When AI tools contribute to your research, they should be disclosed in the acknowledgments section rather than assigned CRediT (Contributor Roles Taxonomy) roles, because CRediT is designed exclusively for human contributors who can take accountability for the work.
Where AI Assistance Belongs
Most major publishers (including Nature, Science, and Elsevier journals) now require authors to declare AI tool usage in a dedicated statement within the acknowledgments or methods section. This disclosure should specify which tools were used and for what purposes—for example, "ChatGPT was used to improve the clarity of the abstract" or "Grammarly assisted with grammar checking in the manuscript draft."
Why AI Tools Don't Get CRediT Roles
The CRediT taxonomy includes 14 standardized roles like Conceptualization, Methodology, Writing – Original Draft, and Formal Analysis. These roles carry implicit accountability: the person assigned must be able to defend, explain, and take responsibility for that contribution. Since AI tools cannot verify facts, respond to peer review, or be held accountable for errors, they cannot ethically be listed as contributors with CRediT roles.
Handling Borderline Cases
When AI tools play a substantial role—such as generating code, analyzing datasets, or drafting significant portions of text—transparency becomes even more critical. Describe the tool's contribution in enough detail that readers and reviewers can assess its impact on the work's validity. For instance, if you used an AI assistant to help structure your literature review, note that human authors verified all sources and claims. WisPaper's TrueCite automatically verifies citations against real published metadata while you write, helping ensure that AI-suggested references are legitimate and properly formatted.
The Human Accountability Standard
All authors listed with CRediT roles must meet ICMJE authorship criteria: substantial contributions to conception/design or data acquisition/analysis, drafting or critical revision, final approval, and agreement to be accountable. AI tools, regardless of sophistication, cannot meet the accountability requirement—making transparent acknowledgment rather than authorship credit the appropriate path.
