Transparency of AI in science: GAIDeT taxonomy

BSPU researchers — Yana Sychikova (Vice-Rector for Research) and Natalia Tsybuliak (Department of Applied Psychology and Speech Therapy) — together with independent Japanese researcher Jaime A. Teixeira da Silva and Serhii Nazarovets from Borys Grinchenko Kyiv Metropolitan University, proposed a solution. Their taxonomy classifies research tasks into macro- and micro-levels: from hypothesis formulation and literature review writing to data analysis, programming, or ethical risk assessment. For each task, GAIDeT specifies whether it can be partly performed by AI and how much human oversight is required.

A Tool for Researchers

Alongside the taxonomy, the team developed an interactive GAIDeT Declaration Generator – a free online tool that helps authors quickly generate a transparent statement on AI use in their research. Such a declaration can be included in manuscripts under the CRediT (author contributions) section. This practice strengthens trust in publications and helps avoid misunderstandings.

Significance for the Academic Community

GAIDeT is designed not only for authors but also for editors, publishers, and funding institutions. It promotes transparency, ethics, and reproducibility in the use of artificial intelligence, while safeguarding a key principle: the final responsibility always remains with humans.