Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The Pursuit of Fairness in Artificial Intelligence Models: A Survey

Tahsin Alamgir Kheya, Mohamed Reda Bouadjenek, Sunil Aryal

TL;DR

This survey addresses the challenge of fairness in AI by outlining a comprehensive taxonomy of fairness and bias definitions, documenting data-, human-, and model-driven sources of unfairness, and surveying practical mitigation strategies. It details group, individual, and causal notions of fairness, including intersectional and sufficiency metrics, and discusses real-world case studies across criminal justice, hiring, finance, healthcare, and education. The authors emphasize a multi-pronged approach—pre-, in-, and post-processing; data debiasing; adversarial and causal methods; regularization; and fairness-aware optimization—while highlighting trade-offs between fairness and accuracy. They also explore ethical guidelines, user experience implications, and governance considerations, arguing for ongoing, responsible development of fair AI with transparency and accountability.

Abstract

Artificial Intelligence (AI) models are now being utilized in all facets of our lives such as healthcare, education and employment. Since they are used in numerous sensitive environments and make decisions that can be life altering, potential biased outcomes are a pressing matter. Developers should ensure that such models don't manifest any unexpected discriminatory practices like partiality for certain genders, ethnicities or disabled people. With the ubiquitous dissemination of AI systems, researchers and practitioners are becoming more aware of unfair models and are bound to mitigate bias in them. Significant research has been conducted in addressing such issues to ensure models don't intentionally or unintentionally perpetuate bias. This survey offers a synopsis of the different ways researchers have promoted fairness in AI systems. We explore the different definitions of fairness existing in the current literature. We create a comprehensive taxonomy by categorizing different types of bias and investigate cases of biased AI in different application domains. A thorough study is conducted of the approaches and techniques employed by researchers to mitigate bias in AI models. Moreover, we also delve into the impact of biased models on user experience and the ethical considerations to contemplate when developing and deploying such models. We hope this survey helps researchers and practitioners understand the intricate details of fairness and bias in AI systems. By sharing this thorough survey, we aim to promote additional discourse in the domain of equitable and responsible AI.

The Pursuit of Fairness in Artificial Intelligence Models: A Survey

TL;DR

This survey addresses the challenge of fairness in AI by outlining a comprehensive taxonomy of fairness and bias definitions, documenting data-, human-, and model-driven sources of unfairness, and surveying practical mitigation strategies. It details group, individual, and causal notions of fairness, including intersectional and sufficiency metrics, and discusses real-world case studies across criminal justice, hiring, finance, healthcare, and education. The authors emphasize a multi-pronged approach—pre-, in-, and post-processing; data debiasing; adversarial and causal methods; regularization; and fairness-aware optimization—while highlighting trade-offs between fairness and accuracy. They also explore ethical guidelines, user experience implications, and governance considerations, arguing for ongoing, responsible development of fair AI with transparency and accountability.

Abstract

Artificial Intelligence (AI) models are now being utilized in all facets of our lives such as healthcare, education and employment. Since they are used in numerous sensitive environments and make decisions that can be life altering, potential biased outcomes are a pressing matter. Developers should ensure that such models don't manifest any unexpected discriminatory practices like partiality for certain genders, ethnicities or disabled people. With the ubiquitous dissemination of AI systems, researchers and practitioners are becoming more aware of unfair models and are bound to mitigate bias in them. Significant research has been conducted in addressing such issues to ensure models don't intentionally or unintentionally perpetuate bias. This survey offers a synopsis of the different ways researchers have promoted fairness in AI systems. We explore the different definitions of fairness existing in the current literature. We create a comprehensive taxonomy by categorizing different types of bias and investigate cases of biased AI in different application domains. A thorough study is conducted of the approaches and techniques employed by researchers to mitigate bias in AI models. Moreover, we also delve into the impact of biased models on user experience and the ethical considerations to contemplate when developing and deploying such models. We hope this survey helps researchers and practitioners understand the intricate details of fairness and bias in AI systems. By sharing this thorough survey, we aim to promote additional discourse in the domain of equitable and responsible AI.
Paper Structure (90 sections, 16 equations, 6 figures, 1 table)

This paper contains 90 sections, 16 equations, 6 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Survey
  3. Conceptualizing Fairness and Bias in ML
  4. Fairness in Machine Learning
  5. Group fairness
  6. Demographic Parity
  7. Conditional Statistical Parity
  8. Individual fairness
  9. Fairness through Unawareness
  10. Fairness through Awareness
  11. Separation Metrics
  12. Predictive Equality
  13. Equal Opportunity
  14. Balance for the Negative Class
  15. Balance for the Positive Class
  16. ...and 75 more sections

Figures (6)

  • Figure 1: Number of papers published in this topic over the years. Data acquisition process to plot this graph is provided in Section \ref{['psp']}.
  • Figure 2: Proposed taxonomy of fairness in the machine learning context
  • Figure 3: Graph that exhibits unresolved discrimination.
  • Figure 4: Proposed taxonomy of observed biases in the machine learning pipeline
  • Figure 5: The three crucial stages in the development of an ML model
  • ...and 1 more figures