When AI use becomes the norm: Researcher perspectives on AI disclosure policy and practice

Area of Science:

• Computational Social Science
• Bioinformatics
• Academic Publishing

Background:

• Growing mandates for artificial intelligence (AI) disclosure in academic publishing contrast with observed research practices.
• Limited understanding exists regarding researchers' perceptions, navigation strategies, and identified limitations of current AI disclosure policies.

Purpose of the Study:

• To explore researchers' lived experiences and perspectives on artificial intelligence (AI) disclosure requirements.
• To identify systemic limitations and disciplinary variations in AI disclosure practices within academic research.

Main Methods:

• Conducted semi-structured interviews with 14 researchers in bioinformatics and computational social science.
• Employed reflexive thematic analysis to interpret qualitative data on AI disclosure experiences.

Main Results:

• Identified fragmented and inconsistently enforced AI disclosure requirements.
• Revealed systemic limitations including scope ambiguity, research integrity risks, and disincentives for honest reporting.
• Highlighted a transparency paradox where honest AI disclosure may lead to professional penalties and disciplinary norms resisting uniform governance.

Conclusions:

• The compliance gap in AI disclosure stems from the interplay of structural conditions and ethical obligations, exacerbated by unverifiable self-reporting and disciplinary variations.
• Advocates for a structured AI contribution taxonomy as a practical alternative to current disclosure methods.
• Recommends field-sensitive adaptation for effective AI disclosure governance over uniform implementation across diverse research communities.