Research Integrity and Generative AI

This White Paper explores the implications for research integrity of recent advances in generative artificial intelligence (AI). Our objective is to support researchers in safely and responsibly understanding and managing the issues that may arise for research activities that involve generative AI tools. Our scope includes generative AI technologies that facilitate generation of text, computer code, images, artwork, and other media, and particularly those that involve generation from a linguistic (user-specified) prompt. The White Paper covers five main topics: A summary of the characteristics and capabilities of generative AI tools. A review of the enduring role of established Research Integrity principles, which continue to guide the conduct of all research, whether or not it involves the use of AI tools at any stage. The identification of seven key areas of elevated risk for research integrity in misuse of generative AI, including research falsification, misinformation, transparency, reproducibility, and bias. The identification of emerging opportunities for progress in the responsible conduct of research using generative AI with potential benefits to research, researchers, and wider society. The suggestion of selected actions to assist in maintaining or enhancing research standards and integrity in the face of both positive and negative disruptions that generative AI tools, such as ChatGPT, may cause to research activities.