How do varied populations interact with embodied conversational agents? Findings from inner-city adolescents and prisoners
Two studies were conducted to identify individual characteristics that predict behavioral responses to violence prevention interventions. These studies used embodied conversational agents (ECAs) to create hypothetical social situations (called virtual vignettes) to assess interpersonal competency skills. One study was of male inner-city African–American adolescents, and the second was of male prisoners in a state correctional system. In pre- and post-intervention sessions, participants interacted with an ECA that tried to entice them into making risky decisions. The virtual vignette sessions tested participants’ negotiation and conflict resolution skills. Results showed differing tendencies for participants to be engaged by the virtual vignettes. The vignettes were sufficiently realistic to elicit differences in behavior among the adolescents, but generally not for the prisoners. Prior acceptance, accessibility, and usability data suggest that most users readily accept ECAs as valid conversational partners. The evidence presented here suggests that the technology – or the setting in which the technology is used – is not by itself sufficient to actively engage users. The usefulness of virtual vignettes to adequately predict future behavior may be at least partially influenced by participant characteristics.
Hubal, R., Fishbein Launse, D., Sheppard, MS., Paschall, M., Eldreth, D., & Hyde, CT. (2008). How do varied populations interact with embodied conversational agents? Findings from inner-city adolescents and prisoners. Computers in Human Behavior, 24(3), 1104-1138. https://doi.org/10.1016/j.chb.2007.03.010