Generative AI and Project Risk Management: Human-AI Collaboration Under Uncertainty

Generative artificial intelligence (GenAI) is increasingly entering project-management activities that involve large volumes of unstructured information, uncertain conditions, and consequential managerial judgment. Project risk management is a particularly important setting because AI may help project teams identify risks, synthesize dispersed evidence, generate scenarios, and formulate response alternatives, yet the consequences of misplaced reliance can be substantial. This integrative literature review develops a process-based framework for human-GenAI collaboration in project risk management. The framework links project risk information to GenAI-assisted risk identification, GenAI-supported assessment and scenario analysis, human verification and contextual interpretation, human-AI response development, human accountable choice, and monitoring and feedback. The article further distinguishes AI capability from AI authority and proposes four interacting dimensions for determining appropriate GenAI authority: uncertainty, risk consequences, verifiability, and managerial expertise. A four-level authority continuum is introduced, ranging from informational support to delegated decision authority. The synthesis suggests that high uncertainty can increase the value of GenAI for exploration while simultaneously reducing the justification for autonomous AI choice. The article concludes that successful adoption should be evaluated through the quality of the human-AI decision architecture rather than the amount of human work replaced. Keywords: generative artificial intelligence, project risk management, human-AI collaboration, uncertainty, decision authority, risk governance, managerial judgment