Ethnic Bias in AI-Driven Virtual Recruitment
Author
Nielsen, Sofie Julsgaard
Term
4. term
Education
Publication year
2026
Abstract
This thesis examines whether virtual reality (VR) can be used as a research tool to explore the relationship between implicit ethnic bias and hiring decisions. Participants conducted unstructured job interviews with four virtual candidates in a VR simulation, then indicated their hiring choices and completed questionnaires on presence and co‑presence. They also took an Implicit Association Test (IAT) to assess implicit preferences for white‑presenting versus Arab‑presenting individuals. With a small sample (N=15), the study found no statistically significant relationship between IAT scores and hiring choices. However, descriptive patterns aligned with implicit bias theory: participants with an implicit preference for white‑presenting individuals more often selected such candidates, while those with neutral or Arab preferences showed more varied selections. Participants reported a generally moderate sense of immersion and social engagement in the VR environment. Despite the lack of statistical significance—likely due to limited power—the findings suggest that VR can serve as a controlled yet immersive setting for studying sensitive social processes in recruitment.
Denne afhandling undersøger, om virtuel reality (VR) kan bruges som forskningsværktøj til at belyse sammenhængen mellem implicit etnisk bias og ansættelsesbeslutninger. Deltagere gennemførte ustrukturerede jobsamtaler med fire virtuelle kandidater i en VR-simulation, hvorefter de angav, hvem de ville ansætte, og udfyldte spørgeskemaer om nærvær og socialt samvær. Derudover tog de en Implicit Association Test (IAT) for at måle implicitte præferencer for hvidt‑fremtrædende versus arabisk‑fremtrædende personer. Med et lille sample (N=15) fandt studiet ingen statistisk signifikant sammenhæng mellem IAT‑scorer og valg af kandidater. Deskriptive mønstre pegede dog i retning af implicit bias‑teori: deltagere med implicit præference for hvidt‑fremtrædende personer valgte oftere sådanne kandidater, mens deltagere med neutral eller arabisk præference fordelte valgene mere bredt. Deltagerne rapporterede generelt moderat følelsesmæssigt nærvær og med‑nærvær i VR‑miljøet. På trods af manglende signifikans, sandsynligvis pga. begrænset statistisk styrke, indikerer resultaterne, at VR kan fungere som et kontrolleret, men engagerende miljø til at studere følsomme sociale processer i rekruttering.
[This apstract has been generated with the help of AI directly from the project full text]