A hybrid framework for creating artificial intelligence-augmented systematic literature reviews

Abstract

Abstract The integration of artificial intelligence (AI), particularly generative AI (GenAI) and large language models (LLMs), into systematic literature reviews (SLRs) represents a transformative advancement in research methodologies. This paper proposes a hybrid framework combining AI’s computational power with the epistemological rigor of human expertise, anchored in transparency, validity, reliability, comprehensiveness, and reflective agency. Through three interconnected phases—design, study collection, and interpretation—the framework employs AI model selection, knowledge base curation, and iterative prompt engineering to enhance scalability, uncover interdisciplinary connections, and ensure methodological integrity through robust human oversight. It addresses key SLR challenges, including handling vast datasets, ensuring reproducibility, and maintaining epistemic rigor while leveraging advanced AI capabilities. Key innovations include cyclical validation, inter-model comparisons, and sensitivity testing to enhance trustworthiness and mitigate biases. The framework aligns AI processes with ethical standards and research objectives by emphasizing domain-specific LLMs, reliability metrics, and standardized reporting protocols. It establishes SLRs as a foundation for advancing knowledge in complex, interdisciplinary research landscapes, harmonizing AI efficiency with human expertise.

Topics & Concepts