Computer-Generated Anthropomorphic Agents: Avatars and Agents
Computer-generated anthropomorphic characters are computer agents or avatars that have been provided human traits or qualities, such as computer generated faces (Gong & Nass, 2007). Computer-generated anthropomorphic characters are frequently imbued with the ability to speak through either computer-generated speech or prerecorded natural speech (Gong & Nass, 2007). These virtual humans are a new and increasingly prominent type of digital communicator, and are being used in many capacities. For instance, virtual humans have been utilized as computer interface agents (Gong & Nass, 2007; Lee & Nass, 2002), as avatars or virtual representations of self in video games (Eastin, 2006), in virtual environments (Bailenson, Beall, Loomis, Blascovich, & Turk, 2005), and for the purposes of computer-mediated communication (Lee, 2004). Computer-generated anthropomorphic characters are usually categorized as being one of two types, either agents, which are computer-controlled characters, or avatars, which are user-controlled (see Nowak, 2004; Eastin, Appiah, & Cichirillo, in press; Gong & Nass, 2007). Computer-generated characters are being used as digital communicators on websites as well as in computer games and applications with increasing frequency (Gong & Nass, 2007). A major overlooked effect of the proliferation of virtual human characters is that individuals who interact with them are increasingly being exposed to racial entities that are not real humans or real-human representations (Gong, Appiah & Elias, 2007).
Computer-generated anthropomorphic characters are often imbued with basic social identities that frequently include race (Baylor & Kim, 2003; Gong, Appiah, & Elias, 2007). As a result, an important theoretical consideration establishes how ethnicity operates in virtual contexts between source and viewer as opposed to the more traditional communication contexts involving real humans or humans interacting with mediated real-human entities and representations. Simply, do virtual human representations stimulate traditional racial responses of ingroup and outgroup members as often found with real-human entities (Gong, Appiah, & Elias, 2007)? Answers to this and similar questions hold significant implications for individuals of all ethnicities, particularly Blacks.
Ostensibly, Blacks have made significant inroads with respect to their representations in advertisements in traditional media, most notably television and print (Wilkes & Valencia, 1989; Zinkhan, Qualls & Biswas, 1990). Subsequent studies have revealed, however, that Blacks in advertisements are still being utilized in restricted roles (Entman & Rojecki, 2000), for short time periods (Greenberg & Brand, 1994), and in racially-integrated groups (Wilkes & Valencia, 1989). Additionally, Blacks who do not have expert qualifications or who are not recognizable celebrities have been disproportionately used to advertise low-end, inexpensive products (Bang & Reece, 2003). Ironically, despite a common perception of Blacks living in the U.S. being destitute, they currently have more disposable income than they did in the past, and outspend non-Black consumers in several high margin categories (Cooper, 2004). Continued failure to acknowledge the importance of this demographic could be to the detriment of many organizations. Additionally, a lack of understanding of how race affects consumer attitudes and communication of individuals of all ethnicities in today’s current information society could also prove deleterious.
In today’s online environment consumers are ever-exposed to racial entities that are computer-generated in a multitude of digital communication environments. There exists, however, limited insight as to the impact of virtual race and ethnicity, particularly in e-commerce settings. Additionally, research has yet to examine ethnic ambiguity as a key predictor of consumer attitudes toward products featured on commercial websites. What the research has shown, however, is that people automatically and unconsciously follow the same social rules when they interact with computers as when they interact with humans (Nass & Moon, 2000; Reeves & Nass, 1996). Nass and colleagues, in the media equation discourse community, have established that social rules such as displays of ingroup favoritism, politeness, reciprocity, gender stereotypes, and personality are applied in much the same way when humans interact with computers as when they interact with other humans (Gong, Appiah, & Elias, 2007). Moreover, with respect to race and ethnicity, Nass and colleagues (2000) have found that agents that are the same ethnicity as participants tend to elicit more positive ratings of social attraction and trustworthiness, as well as greater conformity than agents of a different ethnicity. A noticeable trend that has been identified by the scant research is an ingroup preference for same-race computer agents. This seems to follow the same pattern of results as that found among the growing body of studies examining real humans (for review see Gong, Appiah, & Elias, 2007). For instance, Baylor and Kim (2003) found that among computer-generated Black and White virtual pedagogical agents, users rated the same-race agents as more engaging and affable. Baylor, Shen, and Huang (2003) also found same-race preference in direct choice-making among Black and White virtual pedagogical agents. However, since the latest cultural trend among marketers is to use ethnically ambiguous models in advertising (Arlidge, 2004), the question arises, how might this process work for agents that are neither White nor Black but racially or ethnically ambiguous?