AI ethics

In short

AI ethics is the study and practice of making AI fair, transparent, accountable and respectful of people's rights. It asks not just whether an AI system works, but whether it should be used and who it might harm.

Also called: Ethics of AI, Ethical AI, Responsible AI
Topic: Safety, ethics and risk
Level: Beginner
First described: Ethical concerns about computing are decades old, but the first intergovernmental standard on AI was the OECD AI Principles, adopted in May 2019².

AI ethics covers questions such as: Is this system biased against some groups? Can people find out why it made a decision about them? Who is responsible when it goes wrong? Does it respect privacy?

Hundreds of organisations have written ethics principles. A 2019 study in Nature Machine Intelligence mapped these guidelines and found them converging on five principles: transparency, justice and fairness, non-maleficence (avoiding harm), responsibility and privacy. It also found sharp disagreement about what those words mean in practice and how to apply them¹.

Governments have signed up to shared texts. The OECD AI Principles, adopted in May 2019 and updated in May 2024, were the first intergovernmental standard on AI. They include human rights and fairness, transparency and explainability, robustness and safety, and accountability, and have 47 adherents². In November 2021, UNESCO's 193 member states adopted a Recommendation on the Ethics of AI, built around human rights and human dignity³.

Principles are not law. In the UK, ethical problems with AI usually reach the courts through existing law, such as equality and data protection rules. In 2020 the Court of Appeal found South Wales Police had not done enough to check that its facial recognition software was free of racial or gender bias⁴.

AI ethics overlaps with AI safety but is not the same. Ethics focuses on fairness, rights and accountability today; safety work also covers accidents, misuse and loss of control.

Example

A council wants to use an AI tool to flag families for social care visits. An ethics review would ask whether the tool flags some communities more than others, whether families can challenge a decision, and whether a human makes the final call.

Why it matters

AI is already used to screen job applications, set prices and help with policing. Ethics is the set of questions that decides whether those systems treat people fairly.

Questions people ask

What are the main principles of AI ethics?

A 2019 review of ethics guidelines worldwide found five recurring principles: transparency, justice and fairness, avoiding harm, responsibility and privacy¹. The OECD's principles add inclusive growth, robustness and safety, and accountability².

Is AI ethics the law?

Mostly not on its own. Ethics principles from the OECD and UNESCO are voluntary²³. But many ethical concerns are covered by existing law. In the UK, the Court of Appeal ruled in 2020 that South Wales Police breached its equality duty by not checking its facial recognition software for bias⁴.

What is the difference between AI ethics and responsible AI?

They are often used interchangeably. AI ethics usually refers to the principles and debates; responsible AI usually refers to how companies put those principles into practice through reviews, testing and documentation.

Related terms

Sources

Back to the glossary