instances of "hallucination citations" have been observed, where references are cited that do not exist, with some papers containing up to 10 to 15 such citations [2] Group 2: Editorial Policies and Practices - The journal has implemented a checkbox for authors to declare the use of generative AI, requiring detailed information about the AI model used, its purpose, and its application in the manuscript [1] - Papers with significant undisclosed AI content will be rejected, and institutions may be informed for educational purposes on proper AI usage [3] Group 3: Acceptable and Unacceptable AI Uses - Generative AI is not permitted to replace researchers' work in generating scientific insights, analyzing data, or drawing conclusions [5] - AI can be used to improve grammar and language expression in writing, as well as to summarize existing research [7] Group 4: Challenges and Future Directions - The journal is establishing a research integrity committee to address the impact of generative AI on academic integrity, with a focus on monitoring ongoing cases and adapting to new guidelines [8] - Researchers and editors are advised to treat AI as an auxiliary tool, being cautious of its limitations and potential risks, including the de-skilling of future researchers [9]