Human-AI collaboration to identify literature for evidence synthesis

• Published in: Cell Reports Sustainability Vol. 1; no. 7; p. 100132
• Main Authors: Scott Spillias, Paris Tuohy, Matthew Andreotta, Ruby Annand-Jones, Fabio Boschetti, Christopher Cvitanovic, Joseph Duggan, Elisabeth A. Fulton, Denis B. Karcher, Cécile Paris, Rebecca Shellock, Rowan Trebilco
• Format: Journal Article
• Language: English
• Published: Elsevier 01.07.2024
• Subjects: artificial intelligence; large language models; natural-language; processing; scientific publication; systematic review
• ISSN: 2949-7906
• DOI: 10.1016/j.crsus.2024.100132

Systematic approaches to evidence synthesis can improve the rigor, transparency, and replicability of a literature review. However, these systematic approaches are resource intensive. We evaluate the ability of ChatGPT to undertake two stages of evidence syntheses (searching peer-reviewed literature and screening for relevance) and develop a collaborative framework to leverage both human and AI intelligence. Using a scoping review of community-based fisheries management as a case study, we find that with substantial prompting, the AI can provide critical insight into the construction and content of a search string. Thereafter, we evaluate five strategies for synthesizing AI output to screen articles based on predefined inclusion criteria. We find that low omission rates (<1%) of relevant literature by the AI are achievable, which is comparable to human screeners. These findings suggest that generalized AI tools can assist reviewers to accelerate the implementation and improve the reliability of literature reviews, thus supporting evidence-informed decision-making. Science for society: This study underscores the transformative role AI can play in streamlining systematic literature reviews, a cornerstone of evidence-based decision-making across disciplines. By demonstrating that AI can closely match human accuracy in filtering relevant studies, this research paves the way for more efficient, rigorous syntheses of scientific knowledge. Looking ahead, this work invites transdisciplinary collaboration, combining expertise in AI, data science, and various research fields to further refine these tools. The potential societal impact is substantial; policymakers, practitioners, and researchers can leverage these advancements to more rapidly inform actions on urgent issues related to environmental management.