Evaluating the Capabilities of Large Language Models in the Literature Review Process

We examined the effectiveness of the large language models GPT-5 and Gemini in conducting the literature search, literature relevance screening, and literature quality screening tasks for a literature review. Results showed that while these models are not yet ready to fully replace human involvement in the literature review process, they may be able to play a supporting role in these tasks, potentially reducing the workload of human reviewers. The utility of these models is likely to improve over time as the models improve, and thus future research should continue to evaluate various models’ abilities to support in literature review tasks.