The chapter looks into the developing bond between artificial intelligence (AI) and edge computing. In particular, the idea of using AI to intelligently offload computations. As the number of latency-sensitive applications have increased and the use cases for smart devices has widened, resource allocation at the edge has become critical. We discuss AI-based methods that intelligently determine what and when to transfer compute-intense tasks from resource-constrained edge devices to nearby edge servers or cloud environments. Pragmatic methods, RL optimization procedures and ML research exercises are the main focus of the standardized testing. To illustrate real-world examples, smart cities, autonomous vehicles, and industrial IoT are further explored. This chapter focuses on the development of a new hybrid offloading framework, synthesizing some of the greatest qualities of the predictive analytic and real-time learning to put into practice. These challenges including device heterogeneity, network variability, privacy, etc., are elaborated. Finally, in the concluding chapter, we argue the need for open problems that inform the path toward a sustainable, secure, AI-enabled edge computing.