AI ethics is the applied branch of ethics that examines the moral responsibilities of those who design, deploy, and use AI systems, as well as the societal consequences that follow from those systems operating at scale. It draws on philosophy, law, social science, and technical research to address questions of fairness (does the system treat different groups equitably?), accountability (who is responsible when the system causes harm?), transparency (can its decisions be explained?), privacy (what data was used and how is it protected?), and broader concerns such as environmental cost, labor displacement, and concentration of power.

Think of it like building codes for software. Just as structural engineering codes exist not because architects are malicious but because systems affect many people who have no say in how they are built, AI ethics establishes norms to protect those affected by AI decisions.

A concrete example is auditing a hiring model for disparate impact: a system trained on historical data may encode past discrimination, systematically scoring applicants from certain demographic groups lower even when those features are not included explicitly. Addressing this requires both technical interventions (re-weighting training data, fairness-aware loss functions) and governance interventions (independent audits, documentation requirements).

AI ethics intersects with Bias (a central technical manifestation of ethical failure), AI Alignment (ensuring systems do what humans intend), AI Governance (the regulatory and policy layer), Hallucination (which raises honesty and reliability concerns), and RLHF (a technique partly motivated by ethical goals around helpfulness and harmlessness).

You will encounter AI ethics in responsible AI frameworks published by organizations like the OECD and IEEE, in corporate AI use policies, in model cards, and in academic venues such as the ACM FAccT conference.