ABSTRACT
Increasingly, humans are interacting not only with other humans, but also with artificial agents – and there is a growing effort to develop artificial team-mates for different team-based activities. Related work comparing human-human and humancomputer conversational interactions, competitive interactions, and cooperative interactions shows that there are significant differences between the way humans feel about, treat, and behave when interacting with humans – compared to the way they feel about and act towards/with artificial agents. One issue that has not been sufficiently studied in the related work is whether there are differences in player enjoyment when playing a team-mate game as a member of a human-human team versus a human-agent team – and if so, how to explain such differences. A number of empirical comparative studies were conducted in which participants played teammate games with either an artificial team-mate – or with a human team-mate who was connected over the network from another location. In fact, unknown to the participants, the “human” team-mate was the same as the artificial team-mate, so gameplay behavior was constant under all conditions. Participants consistently made claims about increased enjoyment when the team-mates was perceived to be human. This paper summarizes these results and proposes an explanation in terms of players’ inability to imagine artificial agents are capable of certain states or behaviors.
