Why AI Ethics Matters
In 2018, a man in Detroit named Robert Williams was arrested at his home in front of his daughter and wife. Police had matched his face to a surveillance photo using an AI system — and the system was wrong. Williams spent 30 hours in jail before the mistake was discovered. He had done nothing wrong. The AI had failed him. This is not a hypothetical. It is not science fiction. It happened, it keeps happening, and it is one reason why the question of how AI is built and used is not just technical — it is deeply human. That is what AI ethics is about.
What Is AI Ethics?
Ethics is the study of what is right and wrong, fair and unfair, and who bears responsibility when things go wrong. AI ethics applies that same rigorous thinking to the systems we build and deploy. AI now makes or shapes decisions in medicine — flagging cancer scans. In criminal justice — estimating recidivism risk. In hiring — screening resumes. In lending — approving loans. In education — recommending content. In each of these, a flawed or unfair system does not just produce a bad answer on a test: it alters someone's life. AI ethics asks: Are these systems fair? Are they honest? Who benefits and who is harmed? Who is responsible when something goes wrong?
AI ethics is the study of the moral principles that should guide how AI systems are designed, deployed, and governed — including questions of fairness, accountability, transparency, and harm.
Here is a useful frame: every AI system encodes choices. Someone decided what data to collect, what to optimize for, who to test on, and how to deploy the result. Each of those choices reflects values, whether or not the builders thought of them that way. A system built to maximize engagement on a social platform encodes the value 'time on screen matters most' — and that choice has consequences for mental health, political polarization, and what information spreads. Ethics is not about blaming technologists — most are genuinely trying to build good things. It is about making those value choices visible, deliberate, and accountable.
Five Pillars of AI Ethics
Researchers and organizations around the world have converged on a handful of core principles that most ethical AI frameworks share. You will encounter all of them across this module. Fairness: The system should treat similar cases similarly, and not systematically disadvantage any group. Transparency: Stakeholders should be able to understand how and why a system makes its decisions. Accountability: When something goes wrong, someone must be answerable for it. Privacy: Systems should collect and use personal data only for legitimate, consented purposes. Beneficence and non-maleficence: Systems should do good and avoid doing harm. These principles sound simple. Applying them is hard — they sometimes conflict with each other, and they require ongoing judgment, not just a checklist.
Many teams treat ethics as a review step after a system is built. That approach nearly always fails. Ethical choices are embedded in data collection, in how a problem is framed, in which metric is optimized. Ethics must be part of the design process from the very beginning.
Flashcards — click each card to reveal the answer
Robert Williams was arrested because of a flawed AI system. What principle of AI ethics does this case most directly illustrate as being violated?
Why do researchers say 'ethics must be baked in from the start' rather than reviewed at the end?
Map an AI Decision
- Think of one AI system you interact with — a recommendation engine, a search result, a spam filter, a content moderation system.
- Write down: What decision does it make? What data do you think it uses?
- Now apply the five pillars. For each pillar, write one sentence: Is this system likely doing well or poorly on this dimension? Why do you think so?
- Finally, write one question you would ask the system's builders if you could sit down with them.
- Share your question with a classmate or partner and discuss whether your questions overlap.