Electronic Thesis/Dissertation
 

Learning to Mitigate Bias in AI Design Practices

Open Access Deposited

Balancing Code and Context

comparative studies across organizational contexts

This basic qualitative study sought to understand how AI designers learn to recognize bias in AI systems and develop, implement, and evaluate bias mitigation strategies. Framed by Informal Learning Theory and Socio-Technical Systems Theory, ten AI designers with experience in bias mitigation within their organizations took part in semi-structured, in-depth interviews that required them to reflect on the factors that triggered, facilitated, and constrained their learning about bias mitigation.For these participants, formal learning programs rarely proved sufficient. Instead, they learned primarily through informal processes. Personal encounters with biased technology, professional incidents that made bias concrete, and ethical responsibility that transformed bias mitigation from a technical requirement into a moral imperative all served as triggers that initiated learning. Organizations that acknowledged bias as a reality with executives who committed to Responsible AI, and diverse teams created environments where informal learning could flourish. However, barriers persisted

and cross-national studies exploring how cultural and regulatory contexts influence learning. Implications for practice underscored the importance of diverse team composition as an epistemological advantage rather than simply a fairness goal, and the need for experiential training approaches over passive formats. For educational institutions, industry partnerships could expose students to real-world bias mitigation challenges before entering the workforce. This study sought to address a gap in the literature at the intersection of AI bias mitigation and informal learning. As AI systems continue to influence critical decisions in healthcare, employment, financial services, and public safety, understanding how those who design these systems learn to recognize and address bias becomes increasingly urgent. This study offers one contribution to that understanding.

longitudinal studies documenting how bias-mitigation expertise develops over time

business demands competed with the sustained engagement required by informal learning methods. Several recommendations for future research include, but are not limited to

quantitative studies examining the prevalence of informal learning patterns across the AI industry

Author Language Keyword Date created Type of Work License
  • All rights reserved
Rights statement GW Unit Degree Advisor Committee Member(s) Persistent URL

Notice to Authors

If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.

Thumbnail Title Date Uploaded Visibility Actions
Preview of Rhoads_gwu_0075A_17787.pdf File 2026-06-24 Embargo