Application, risks and countermeasures of generative artificial intelligence in scientific research

Generative artificial intelligence, with its powerful language understanding and generation capabilities, has demonstrated tremendous potential in numerous fields and has been widely applied in scientific research, enhancing the overall efficiency of scientific endeavors. However, as generative AI tools becomes more deeply integrated, its potential risks in the field of scientific research are also increasingly prominent. This paper elaborated on the application status of generative AI in scientific research and discussed the risks that may arise when it is involved in scientific research, including issues related to literature extraction, originality, content writing, image generation, peer review quality, and value judgment. These deficiencies impact the quality and output of scientific research results, exacerbate academic misconduct, pose a serious threat to academic integrity, and may even lead to information security issues. This paper pointed out that scientific journals must be highly vigilant about the potential risks posed by generative artificial intelligence (GAI) technology and should promptly adopt effective countermeasures. Doing so will enhance their capability to identify and handle AI-generated content (AIGC), ensure the responsible use of generative AI tools, maintain scientific integrity, and safeguard the rigor and scientific quality of academic publishing. It also emphasized that current AI lacks the capacity for value judgment and social context awareness, proposing corresponding recommendations to clarify that ultimate discretion and primary responsibility in research activities ultimately rest with humans.