AI ‐Assisted Scientific Figures in Scholarly Publishing: A Risk‐Based Taxonomy by Evidentiary Function

2026/06/27 by Malcolm Koo, Ming‐Chi Lu
Computer Science · Medicine · Social Sciences · #Artificial Intelligence in Healthcare and Education #Computational and Text Analysis Methods #Research Data Management Practices

paper · doi:10.1002/leap.2085

openalex publication_date 2026/06/27 · openalex created_date 2026/06/28 · openalex updated_date 2026/07/22

Abstract

Summary AI‐assisted figures should be assessed by what they claim to represent, not by AI use alone. Existing AI‐image policies may not offer enough granularity for diverse figure types in scholarly publishing. The taxonomy classifies figures as explanatory, analytical, simulated, or evidentiary. Each figure type carries distinct integrity risks and requires proportionate editorial safeguards. Figure‐level classification could improve disclosure, auditability, labeling, and preservation of original observations.

Citations

• GAIDeT (Generative AI Delegation Taxonomy): A taxonomy for humans to delegate tasks to generative artificial intelligence in scientific research and publishing

Related

• AI-Generated Figures in Academic Publishing: Policies, Tools, and Practical Guidelines
• Beyond AI disclosure: Claim accountability and responsible research in scholarly publishing
• How AI use in scholarly publishing threatens research integrity, lessens trust, and invites misinformation
• Provenance, Not Prohibition: A Framework for AI in Scholarly Publishing
• Artificial Intelligence (AI) guidance for authors, peer reviewers, and editors: A content analysis of journal policies