Future cellular communication is envisioned to rely on intelligent networks enabled by Artificial Intelligence (AI)/Machine Learning (ML) methods. However, research on AI/ML-based networks is still in its infancy, and existing works do not address the whole process of integrating AI technologies into 5G and beyond networks. This systematic review assesses the current state of AI and ML applications for future wireless networks, focusing on four main topics: task offloading, resource allocation, spectrum management, and channel estimation. This work confirms that research in this area faces numerous challenges, including the lack of reliable datasets, limited information on tools and libraries, hardware impact, and simulation platforms. These factors make it difficult to compare and validate the performance of different AI/ML approaches, impairing their progress into standardized solutions. Additionally, the cost of deploying AI is often overlooked, undermining the credibility of the reported gains. Moreover, current surveys rarely analyze the types of datasets used, the runtime cost of models, or the simulation and ML tools adopted. These omissions hinder reproducibility and obscure the trade-offs between performance and implementation cost. The outcome of this review emphasizes the need for more efforts to address these challenges and advance the field of AI/ML-based network research.