Consider these research scenarios - click to categorise as acceptable, questionable, or problematic
A non-native English speaker uses AI to improve the grammar, clarity, and academic tone of their manuscript while keeping all scientific content and arguments unchanged, acknowledging AI assistance in their submission.
A researcher inputs their broad research interests and preliminary findings to AI, which suggests specific research questions and hypotheses. They select and refine the most promising ideas for their next study.
A researcher uses AI to write first drafts of methodology and results sections based on their data and notes, then extensively revises and fact-checks before submission, noting AI assistance in acknowledgments.
A researcher feeds abstracts of 50 relevant papers to AI and asks it to identify key themes and gaps in the literature, using this analysis as a starting point for their own comprehensive review.
A reviewer uses AI to help structure their review comments and suggest additional points to consider, while keeping the actual scientific judgment and recommendations entirely their own.
A researcher with dyslexia relies on AI to help structure and organise their written work, as traditional writing is extremely challenging for them. They don't disclose this AI use, considering it an accessibility accommodation rather than academic assistance.
A researcher uploads their statistical results to AI and asks it to explain patterns and suggest interpretations, using some of these insights in their discussion while independently verifying the key claims.
A researcher uses AI to help structure and write sections of a grant proposal, ensuring all scientific content reflects their expertise, and clearly stating AI assistance was used in the application.
A research team uses AI as a brainstorming partner to generate creative approaches to a research problem, treating AI suggestions as starting points for team discussion and independent evaluation.