Artificial Intelligence (AI) Driven Networks: Challenges and Solutions Abstract: Artificial Intelligence (AI) such as Machine learning is able to discover the hidden and complex relationships from input data to the network output. In the literature, AI has been explored to address multiple network issues such as routing and clustering, network scheduling, network security, network quality of service (QoS) as well as dynamic spectrum access. It has drawn much attention as a key tool for the design of future wireless networks including 5G. However, the evolution towards AI-based data driven network is at the early stage. There are still significant challenges in the area. For example, how to choose appropriate AI algorithms for different networking applications such as Internet of Things (IoT), connected health, smart cities and connected/autonomous vehicle? What types of data should be collected from networks so that the network behavior/pattern could learned to optimize whole network performance? How the AI could re-design the networks and secure them, and etc.. In this work, we will present these challenges in details and discuss the potential solutions that could overcome performance limitations by exploring AI as a tool for the future wireless networks.