Paper Number
ECIS2026-2518
Paper Type
SP
Abstract
Anthropomorphic AI agents (AAAs) increasingly display social identity cues such as gender, ethnicity, and disability status, raising questions about their influence on intergroup relations. While existing research emphasizes organizational and interpersonal outcomes, this study explores whether intergroup contact with AAAs affects attitudes toward the social groups they represent. Drawing on Computers are Social Actors framework and Intergroup Contact Theory; we conducted a randomized experiment with 229 UK-based Instagram users who interacted with AI-generated profiles either displaying or omitting transgender identity cues. Results indicate that AAAs with social identity cues triggered significantly higher anthropomorphism, reduced intergroup anxiety, and increased empathy towards human transgender individuals. These preliminary findings suggest AAAs function in dual capacity, (1) as entities subject to social expectations and (2) as representatives shaping perceptions of social groups, with implications for responsible AI design in public-facing contexts.
Recommended Citation
Voronin, Georg, "Computers Are Soci(Et)Al Actors: Extending Intergroup Contact Theory To Anthropomorphic AI Agents" (2026). ECIS 2026 Proceedings. 10.
https://aisel.aisnet.org/ecis2026/ai_anthro/ai_anthro/10
Computers Are Soci(Et)Al Actors: Extending Intergroup Contact Theory To Anthropomorphic AI Agents
Anthropomorphic AI agents (AAAs) increasingly display social identity cues such as gender, ethnicity, and disability status, raising questions about their influence on intergroup relations. While existing research emphasizes organizational and interpersonal outcomes, this study explores whether intergroup contact with AAAs affects attitudes toward the social groups they represent. Drawing on Computers are Social Actors framework and Intergroup Contact Theory; we conducted a randomized experiment with 229 UK-based Instagram users who interacted with AI-generated profiles either displaying or omitting transgender identity cues. Results indicate that AAAs with social identity cues triggered significantly higher anthropomorphism, reduced intergroup anxiety, and increased empathy towards human transgender individuals. These preliminary findings suggest AAAs function in dual capacity, (1) as entities subject to social expectations and (2) as representatives shaping perceptions of social groups, with implications for responsible AI design in public-facing contexts.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.