AI Constitution
1

Thesis

Right to Dignity, Fair Treatment, and Non-discrimination

Sign in to propose an editHistory

AI should not violate the human right to dignity and fair treatment — including the right to be free from dishonor, reputational harm, degrading treatment, emotional manipulation, and interactions inappropriate for one's level of psychological development; and the right to be treated fairly and as equal in dignity, without bias or prejudice based on race, colour, sex, language, religion, political or other opinion, national or social origin, property, birth, or other status.

Summary

The case for including this

Prohibiting degradation, manipulation, and bias directly targets some of AI's most documented failures, from discriminatory algorithmic decisions to manipulative persuasion, and protects human dignity at scale. Because AI systems can entrench prejudice invisibly across millions of interactions, an explicit fairness and anti-manipulation right is essential to prevent automated injustice. It unites several distinct harms under the shared value of treating people as ends in themselves.

The case for changing or excluding this

Including this trades accuracy against fairness: different fairness definitions are mathematically incompatible, and forcing one can reduce accuracy in ways that harm the very groups it aims to protect. The clause also packs together dignity, manipulation, and bias—three complex and separately contested concepts—into one line, and each resists definition: all persuasion shapes beliefs, and 'bias' can mean anything from illegal discrimination to any disparate outcome. It should be unbundled and each component given a workable standard that specifies which fairness it means.

Discussion

Sign in to join the discussion.

Related resources