The most richly diverse field of facial recognition is Emotion AI. This field builds a relationship between artificial intelligence and human emotional intelligence. Emotion AI applies identity recognition to decipher facial expressions, voice patterns, and even physiological signs to comprehend human feelings. This chapter explores the primary limitations and challenges of Emotion AI. It concentrates on data limitations and bias, technological challenges in implementing Emotion AI technology in real life, ethical hazards, and legal problems concerning trust in human and AI interactions. We use case studies and applications in health care, education, marketing, and public safety as a vehicle for illustrating the potential and risks associated with the use of these applications. We also stress the necessarily interdisciplinary nature of Emotion A.I., overlapping AI research with psychology, human computer interaction (HCI), and ethics and stewarship. We conclude by providing a roadmap based on future directions which highlight the importance of multimodal approaches (viz. AI deployment in an interdisciplinary manner), explainable and transparent architectures, culturally aware modelling, and sustainable practices. In combination with technical invention, an ethical consideration and human-centered practice of design, Emotion A.I. has the potential to develop/invest toward a transformative yet sustainable changes to human wellness and interaction across disciplines.