Disclosing artificial intelligence use in scientific research and publication: When should disclosure be mandatory, optional, or unnecessary?

Abstract

Currently there is a broad consensus among scholars that artificial intelligence (AI) tools can be used in research and publication, and that their use should be disclosed. Publishers and influential organizations, like the International Committee of Medical Journal Editors, have developed different and sometimes contradictory disclosure policies. We review some of these policies, examine the ethical reasons for disclosing AI use in research, and develop a framework for disclosure. We distinguish between mandatory, optional, and unnecessary disclosure of AI use, arguing that disclosure should be mandatory only when AI use is intentional and substantial. AI use is intentional when it is directly employed with a specific goal or purpose in mind. AI use is substantial when it 1) produces evidence, analysis, or discussion that supports or elaborates on the conclusions/findings of a study; or 2) directly affects the content of the research/publication. To support the application of our framework, we state three criteria for identifying substantial AI uses in research: a) using AI to make decisions that directly affect research results; b) using AI to generate content, data or images; and c) using AI to analyze content, data or images. Disclosure should be mandatory when AI use meets one of these criteria.