One of the most important applications of Appreciative Inquiry to ethical AI is in the domain of bias.

The conventional approach to AI bias is diagnostic: audit the system, identify where bias exists, measure its magnitude, implement mitigations. This is necessary work. But it is incomplete.

Appreciative Inquiry asks a complementary question: in AI systems and processes where fairness has been achieved: or meaningfully approached: what conditions made that possible?

This question opens up a different kind of investigation. It looks not just for what went wrong, but for what went right. It surfaces the human decisions, the structural conditions, and the cultural values that produced more equitable outcomes: and asks how those can be replicated, scaled, and made permanent.

Research in this area consistently finds that AI systems tend to be fairer when:
– The teams building them are genuinely diverse: in demographics, disciplinary background, and life experience
– Community members who will be affected by the system are involved from the earliest stages of design
– Ethical review is built into the development process, not bolted on at the end
– There is psychological safety to raise concerns and slow down
– Leadership actively models curiosity about impact, not just enthusiasm for capability

Each of these conditions is something that can be cultivated and strengthened. Appreciative Inquiry provides a methodology for doing so: by finding the examples where these conditions existed, understanding them deeply, and designing organisations where they are consistently present.

This does not mean ignoring existing bias. It means pursuing both tracks simultaneously: diagnosing and mitigating current harms while building the organisational conditions that make future harms less likely.

Reflection question: In your experience, what conditions have produced the most equitable or human-centred technology decisions? What do those conditions have in common?