The Transparency Card: A Structured Reporting Framework for Agent-Assisted Research
Technical report (Zenodo), 2026. doi:10.5281/zenodo.19772806
Abstract
As AI agents become active contributors to scientific research, the question of how to report their involvement becomes urgent. This report introduces two complementary reporting instruments. The Transparency Card provides comprehensive, task-level disclosure of AI contributions. The Agentic R&D Card offers a compact, phase-level summary suitable for routine inclusion as a publication appendix. Both instruments are structured, machine-readable, and designed for incremental adoption.
Related work
- A benchmark of expert-level academic questions to assess AI capabilitiesNature, 2026
- Modeling the impact of research data unavailability on scienceJournal of Informetrics, 2026
- Agentic-JEPA: A Self-Supervised World Model for Planning in Text-Based Agent EnvironmentsPreprint, 2026
- Media Bias Bias-Mitigated Dataset (MBBMD): A Hierarchical, Perspectivist, and Counterfactually-Augmented Corpus for Bias Detection in Spanish NewsProcesamiento del Lenguaje Natural (SEPLN) 2026, 2026