Postoperative delirium (POD) is associated with a variety of poor prognoses and has important medical and public health implications. POD can occur in surgical populations of different ages and types of surgeries. Accurate prediction of POD can help to improve the prognosis of patients. The occurrence of POD is related to multiple factors, and the existing prediction tools are limited in number and deficient in predictive ability. With the rising use of artificial intelligence in the medical community, new methods for the prediction of POD have been provided. In this review, we summarize the current research on the prediction of POD by artificial intelligence methods in recent years, reveal their advantages and disadvantages, and discuss the prospects for providing new ideas for the diagnosis and treatment of POD.