Navigating Ethical Dissonance in Algorithmic Decision-Making: Lived Experiences of Data Scientists in the Financial Sector

Abstract

As algorithmic decision-making becomes central to data science practice, questions of ethics and bias have emerged as critical concerns across sectors. Within the financial industry, practitioners face increasing responsibility for managing the consequences of automated systems, yet their lived experiences remain poorly understood. While much of the existing literature addresses technical solutions for fairness and transparency, it offers limited insight into how data scientists interpret and respond to ethical dilemmas in practice. This study investigates: How do practitioners experience and navigate ethical tensions related to algorithmic bias in real-world settings? Using a phenomenological approach, this research explores the experiences of ten data scientists in financial institutions.  Data were collected through in-depth, semi-structured interviews and analyzed thematically using Interpretative Phenomenological Analysis (IPA). The findings reveal that participants often experience ethical dissonance, driven by conflicts between institutional performance metrics and personal ethical values. Practitioners develop informal coping strategies such as silent resistance, moral rationalization, and subtle intervention to reconcile these tensions. These results suggest that ethical engagement is shaped not only by policy and design, but also by individual interpretation and organizational culture. The study highlights the need for structural support systems that empower ethical reflection within data-driven environments. These insights contribute to a more human-centered understanding of ethics in algorithmic systems and offer new directions for future interdisciplinary research.