Kessia Nepomuceno | Searcharxiv

arXiv2025

A Systematic Mapping on Software Fairness: Focus, Trends and Industrial Context

Context: Fairness in systems has emerged as a critical concern in software engineering, garnering increasing attention as the field has advanced in recent years. While several guidelines have been proposed to address fairness, achieving a comprehensive understanding of research solutions for ensuring fairness in software systems remains challenging. Objectives: This paper presents a systematic literature mapping to explore and categorize current advancements in fairness solutions within software engineering, focusing on three key dimensions: research trends, research focus, and viability in industrial contexts. Methods: We develop a classification framework to organize research on software fairness from a fresh perspective, applying it to 95 selected studies and analyzing their potential for industrial adoption. Results: Our findings reveal that software fairness research is expanding, yet it remains heavily focused on methods and algorithms. It primarily focuses on post-processing and group fairness, with less emphasis on early-stage interventions, individual fairness metrics, and understanding bias root causes. Additionally fairness research remains largely academic, with limited industry collaboration and low to medium Technology Readiness Level (TRL), indicating that industrial transferability remains distant. Conclusion: Our results underscore the need to incorporate fairness considerations across all stages of the software development life-cycle and to foster greater collaboration between academia and industry. This analysis provides a comprehensive overview of the field, offering a foundation to guide future research and practical applications of fairness in software systems.

cs.SE↗

arXiv2025

The AI Fairness Myth: A Position Paper on Context-Aware Bias

Defining fairness in AI remains a persistent challenge, largely due to its deeply context-dependent nature and the lack of a universal definition. While numerous mathematical formulations of fairness exist, they sometimes conflict with one another and diverge from social, economic, and legal understandings of justice. Traditional quantitative definitions primarily focus on statistical comparisons, but they often fail to simultaneously satisfy multiple fairness constraints. Drawing on philosophical theories (Rawls' Difference Principle and Dworkin's theory of equality) and empirical evidence supporting affirmative action, we argue that fairness sometimes necessitates deliberate, context-aware preferential treatment of historically marginalized groups. Rather than viewing bias solely as a flaw to eliminate, we propose a framework that embraces corrective, intentional biases to promote genuine equality of opportunity. Our approach involves identifying unfairness, recognizing protected groups/individuals, applying corrective strategies, measuring impact, and iterating improvements. By bridging mathematical precision with ethical and contextual considerations, we advocate for an AI fairness paradigm that goes beyond neutrality to actively advance social justice.

cs.CY↗