What is HAI?

What is Human-Agent Interaction? We provide the following for researchers who are considering joining our HAI community. As we envision HAI to be a broad and inviting area of research that will continue to evolve with the changing research landscape, this should be taken as a guiding discussion only and any relevant work is welcome.

What is an agent? An agent is an object or technology that people interact with as if it is able to act with its own purposes, motivations, and intentions. Having autonomy is not sufficient. For example, people generally do not interact with telephones or elevators as agents although there is autonomy behind the scenes. Conversely, some people talk to their cars or boats as if they have their own intentions or personalities, even if this is not the case. It is the point where people shape their interactions with an object or technology as if it has purposes, motivations, or intentions, where we have human-agent interaction. Often, technologies are designed to encourage people to give them agency, for example, by using social methods such as speech or gestures, or having a personality, although this is not necessary for agency.

People attribute agency to many things and technologies in their everyday lives, which is a common area of research, for example, with robots, virtual characters and video games, virtual reality, smart cars, and so forth. By considering the broader field of Human-Agent Interaction, we have an opportunity to transcend these traditional categories and to share knowledge and learn from each other, regardless of the embodiment of our agent. Findings in human-robot interaction are likely to have implications for many kinds of agents.

The HAI conference aims to bring together an interdisciplinary community on agents. In addition to work on agents themselves, this conference also includes novel technologies and algorithms for use by agents, human-human studies that can inform interactive agent design, how psychological and cognitive models of agency are relevant to
agents, and so forth.