Narrative contextualisation: Overriding the artificiality of Artificial Humane Agents
Abstract. The recent advancements in creating humane artificial characters with seemingly self-emergent social behavior have reached the technological and artistic level that challenges the argument that bridging the uncanny abyss between the non-human and humane artificial agents is still beyond the possible. A great part of the existing research on the topic has focused on developing the social communication skills related to face-to-face interaction between human and non-human agents, such including combinations of verbal and non-verbal expressiveness. These aspects alone continue requiring extensive efforts from the scientific community, while relatively little attention has been focused on the contextual aspects of such face-to-face settings. This talk argues that the challenge of reaching a sufficient level of human-likeness of humane artificial agents, this is, the point where one would trust an artificial car dealer, or a virtual medical doctor, lies elsewhere than in the technically detailed appearance of the artificial agent. Instead, the believability of such an agent rests in the emotive-cognitive contextualization of the interaction, in order words, taking into account the implicit and explicit psychological dependencies related to the context. To tackle this issue, the research on narratives and storytelling is proposed as a framework to facilitate the endeavor of modeling contextual aspects into the adaptive interaction dynamics of an artificial character designed to elicit trustfulness and confidence in the human agent. A selection of case studies is discussed in the light of cognitive neuroscientific findings on artificial agents and narrative contextualization, particularly from the point of view of societal and medical applications.
The conference website: http://nordichi2020.org/