Six rules for people who build AI
Ethics is about what is morally right and what is morally wrong. It is not the same as the law. Something can be legal but not ethical. No law says you must tell a classmate that her bag is open, but telling her is the kind thing to do. Designers of AI should follow six ethical rules. Be inclusive: AI tools should work for all users, not only a chosen group. Be unbiased: rely on facts, not on opinions. Be beneficial: use AI to help society, never to exploit it. Be transparent: explain openly how the tool works. Be protective: think about the privacy and security of users. Be accountable: if the tool does not work as it should, own up to the mistake, say what went wrong and describe the fix. Following these rules helps ease people’s fear and builds trust.
| Rule | What it means |
|---|---|
| inclusive | works for all users |
| unbiased | based on facts, not opinions |
| beneficial | helps society |
| transparent | explains how it works |
| protective | keeps privacy and security |
| accountable | admits mistakes and fixes them |
| The law | Ethics | |
|---|---|---|
| asks | what must I do? | what is right? |
Rule cards. Make six cards, one for each rule, with a small symbol. A partner reads these situations and you hold up the rule that was followed or broken: an app explains how it chooses suggestions; a reading tool does not work for pupils who need large text; a company hides a mistake; an AI shares users’ private messages; a tool is built to trick people into buying; an AI is trained on facts, not opinions.