What does it mean to use artificial intelligence well in realist research? And where should we draw the red lines?
In this timely and balanced conversation, Alejandro is joined by Professor Andrew Booth, Professor in Evidence Synthesis at the University of Sheffield, whose work spans systematic reviews, qualitative evidence synthesis, information retrieval, and realist synthesis.
Andrew approaches AI not as a solution to every methodological challenge, nor as something researchers should simply reject, but as a potentially useful assistant. Drawing on his extensive experience in evidence synthesis and realist research, he considers where AI might genuinely enhance existing research practices and where its apparent convenience can become misleading.
But the conversation also tackles the risks. Andrew discusses hallucinated references, the persuasive plausibility of AI-generated answers and the danger of producing neat, fully formed CMO configurations when the evidence itself may bepartial, ambiguous, or better represented through incomplete configurations that require human synthesis and discussion.
Alejandro and Andrew also explore one of the deeper concerns surrounding AI: what happens to researchers’ intellectual development and academic voice when too much of the process begins with a blank page handed over to a machine? Andrew concludes with three practical principles for responsible AI use in realist research.
Whether you are a realist evaluator, realist reviewer, evidence synthesis researcher, information specialist, student, supervisor, or PhD researcher, this episode offers a practical and refreshingly nuanced discussion of how we might engage with AI without allowing it to replace the distinctly human work of thinking, judging, interpreting and explaining.
