Artificial intelligence (AI) draws its power from big data. However, accessing and processing big data may not always be possible due to both confidentiality and hardware requirements for high computational performance. Federated learning (FL) is a new concept proposed to solve the aforementioned privacy & big data dilemma. FL is also a framework that performs updating of the parameters of a common AI model trained by the different participants and then combining the updated parameters through the coordinator while protecting data privacy. Due to the modular design of the FL concept, the workload is shared among the participants while protecting data privacy. It also provides advantages like scalability in terms of collaborator count and higher performance and lower execution time for some sort of problems. Depending on the similarity of the feature and sample spaces of the collaborators, there are some FL approaches such as horizontal, vertical and transfer. FL is applicable to any field in which machine learning methods are utilized and the data privacy is an important issue. Healthcare services, transportation sector, financial technologies and natural language processing are the prominent fields where horizontal FL concept is applied. On the other hand, AI-based collaborations between the sectors can be developed with vertical and transfer FL concepts.