AI Ethics and Society
AI now helps decide who gets a loan, a job interview or a medical test. Explore where unfairness comes from, why "fair" has more than one meaning, what privacy means when data is everywhere, and who is responsible when AI gets it wrong.
Start this course →Lessons
- 1Why AI raises new questionsSee why AI decisions need ethical thinking, and who is affected.6 min
- 2Where unfairness comes fromTrace bias back to data, and see why "fair" has more than one meaning.8 min
- 3Privacy in a world of dataUnderstand consent, anonymisation and its limits, and ways to learn from data more privately.7 min
- 4Who is responsible?Know what transparency, explainability, oversight and accountability mean in practice.7 min
What you will learn
- AI repeats a decision rule thousands of times, so a flaw becomes a pattern.
- Good ethics starts by asking who the stakeholders are.
- Many questions have no single right answer, so the reasoning must be open to discussion.
- Bias enters through historical bias, sampling bias and measurement bias.
- Simply deleting a sensitive column often does not remove bias.
- Different definitions of fairness can conflict, so choosing one is a value judgement.
- Privacy is control over information about yourself; meaningful consent requires understanding.
- Anonymisation can fail through re-identification.
- Methods such as federated learning reduce how much raw data must be collected.
- Transparency: say when AI is used. Explainability: give understandable reasons.
- Human oversight keeps people able to check and override.
- Accountability means someone answers for outcomes; governments are now writing AI laws.
Terms explained in this course
- ethics
- Careful thinking about what is right and wrong, and how we should act.
- stakeholder
- Anyone affected by a decision or a system, not only the people who build or buy it.
- historical bias
- Unfairness already present in past decisions, which a model copies when it learns from them.
- sampling bias
- When the data collected does not represent everyone the system will be used on.
- measurement bias
- When the thing measured is a poor stand-in for what we actually care about, and the gap differs between groups.
- fairness
- In AI, the idea that a system should not treat people worse because of who they are. There are several precise definitions, and they can conflict.
- privacy
- People's ability to control information about themselves.
- consent
- Agreement given freely, knowing what you are agreeing to.
- anonymisation
- Removing information that identifies people from a dataset.
- re-identification
- Working out who someone is in a supposedly anonymous dataset by combining it with other information.
- surveillance
- Watching or recording people's activities, often at large scale.
- transparency
- Being open about when AI is used, what data it uses and how it is meant to work.
- explainability
- The ability to give a reason for a particular AI decision that a person can understand.
- accountability
- Clear responsibility for a system's outcomes, including someone who answers for mistakes and can fix them.
- human oversight
- Keeping people able to check, override or stop an AI system's decisions.