LLMs for Requirements Engineering: A Meta-Analysis on Trends and Challenges

Requirements Engineering (RE) is a critical yet error-prone phase of software development, where poor requirements management contributes to project failures. Large Language Models (LLMs) offer potential to automate tasks such as requirement generation, analysis, and validation. This meta-analysis examines 29 overview studies and 93 primary research papers to characterize current LLM applications in RE. We identify trends in task automation, model usage, evaluation practices, and domain-specific constraints, highlighting both opportunities and limitations. While LLMs can support drafting and preliminary analysis, human oversight remains essential, and challenges persist in reproducibility, industrial adoption, and methodological rigor. Emerging approaches—including retrieval-augmented generation, multi-agent frameworks, and formal-method integration—point to promising directions for advancing LLM-assisted RE.