To Affinity and Beyond: Interactive Digital Humans as a Human Computer Interface

The field of human computer interaction is increasingly exploring the use of more natural, human-like user interfaces to build intelligent agents to aid in everyday life. This is coupled with a move to people using ever more realistic avatars to represent themselves in their digital lives. As the ability to produce emotionally engaging digital human representations is only just now becoming technically possible, there is little research into how to approach such tasks. This is due to both technical complexity and operational implementation cost. This is now changing as we are at a nexus point with new approaches, faster graphics processing and enabling new technologies in machine learning and computer vision becoming available. I articulate the issues required for such digital humans to be considered successfully located on the other side of the phenomenon known as the Uncanny Valley. My results show that a complex mix of perceived and contextual aspects affect the sense making on digital humans and highlights previously undocumented effects of interactivity on the affinity. Users are willing to accept digital humans as a new form of user interface and they react to them emotionally in previously unanticipated ways. My research shows that it is possible to build an effective interactive digital human that crosses the Uncanny Valley. I directly explore what is required to build a visually realistic digital human as a primary research question and I explore if such a realistic face provides sufficient benefit to justify the challenges involved in building it. I conducted a Delphi study to inform the research approaches and then produced a complex digital human character based on these insights. This interactive and realistic digital human avatar represents a major technical undertaking involving multiple teams around the world. Finally, I explored a framework for examining the ethical implications and signpost future research areas.