Editorial · CASRAI · Generative AI use and disclosure
AI Literature Review and Research Synthesis Tools: Elicit, Consensus, Rayyan and Disclosure Requirements
A new generation of artificial intelligence tools has entered the research workflow at the literature review and evidence synthesis stage. Elicit, developed by Ought (now the Elicit PBC), uses language models to extract findings from papers and assist with systematic reviews. Consensus provides AI-driven synthesis of evidence from published research. Rayyan offers machine learning-assisted screening for systematic review teams. SciSpace and Perplexity offer research assistant capabilities drawing on published literature. As these tools become more capable and more widely used, the question of when and how their use should be disclosed in research outputs has become urgent. COPE guidance, emerging journal AI tools policies, and funder transparency expectations all point towards a disclosure norm, but the details remain contested and inconsistently applied across disciplines and publishers.
Related editorial in this domain
More on Generative AI use and disclosure
NIST AI RMF & the Generative AI Profile Explained
The NIST AI Risk Management Framework is a voluntary US framework for managing AI risk; its Generative AI Profile (AI 600-1) adapts it to generative systems. A neutral explainer of how organisations use them.
AI Model Documentation: Datasheets and Model Cards
Model cards and datasheets for datasets are structured documentation that record what an AI model or dataset is, how it was built and where it should not be used. This article explains each, what they record, and how they support reproducibility and accountability.
Why generative AI cannot be an author — and what to disclose instead
Every major publisher and integrity body now agrees a generative AI tool cannot be an author. The reasoning rests on accountability, and it points directly at what authors should do instead: disclose the use clearly and verify the output.
