Fairness and Machine Learning

Limitations and Opportunities

by Barocas, Hardt, Narayanan

| ISBN: 9780262376532 | Copyright 2023

Click here to preview

Instructor Requests

Digital Exam/Desk Copy Print Desk Copy Ancillaries

An introduction to the intellectual foundations and practical utility of the recent work on fairness and machine learning.

Fairness and Machine Learning introduces advanced undergraduate and graduate students to the intellectual foundations of this recently emergent field, drawing on a diverse range of disciplinary perspectives to identify the opportunities and hazards of automated decision-making. It surveys the risks in many applications of machine learning and provides a review of an emerging set of proposed solutions, showing how even well-intentioned applications may give rise to objectionable results. It covers the statistical and causal measures used to evaluate the fairness of machine learning models as well as the procedural and substantive aspects of decision-making that are core to debates about fairness, including a review of legal and philosophical perspectives on discrimination. This incisive textbook prepares students of machine learning to do quantitative work on fairness while reflecting critically on its foundations and its practical utility.

• Introduces the technical and normative foundations of fairness in automated decision-making
• Covers the formal and computational methods for characterizing and addressing problems
• Provides a critical assessment of their intellectual foundations and practical utility
• Features rich pedagogy and extensive instructor resources

Expand/Collapse All
Cover (pg. Cover)
Contents (pg. v)
Preface (pg. ix)
Online Materials (pg. xiv)
Acknowledgments (pg. xv)
1. Introduction (pg. 1)
Demographic Disparities (pg. 3)
The Machine Learning Loop (pg. 5)
The State of Society (pg. 6)
The Trouble with Measurement (pg. 8)
From Data to Models (pg. 11)
The Pitfalls of Action (pg. 13)
Feedback and Feedback Loops (pg. 14)
Getting Concrete with a Toy Example (pg. 17)
Justice beyond Fair Decision Making (pg. 20)
Our Outlook: Limitations and Opportunities (pg. 22)
Bibliographic Notes and Further Reading (pg. 23)
2. When Is Automated Decision Making Legitimate? (pg. 25)
Machine Learning Is Not a Replacement for Human Decision Making (pg. 26)
Bureaucracy as a Bulwark against Arbitrary Decision Making (pg. 28)
Three Forms of Automation (pg. 30)
Mismatch between Target and Goal (pg. 36)
Failing to Consider Relevant Information (pg. 38)
The Limits of Induction (pg. 41)
A Right to Accurate Predictions? (pg. 43)
Agency, Recourse, and Culpability (pg. 44)
Concluding Thoughts (pg. 47)
3. Classification (pg. 49)
Modeling Populations as Probability Distributions (pg. 50)
Formalizing Classification (pg. 51)
Supervised Learning (pg. 56)
Groups in the Population (pg. 58)
Statistical Nondiscrimination Criteria (pg. 60)
Independence (pg. 61)
Separation (pg. 63)
Sufficiency (pg. 67)
How to Satisfy a Nondiscrimination Criterion (pg. 70)
Relationships between Criteria (pg. 71)
Case Study: Credit Scoring (pg. 74)
Inherent Limitations of Observational Criteria (pg. 79)
Chapter Notes (pg. 80)
4. Relative Notions of Fairness (pg. 83)
Systematic Relative Disadvantage (pg. 83)
Six Accounts of the Wrongfulness of Discrimination (pg. 85)
Intentionality and Indirect Discrimination (pg. 87)
Equality of Opportunity (pg. 88)
Tensions between the Different Views (pg. 93)
Merit and Desert (pg. 95)
The Cost of Fairness (pg. 98)
Connecting Statistical and Moral Notions of Fairness (pg. 100)
The Normative Underpinnings of Error Rate Parity (pg. 105)
Alternatives for Realizing the Middle View of Equality of Opportunity (pg. 109)
Summary (pg. 110)
5. Causality (pg. 113)
The Limitations of Observation (pg. 114)
Causal Models (pg. 117)
Causal Graphs (pg. 120)
Interventions and Causal Effects (pg. 123)
Confounding (pg. 124)
Graphical Discrimination Analysis (pg. 127)
Counterfactuals (pg. 132)
Counterfactual Discrimination Analysis (pg. 138)
Validity of Causal Modeling (pg. 143)
Chapter Notes (pg. 149)
6. Understanding United States Antidiscrimination Law (pg. 151)
History and Overview of US Antidiscrimination Law (pg. 152)
A Few Basics of the American Legal System (pg. 157)
How the Law Conceives of Discrimination (pg. 163)
Limits of the Law in Curbing Discrimination (pg. 167)
Regulating Machine Learning (pg. 172)
Concluding Thoughts (pg. 182)
7. Testing Discrimination in Practice (pg. 185)
Part 1: Traditional Tests for Discrimination (pg. 186)
Audit Studies (pg. 186)
Testing the Impact of Blinding (pg. 190)
Revealing Extraneous Factors in Decisions (pg. 192)
Testing the Impact of Decisions and Interventions (pg. 193)
Purely Observational Tests (pg. 194)
Taste-Based and Statistical Discrimination (pg. 198)
Studies of Decision-Making Processes and Organizations (pg. 200)
Part 2: Testing Discrimination in Algorithmic Systems (pg. 202)
Fairness Considerations in Applications of Natural Language Processing (pg. 203)
Demographic Disparities and Questionable Applications of Computer Vision (pg. 204)
Search and Recommendation Systems: Three Types of Harms (pg. 206)
Understanding Unfairness in Ad Targeting (pg. 208)
Fairness Considerations in the Design of Online Marketplaces (pg. 210)
Mechanisms of Discrimination (pg. 212)
Fairness Criteria in Algorithmic Audits (pg. 214)
Information Flow, Fairness, Privacy (pg. 215)
Comparison of Research Methods (pg. 217)
Looking Ahead (pg. 219)
Chapter Notes (pg. 219)
8. A Broader View of Discrimination (pg. 221)
Case Study: The Gender Earnings Gap on Uber (pg. 221)
Three Levels of Discrimination (pg. 225)
Machine Learning and Structural Discrimination (pg. 229)
Structural Interventions for Fair Machine Learning (pg. 234)
Organizational Interventions for Fairer Decision Making (pg. 238)
Concluding Thoughts (pg. 245)
Chapter Notes (pg. 247)
Appendix: A Deeper Look at Structural Factors (pg. 248)
9. Datasets (pg. 251)
A Tour of Datasets in Different Domains (pg. 252)
Roles Datasets Play (pg. 260)
Harms Associated with Data (pg. 271)
Beyond Datasets (pg. 274)
Summary (pg. 282)
Chapter Notes (pg. 282)
References (pg. 285)
Index (pg. 311)

Solon Barocas

Solon Barocas is a Principal Researcher in the New York City lab of Microsoft Research, where he is a member of the Fairness, Accountability, Transparency, and Ethics in AI (FATE) research group. He is Adjunct Assistant Professor in the Department of Information Science at Cornell University and Faculty Associate at the Berkman Klein Center for Internet & Society at Harvard University.

Moritz Hardt

Moritz Hardt is Director of Social Foundations of Computation at the Max Planck Institute for Intelligent Systems and coauthor of Patterns, Predictions, and Actions: Foundations of Machine Learning.

Arvind Narayanan

Arvind Narayanan is Professor of Computer Science at Princeton University and director of the Center for Information Technology Policy. His work was among the first to show how machine learning reflects cultural stereotypes, and he led the Princeton Web Transparency and Accountability Project to uncover how companies collect and use our personal information.

Instructors Only
You must have an instructor account and submit a request to access instructor materials for this book.
Go paperless today! Available online anytime, nothing to download or install.


  • Bookmarking
  • Note taking
  • Highlighting