There is something almost automatic about how organisations respond to ethical AI challenges: identify the problem, analyse its root causes, develop solutions, implement them, evaluate results.

This is the deficit model: and it is deeply embedded in how institutions think. It has genuine value. Rigorous problem analysis has produced important advances in AI safety, fairness, and accountability.

But the deficit model also has costs that are rarely examined.

When we begin with problems, we frame people as part of the problem. AI teams become sources of bias to be audited. Executives become risks to be managed. The entire conversation is organised around insufficiency: what is lacking, what is broken, what must be fixed.

This framing generates defensiveness. People who feel blamed do not collaborate openly. People who feel surveilled do not experiment freely. Organisations that are constantly in crisis mode do not develop the reflective capacity needed for genuine ethical development.

David Cooperrider, who developed Appreciative Inquiry in the 1980s, made a foundational observation: organisations move in the direction of the questions they ask.

If you ask “what’s wrong with our AI?” you will find things that are wrong. If you ask “when have we made the most ethical AI decisions, and what made that possible?”: you will find something different, and equally true.

This is not positive thinking. It is a deliberate methodological choice about where to direct attention and energy. The insight is that the conditions that produce ethical AI already exist, at least partially, in most organisations. People care. Values exist. Good decisions get made.

The task of Appreciative Inquiry is to find those conditions, understand them deeply, and build an organisation where they are the norm rather than the exception.

Reflection question: Think of a time when your organisation made a decision you were proud of in relation to AI or technology. What made that possible?