Software Fairness Debt: Building a Research Agenda for Addressing Bias in AI Systems
Bibliographic record
Abstract
Ensuring fairness in software systems has become a critical concern in software engineering. Motivated by this challenge, this article explores the multifaceted nature of bias in software systems, providing a comprehensive understanding of its origins, manifestations, and impacts. Through a scoping study, we identified the primary causes of fairness deficiencies in software development and highlighted their adverse effects on individuals and communities, including instances of discrimination and the perpetuation of inequalities. Our investigation culminated in the introduction of the concept of software fairness debt. In addition to defining fairness debt, we propose a socio-technical roadmap that addresses broader aspects of fairness in AI-driven systems. This roadmap is structured around six goals: bridging the gap between research and real-world applications, developing a framework for fairness debt, equipping practitioners with tools and knowledge, improving bias mitigation, integrating fairness tools into industry practice, and enhancing explainability and transparency in AI systems. This roadmap provides a holistic approach to managing biases in software systems through software fairness debt, offering actionable steps for both research and practice. By guiding researchers and practitioners, our roadmap aims to foster the development of more equitable and socially responsible software systems, ensuring fairness is embedded throughout the software lifecycle.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.126 | 0.205 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.008 | 0.007 |
| Science and technology studies | 0.009 | 0.040 |
| Scholarly communication | 0.021 | 0.051 |
| Open science | 0.005 | 0.020 |
| Research integrity | 0.011 | 0.012 |
| Insufficient payload (model declined to judge) | 0.006 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".