Generative Artificial Intelligence for Project Management: A Review of Architectures, Applications, and Governance

초록

Generative artificial intelligence is reshaping contemporary project management (PM) through advances in Large Language Models (LLMs), multimodal foundation models, and autonomous AI agents. Persistent cost and schedule overruns across IT, construction, and infrastructure projects highlight the limitations of traditional PM frameworks and motivate the adoption of adaptive, data-driven decision systems. Between 2020 and 2026, generative AI expanded beyond predictive analytics to support document understanding, long-horizon planning, automated reasoning, and tool-based workflow execution. This review synthesizes evidence from empirical research, industrial applications, and reproducible preprints to evaluate how generative AI enhances planning, scheduling, risk and cost estimation, resource allocation, progress monitoring, document analysis, and stakeholder communication. We propose a unified technical taxonomy covering LLMs, multimodal models, retrieval-augmented systems, and autonomous agents, and evaluate their empirical performance relative to classical machine learning approaches. The review further identifies socio-technical challenges related to hallucination, automation bias, data governance, and human–AI coordination. Collectively, these insights articulate the emerging landscape of generative AI–enabled project management and outline pathways for safe, scalable, and high-impact adoption across complex project ecosystems.

키워드