Beyond Machines and Algorithms: Investigating Bias, Ethnic Attribution, and Social Categorization in Human-Robot Interaction

Author

Jessica Barfield, University of Tennessee at KnoxvilleFollow

Orcid ID

https://orcid.org/0000-0002-1208-6118

Date of Award

5-2024

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Information Sciences

Major Professor

Suzie Allard

Committee Members

Jiangen He, Xiaopeng Xiao, Awa Zhu

Abstract

As robots interact with people in social contexts it is important to design robots that meet the information needs and values of a diverse group of users. However, the design of current robots may not reflect the values and diversity of the users interacting with them and thus may fall short in achieving diversity, equity, and inclusion in human interaction with robots. In my dissertation research I propose that an important factor in achieving diversity, equity, and inclusion for human-robot interaction is the user’s ethnicity in comparison to the perceived ethnicity of a robot when experienced in a social context. Guiding my research is Social Identity Theory which has been used to describe the processes by which people categorize other people into a social class but extended in my dissertation research to human-robot interaction. For this unexplored topic of research, a series of studies are run to investigate the following research questions: i) Do users assign an ethnic identity to robots, and if so, what factors trigger such decisions? ii) Does a match in ethnicity between user and robot lead to the experience of diversity, equity, and inclusion during the user’s interaction with a robot? And iii) Does Social Identity Theory provide a useful theoretical framework to describe human interaction with a robot presented with an ethnic identity? After a manipulation check was run on the independent variables, four factorial experiments were performed, and interviews were conducted to provide anecdotal evidence to guide future research studies on how individuals experience robot ethnicity. The results of the research indicated that users assign an ethnic identity to a robot based on cues to ethnicity, that a robot presented with cues signaling an American ethnicity was anthropomorphized more so than a robot presented with cues signaling a Chinese or Mexican ethnicity and that participants classified robots with a similar ethnicity as themselves as an in-group member resulting in a more favorable evaluation than for robots categorized as an out-group member.

Recommended Citation

Barfield, Jessica, "Beyond Machines and Algorithms: Investigating Bias, Ethnic Attribution, and Social Categorization in Human-Robot Interaction. " PhD diss., University of Tennessee, 2024.
https://trace.tennessee.edu/utk_graddiss/10090