AI, originality, and attribution: Researchers' perspectives on distinguishing contributions

Area of Science:

• Scholarly research practices
• Ethics in artificial intelligence
• Human-computer interaction

Background:

• Artificial intelligence (AI) integration challenges traditional scholarly norms regarding originality, contribution, and authorship.
• Existing policies for AI in research are developing, but understanding researchers' subjective navigation of these ambiguities is lacking.
• This gap impacts the integrity of research involving AI.

Purpose of the Study:

• To explore how researchers perceive and distinguish between human and artificial intelligence contributions in their work.
• To understand the practical strategies researchers employ to navigate attributional ambiguity in AI-assisted research.
• To identify the ethical tensions and needs for clearer guidelines arising from AI integration.

Main Methods:

• Conducted semi-structured interviews with 18 researchers across STEM, Social Sciences, and Humanities.
• Included participants at various career stages: PhD students, Postdoctoral Researchers, and Faculty.
• Analyzed interview data using reflexive thematic analysis, guided by Attribution Theory.

Main Results:

• Researchers primarily view artificial intelligence (AI) as a sophisticated tool, not an independent collaborator.
• Subjective heuristics, like 'gut feelings' and effort-based attribution, are used to manage attributional ambiguity.
• Significant ethical tensions arise, coupled with a strong desire for clearer, more nuanced guidelines.

Conclusions:

• Researchers encounter cognitive and practical challenges when applying traditional integrity norms to AI-assisted work.
• Findings underscore the necessity for critical dialogue and reflective practices.
• Nuanced guidelines are crucial for upholding research integrity and integrating human value with AI capabilities.