No. 249 of 500 No. 56 in AI Ethics & Society

The Ethical Algorithm

The Science of Socially Aware Algorithm Design

By Michael Kearns and Aaron Roth

Published
November 1, 2019
Print length
230 pages
Reading level
Intermediate
ISBN-13
978-0190948207
Publisher
Oxford University Press
Written for
Policymakers, product leaders, data scientists and general readers wanting a rigorous introduction to algorithmic ethics.

Synopsis

Michael Kearns and Aaron Roth, computer scientists at the University of Pennsylvania, argue that many harms caused by data-driven systems are best addressed inside the algorithms themselves, not only through laws and oversight. Writing for nonspecialists, they explain how machine learning models trained on historical data can leak private information or produce unequal outcomes for different groups, and how researchers now define privacy and fairness mathematically so that code can enforce them. Differential privacy, which limits what any analysis can reveal about a single person, serves as the book's model of a rigorous, deployable standard.

Subsequent chapters examine competing definitions of fairness and the cost each imposes on accuracy, including the problem the authors call fairness gerrymandering, in which a model treats broad groups evenly while concentrating harm on a subgroup. The book then turns to game-theoretic settings such as navigation apps and dating markets, to the echo chambers created by recommendation systems, and to false discoveries that arise when researchers reuse the same data repeatedly. A closing chapter considers interpretability, morality and the limits of algorithmic fixes. Readers gain a precise vocabulary for evaluating claims about ethical AI and a realistic view of what technical design can and cannot achieve, without needing mathematics beyond basic intuition.

Scorecard

Editorial score

How it scored

Editorial score: 74.8 / 1005 weighted criteria

  • Influence30%

    21.0/30
  • Authority25%

    22.5/25
  • Practical value20%

    14.0/20
  • Currency15%

    9.8/15
  • Readability10%

    7.5/10
  • Editorial score

    74.8/100

How the rank was computed

Rank No. 249 of 500 comes from a composite of 55.0: observable signals 21.3 of 55 points plus editorial review 33.7 of 45 points.

All 11 inputs
InputPoints
Observable signals21.3/55
Reach (Amazon ratings volume)8.4/16.5
Consensus across best-of lists5.5/13.8
Awards and bestseller evidence0.0/8.3
Reader rating, sample-adjusted6.5/8.3
Amazon sales rank0.9/5.5
Recency0.0/2.8
Editorial review33.7/45
Author authority10.1/11.3
Influence9.5/13.5
Practical value6.3/9.0
Readability3.4/4.5
Currency4.4/6.8

Each criterion is scored from 1 to 10 and weighted, and the points add up to the editorial score. The rank combines observable signals (55 percent) with this score (45 percent). Read the methodology.

Verdict

Two leading computer scientists show how fairness and privacy can be engineered into algorithms, with candid accounting of the trade-offs.

Strengths

  • Rigorous ideas explained without heavy mathematics
  • Authors are leading researchers in the field
  • Clear treatment of fairness-accuracy trade-offs

Keep in mind

Written before the generative AI boom, it does not address large language models, and some readers find its confidence in technical fixes too optimistic.

Key takeaways

  1. Anonymizing data rarely protects privacy; differential privacy offers a provable guarantee by adding carefully calibrated noise to computations.
  2. Fairness and accuracy usually trade off along a frontier, so stakeholders must choose an acceptable balance rather than expect a cost-free fix.
  3. Removing sensitive attributes such as race does not prevent bias, because proxies such as zip codes can reproduce the same outcomes.
  4. Repeatedly testing hypotheses on the same dataset produces false discoveries, a problem that tools from differential privacy can help control.

Notable facts

  • The Brookings Institution hosted a January 2020 public discussion of the book with both authors, moderated by Nicol Turner-Lee. Source (opens in a new tab)
  • Kearns's faculty page lists reviews of the book in Nature and strategy+business and a C-SPAN Book TV segment on it. Source (opens in a new tab)

About the author

Michael Kearns and Aaron Roth

Michael Kearns is professor and National Center Chair in Computer and Information Science at the University of Pennsylvania, founding director of its Warren Center, an Amazon Scholar and a member of the National Academy of Sciences. Co-author Aaron Roth is Penn's Henry Salvatori Professor of Computer and Cognitive Science and co-wrote The Algorithmic Foundations of Differential Privacy with Cynthia Dwork.

Questions

What is The Ethical Algorithm about?
The Ethical Algorithm explains how computer scientists are building social values such as privacy and fairness directly into algorithms. Kearns and Roth cover differential privacy, competing definitions of fairness and their trade-offs with accuracy, strategic behavior around algorithms, recommendation echo chambers and false discoveries from overused data, then discuss what technical design cannot solve, such as moral judgment and interpretability.
Is The Ethical Algorithm worth reading in 2026?
Yes, as a foundation. Its explanations of differential privacy, fairness trade-offs and proxy discrimination still underpin current debates about AI regulation and audits, and differential privacy was adopted for the 2020 U.S. Census. It predates ChatGPT and does not discuss large language models, so it works best paired with a more recent book on generative AI.
Do I need a technical background to read The Ethical Algorithm?
No. The authors wrote it for general readers, using examples and intuition rather than formal proofs. Comfort with basic statistics helps in the chapters on fairness trade-offs and adaptive data analysis, and some readers without a quantitative background find those sections demanding. Engineers and data scientists will find it a fast, well-grounded overview before they turn to the research literature.

Sources

As an Amazon Associate, Best AI Books earns from qualifying purchases. Best AI Books is sponsored by Iternal.ai, whose founder and CEO, John Byron Hanby IV, wrote The AI Strategy Blueprint, our #1 ranked book. How we rank books.