Manipulation of non-verbal interaction style and demographic embodiment to increase anthropomorphic computer character credibility

How do varied populations interact with embodied conversational agents? Findings from inner-city adolescents and prisoners

“…These ECAs are not photo-realistic, but nor are they caricatures. The movements of the ECA were based on motion captured from an adult white male (see Zimmer, et al, 2003), a mismatch with age and ethnicity for the adolescent vignettes that may have influenced results (see Cowell & Stanney, 2005;Kim & Baylor, 2006;Louwerse, et al, 2005). The level of visual realism, in combination with convincing linguistic and emotional models, appears sufficient to engage these participants in short interactions (see Isbister & Doyle, 2002), but is not yet sufficient for valid assessment.…”

Communicating emotions in expressive avatars

“…Giving avatars expressive abilities has been considered beneficial as it potentially leverages the observer's real life experience with social interaction (Cowell and Stanney, 2005). From the real world, emotions of others influence us in our decisions and our own emotional state (Picard, 1997).…”

Gendering the Machine: Preferred Virtual Assistant Gender and Realism in Self-Service

“…An interface agent is, more generally, a computer-generated character designed to interact with users by simulating human appearance and behaviors through artificial intelligence. Dependent on their function, they are referred to as Virtual Customer Service Agent [ 15], Product Recommendation Agent [ 6], Embodied Conversational Agent [ 11], [ 17], Pedagogical Agent [ 18], [19][20][21], [31], or as a human-like or embodied agent/character [12][13], [23][24][25][26][27].…”

“…Research on user preferences for agent gender presents contradictory findings and trends, from a preference for same-sex virtual agents, to a preference for opposite sex virtual agents, and no gender preference at all [ 6], [ 17], [ 22], [ 28]. These differences may be due to user characteristics, but also context [28][29].…”

“…Lee [ 23] for example, found that participants were more open to suggestions made by a virtual character whose implied gender matched the gendered topic. In a gender-neutral task (sorting photographs) Cowell and Stanney [ 17] reported a trend in females towards a preference for a virtual character of the opposite sex. Conversely, Kim and Wei [ 22] found a preference for gender-matched virtual agents, even though the context -a computerbased algebra learning environment -may have been perceived as requiring male expertise.…”