Artificial Intelligence and Large Language Models in Government Document Management: A Systematic Review of Applications, Challenges, and Implementation Strategies

Abstract

Government agencies face escalating challenges in managing vast amounts of documents, including policy papers, legal briefs, and administrative records. Traditional document management systems—reliant on manual classification and keyword-based retrieval—struggle with scalability, accuracy, and security. This study investigates how artificial intelligence (AI), extensive language models (LLMs), can transform government document management by automating classification, enhancing retrieval accuracy, and ensuring compliance with legal standards. A systematic literature review was conducted following the PRISMA guidelines. We searched Google Scholar, IEEE Xplore, ACM Digital Library, and Web of Science for peer-reviewed studies published between 2018 and 2024. Inclusion criteria focused on empirical studies, case analyses, or theoretical frameworks involving LLM or AI applications in government document management. After screening 312 records, 42 studies met the eligibility criteria and were analyzed thematically. Findings reveal that LLMs significantly outperform traditional methods in document classification (accuracy ↑15–30%) and natural language querying (user satisfaction ↑40%). Key applications include automated policy analysis, FOIA request processing, and cross-departmental knowledge retrieval. However, implementation barriers persist: technical complexity (52% of cases), staff resistance (38%), and regulatory compliance issues (29%). LLMs offer transformative potential for government document management, but success hinges on domain-specific training, robust governance frameworks, and stakeholder engagement. This study provides a practical roadmap for public sector agencies to adopt LLMs responsibly, emphasizing phased deployment, ethics oversight, and interoperability with legacy systems.