Should We Collaborate with AI to Conduct Literature Reviews? Changing Epistemic Values in a Flattening World

Summary

The authors revisit whether and how researchers should collaborate with AI to conduct literature reviews. They observe that AI tools accelerate search and screening, especially at scale, but may compromise quality, transparency, and explainability — particularly when based on machine learning or generative AI. Expert systems are seen as less likely to harm these tasks, and any AI method must preserve researchers' ability to critically select, analyze, and interpret the literature.

Contribution

Calls for further reflection on the epistemic values at risk when different AI tools are used at various stages of the review process, and argues that the iterative, scope-redefining nature of reviews requires safeguards for transparency and critical engagement.

Method

Conceptual/critical opinion piece; no empirical evaluation reported in the abstract.

Relevance to RISE

Critical conceptual paper questioning whether AI-assisted literature review changes the epistemic values of the resulting scholarship. Direct counter-weight to the catalog's literature-synthesis projects (storm, open-scholar, paper-qa) — they answer 'how', this paper asks whether IS should.

Critique / open questions

Conceptual piece without empirical comparison of AI-assisted vs. human-only reviews; claims about quality compromise and the relative merits of expert systems are not benchmarked.