Exploring Gender Bias Against Women in STEM: A Qualitative Case Study Analysis

Abstract

The increasing amount of artificial intelligence (AI) employed within STEM ecosystem are raising many concerns for businesses using algorithm-based decision-making systems that are shown to embed gender-related biases. The purpose of this investigation was to analyze whether AI methods of recruiting, evaluating, and promoting individuals may re-introduce or increase structural disparities among genders within STEM fields. This research is grounded in socio-technical systems theory; thus, the researcher defines algorithmic bias as being structural in nature and caused by historically biased datasets, proxy variables, and optimization models, placing predictive accuracy over distributional equality. The methodology included qualitative, integrative research techniques that included a structured review of literature, as well as an analytic review of case studies of biased AI implementation in industry. A comparative analysis of fairness-aware design methodologies, debiasing strategies, and governance modalities was used to help assess their effectiveness in reducing bias. While the results indicate that technical fairness interventions produce statistically significant reductions in measurement biases such as statistical parity and corresponding opportunities, the researchers conclude that they are inadequate without implementing corresponding institutional and governance reforms. The analysis determined that the long-term reduction of gender biases existing in AI systems requires a holistic strategy that includes technical redesigning, inclusive data governance practices, and accountability from organizations. Finally, the researcher presents a theoretically integrated conceptual model for creating an equity-based AI design and implementation framework within STEM ecosystem.