DYNAMIC APPROACHES TO MACHINE LEARNING ETHICS IN SOFTWARE ENGINEERING: A COMPARATIVE EXPLORATION OF PAKISTAN AND SAUDI ARABIA

Abstract

The increasing integration of machine learning (ML) into software engineering has transformed software development while intensifying ethical concerns related to fairness, transparency, accountability, and governance. Grounded in Socio-Technical Systems (STS) Theory and the Responsible AI Framework, this study explores how software professionals operationalize ethical principles throughout the Software Development Life Cycle (SDLC) in Pakistan and Saudi Arabia. A qualitative comparative design was employed using semi-structured interviews with 16 software engineers, AI engineers, machine learning specialists, software architects, and technology managers from organizations actively developing or deploying ML-enabled software. The interview data were analyzed using Braun and Clarke's reflexive thematic analysis. Five interrelated themes emerged: ethical awareness in machine learning development, fairness and bias mitigation, transparency and explainability, organizational governance and regulatory compliance, and cross-cultural perspectives on responsible AI implementation. The findings reveal that ethical machine learning is fundamentally a socio-technical process shaped by interactions among technical practices, organizational governance, professional expertise, and institutional environments. The study extends the application of STS Theory to software engineering, operationalizes the Responsible AI Framework within the SDLC, and proposes a Socio-Technical Ethical Governance Framework to support trustworthy machine learning development across emerging digital economies.