Human vs LLM Workflows in Evidence Synthesis Scoping Reviews
August 28, 2026
A study comparing human and LLM title-and-abstract screening workflows found that no workflow, human or LLM-based, recovered all verified eligible records. The research highlights critical recall-workload trade-offs and consistency issues when using LLMs for complex evidence synthesis.
HOW THIS AFFECTS YOU
●
researcherBe cautious when relying on LLMs for literature screening due to potential recall failures.