Abstract:With the rapid growth of investment in scientific research funds in China, traditional manual audit supervision methods are confronted with low efficiency, poor accuracy, serious lag and other problems when dealing with large-volume and diversified scientific research financial data. Artificial Intelligence (AI) is being gradually applied to the audit supervision of scientific research fund. Based on the life cycle theory, this paper constructs a mechanism framework for AI-enabled audit supervision covering five links of scientific research fund, namely fund application, fund allocation, fund budgeting, fund expenditure and fund reporting, from three levels of technological empowerment, process embedding and supervision effectiveness. Then, combined with publicly reported cases, this paper analyzes the operation mechanism and practical effects of AI technologies in anomaly detection, risk early warning and other aspects, and clarifies the verification and supporting relationship of the cases to the mechanism framework. The study shows that AI not only improves the automation and intelligence level of audit supervision of scientific research fund, but also provides feasible technical support for building an efficient and transparent governance system for scientific research fund. Finally, this paper discusses the practical challenges encountered in the application process and puts forward countermeasures and suggestions.