Digital Human (DH) is an avatar that can reproduce a wide range of human body language to ensure an unprecedented user experience. It is backed by Artificial Intelligence that interprets users’ input and feeds back not just the requirements they need, but can also provide appropriate nonverbal communication as well. A DH bypasses human limitations by providing uninterrupted service. However, DHs raise some concerns including privacy and ethical issues, while impacting social and cultural fronts in terms of job disruption, moral dilemmas, and misinformation, in addition to other effects on humans. In this paper, we review the state-of-the-art of DHs, propose a taxonomy for it, present current implementations, and then train a machine learning model to allow a DH to recognize facial expressions. Finally, we discuss the challenges of the DH before concluding with our perspectives for future directions.