AI Ethics Governance: 9 Key Steps to Go from Zero to Operational

AI Ethics Governance: The Essentials in One Article — Real Code, Diagrams and Concrete Steps, Extracts from a 35-Lesson Course.

AI Ethics Governance: 9 Key Steps to Go from Zero to Operational

Everyone can learn AI Ethics Governance — provided they follow the steps in the right order. We have condensed a complete 35-lesson course into a clear pathway, with the most useful code excerpts.

tl;dr
  • Introduction
  • Fundamental Principles
  • Algorithmic Biases
  • Privacy and GDPR
  • XAI Explainability
~$ cat ./parcours.md # AI Ethics Governance — 10 chapters
01
Introduction
→ Course presentation→ Why AI ethics now?+ 1 more lessons
02
Fundamental principles
→ Fairness (fairness)→ Transparency and explainability+ 1 more lessons
03
Algorithmic biases
→ Understanding biases→ Famous cases — COMPAS, Amazon, Apple Card+ 1 more lessons
04
Privacy and GDPR
→ The GDPR in 10 key articles→ Anonymization and pseudonymization+ 1 more lessons
05
Explainability XAI
→ Why explain a model?→ SHAP, LIME techniques and cousins+ 1 more lessons
06
Regulatory frameworks
→ The EU AI Act in detail→ US, Canada, Asia frameworks+ 1 more lessons
07
AI in sensitive domains
→ AI in healthcare→ AI in justice and police+ 1 more lessons
08
Corporate governance
→ Setting up an AI ethics committee→ Drafting a responsible AI charter+ 1 more lessons
🏁
Final project (+ 2 chapters along the way)
→ You leave with a concrete and demonstrable project

Why AI ethics now?

NOTEObjective — Understand why the topic of AI ethics has moved from the margins to the center of decisions in just a few years. Identify the three major forces that make this discipline indispensable and the three major alerts that made society aware of the risks.

Learning objectives

TIPAt the end of this module — You will be able to explain in two minutes why AI ethics has become a C-suite topic, cite three emblematic cases of misconduct, and identify the financial, legal, and reputational costs of poor algorithmic governance.

Accelerator #1: the generalization of LLMs

In November 2022, the launch of ChatGPT marked a before and after. In two months, the tool reached 100 million users, an absolute record for a consumer service. Suddenly, every employee, every student, every journalist could generate text, code, and analyses.

The consequences are multiple:

WARNINGThe technological leap — Before 2022, AI was mainly a topic for data scientists. After 2022, it became a topic for all employees. The risk became widespread, and therefore regulation had to follow.

Accelerator #2: the arrival of the EU AI Act

Adopted in March 2024, the Regulation (EU) 2024/1689 — nicknamed the “EU AI Act” — is the world’s first general AI law. It applies progressively between 2025 and 2027.

Overview

Timeline

Domino effect

The arrival of a binding standard changes everything: it is no longer about “doing one’s best,” but about documenting, auditing, and certifying.

Accelerator #3: societal and reputational pressure

The media, NGOs (Algorithm Watch, AI Now Institute, La Quadrature du Net), unions, and consumers scrutinize AI uses. A poorly managed incident becomes a scandal within hours on social media.

Recent examples:

Alert #1: COMPAS and the American justice system (2016)

The NGO ProPublica published an investigation in May 2016 on the COMPAS tool (Correctional Offender Management Profiling for Alternative Sanctions), used by American judges to estimate recidivism risk.

NOTEThe finding — ProPublica shows that COMPAS overestimates risk for African-American defendants (false positives twice as frequent as for white defendants). Publisher Northpointe disputes the findings, but the debate has gone global.

Alert #2: Amazon and sexist résumé screening (2018)

Reuters revealed that Amazon developed an internal résumé-screening tool between 2014 and 2017. The model, trained on 10 years of past applications (predominantly male in tech), systematically preferred men’s résumés.

The model penalized any mention containing the word “women’s” (women’s chess club, for example) and favored “masculine” verbs such as “executed” or “captured.” Amazon abandoned the project.

WARNINGKey lesson — A model is never neutral: it inherits the biases of the training data. The Amazon scandal raised awareness among thousands of companies about the risk of indirect discrimination by AI.

Alert #3: Tay and the fragility of chatbots (2016)

Microsoft launched Tay, a Twitter chatbot meant to learn “by chatting.” Within 24 hours, malicious users trained it to produce racist, Holocaust-denying, and sexist statements. Microsoft withdrew the tool.

The real cost of an AI failure

Fairness

NOTEObjective — Understand what is meant by “a fair AI,” distinguish the different mathematical definitions of fairness (demographic parity, equal opportunity, calibration), and recognize that there is no single “correct” definition: some are mathematically incompatible.

Learning objectives

TIPAt the end of this module — You will be able to choose the fairness definition suited to your context, explain why the same algorithm can be called “fair” or “unfair” depending on the chosen criterion, and engage in a structured discussion with stakeholders (business, legal, HR).

What is a “fair” AI?

The word fairness covers a complex reality: ensuring that a system does not unduly disadvantage certain people because of a protected attribute (sex, origin, age, disability, religion…).

Under French law, these attributes are listed in Article 225-1 of the Penal Code and the law of 27 May 2008. Any direct or indirect discrimination based on these criteria is prohibited, even when it originates from an algorithm.

WARNINGIndirect discrimination — An apparently neutral measure can produce a discriminatory effect: for example, requiring a degree from a top school statistically disadvantages candidates from modest backgrounds. An algorithm that learns to prefer this type of profile reproduces discrimination, even without an explicit variable.

The 4 major fairness measures

1. Demographic Parity

The positive prediction rate must be identical across protected groups.

Example: 30 % of men and 30 % of women are approved for a loan.

2. Equal Opportunity

Among people who deserve a positive outcome, the acceptance rate must be equal across groups.

Example: among qualified candidates, 80 % of men and 80 % of women are selected.

3. Equalized Odds

False positive and false negative rates must be identical across groups.

4. Calibration

When the model predicts a 70 % risk, 70 % of actual cases must be observed, identically for all groups.

Kleinberg’s impossibility (2017)

NOTETheorem — As soon as base rates differ between groups (for example: 30 % default rate among young borrowers versus 10 % among seniors), it is mathematically impossible to satisfy both calibration and equalized odds simultaneously. A choice must be made.

This result (Kleinberg, Mullainathan, Raghavan, 2017) explains why the COMPAS debate has no simple correct answer: Northpointe claimed the model was calibrated, ProPublica measured unequal error rates. Both were right… according to their own criterion.

How to choose a definition?

Context Recommended definition Why
Recruitment Equal Opportunity Among the qualified, equal chance of being chosen
Credit Calibration + monitored parity Individual justice + group monitoring
Criminal justice Equalized Odds Presumption of innocence protected
Health (screening) Sensitivity by subgroup Do not miss cases
Targeted advertising Demographic Parity (context-dependent) Avoid discriminatory targeting (housing, employment)
TIPPractical rule — Document your choice in a fairness statement that specifies the definition adopted, the monitored groups, and the tolerated threshold. This documentation is required by future EU AI Act audits for high-risk systems.

Concrete case: the 4/5 rule (US)

In the United States, the EEOC’s Uniform Guidelines on Employee Selection Procedures establish the “four-fifths rule” (1978): the selection rate for a protected group must be at least 80 % of the rate of the most selected group.

Famous cases — COMPAS, Amazon, Apple Card

NOTEObjective — Analyze in depth three emblematic cases that shaped the debate on algorithmic bias: COMPAS (justice), Amazon (recruitment), Apple Card (credit). Understand the mechanics of bias, corporate responses, and lessons for your own projects.

Learning objectives

TIPAt the end of this module — You will be able to recount each case in two minutes, identify responsibilities, and extract 5 practical lessons per case, transferable to your own professional context.

Case 1: COMPAS (Northpointe / Equivant)

The system

COMPAS — Correctional Offender Management Profiling for Alternative Sanctions — is a decision-support tool used since the early 2000s by American judges. It predicts recidivism risk based on 137 questions completed by the defendant and the probation service.

The score (1 to 10) is used to decide parole, bail amount, or sentence length.

The ProPublica investigation (2016)

ProPublica analyzed 7,000 scores in Florida and compared them with re-arrests over two years. Results:

Northpointe’s defense

Northpointe responded that its model is calibrated, hence fair for each individual. The debate reveals Kleinberg’s impossibility: calibration and equalized odds cannot both be satisfied when base rates differ.

The lessons

NOTEFive takeaways
  • Choose and document your fairness definition before deployment
  • Score explainability must be published
  • A proprietary (black-box) model in justice raises democratic issues
  • Unequal base rates often reflect structural inequality
  • Regularly audit with an independent third party

Case 2: Amazon recruitment (2014–2018)

The context

Amazon began developing an internal automated résumé-screening tool for tech positions in 2014. The model was trained on 10 years of applications, with the target “résumés resembling profiles hired over the past 10 years.”

The problem

The hired profiles were overwhelmingly male (the tech sector having historically been male-dominated). The model therefore learned to favor men.

More precisely:

The response

Amazon abandoned the project in 2017 and never deployed the tool “in production.” The information appeared in Reuters in 2018, becoming a global case study.

The lessons

NOTEFive takeaways
  • Never use past performance as a target when it reproduces inequality
  • Removing sex is not enough: track proxies (vocabulary, degrees, clubs)
  • Audit the model on subpopulations before any deployment
  • Involve HR, legal, and ethics functions from the design stage
  • Prefer decision-support (human in the loop) over 100 % automated screening

Case 3: Apple Card (2019)

The trigger

In November 2019, developer David Heinemeier Hansson tweeted that he and his wife share their accounts, yet she received a credit limit 20 times lower than his on the new Apple Card (issued by Goldman Sachs). The tweet went viral.

go-further

This article covers the most useful excerpts — the full AI Ethics Governance course (11 chapters, 35 lessons, corrected exercises and final project) takes you all the way.

./access-the-full-course free course: Claude Cowork

FAQ

How long does it take to learn AI Ethics Governance?
With a structured progression (11 chapters, 35 short and practical lessons), you reach an operational level in a few weeks at 30–60 minutes per day. The key is to practice each concept immediately.
Are there any prerequisites?
Basic computer science knowledge is sufficient. If you can use a terminal and read simple code, you are ready.
Where to start concretely?
Reproduce the commands in this article, then follow the full AI Ethics Governance course: it sequences the 35 lessons in order, with exercises and a final project.

📬 Want to receive this type of guide every week? Subscribe for free — real code, zero fluff.