Emotion (or affective) artificial intelligence (AI) is a hot topic within the greater field of AI, in both, academic as well as practitioner circles. One of the industries with great potential for AI implementation is education. While emotion AI is commonly referred to as a field of growing interest, research in the specific context of education is still in its early stages and publications are few. This paper aims to discuss this emerging field of research on emotion AI in the context of education. The current body of literature can be grouped into three clusters: (1) concept and model development; (2) intelligent tutoring systems; and (3) students' state of mind. The review concludes that emotion-based improvements of learning systems surely hold a lot of promise yet still suffers one major shortcoming: that of appropriate responses to the detected emotions.