Action Learning for AI Ethics: Learning While Doing

An Action Learning Set meets regularly: typically every 4–6 weeks: over a sustained period, often 6–12 months. Each meeting follows a rhythm that balances individual focus with collective support.

The structure of a Set meeting:

Check-in: each member briefly shares where they are with their problem since the last meeting. What actions did they take? What happened? What are they thinking about now?

Individual focus time: each member in turn receives the full attention of the Set. They describe their current challenge, explain where they are stuck, or share a development that has raised new questions.

Questioning: the rest of the Set’s role during this time is almost entirely to ask questions. Not to advise. Not to share their own experience. Not to suggest solutions. To ask questions that help the presenter think.

This is the hardest part of Action Learning for most people: and the most important. We are socially conditioned to offer advice when someone shares a problem. “Have you tried…?” “What about…?” “In my experience…”

These responses feel helpful. But they shift the focus from the presenter’s understanding to the adviser’s. They offer programmed knowledge (P) when what is needed is questioning insight (Q).

Good Action Learning questions are open, genuine, and often simple:
– “What do you most want to understand about this situation?”
– “What assumptions are you making about why this is happening?”
– “What would it mean for you personally if this worked out differently than you expect?”
– “Who else has a stake in this that you haven’t yet talked to?”
– “What are you not saying about this situation?”

Action planning: after the questioning, the presenter identifies what they will do before the next meeting. This is not a comprehensive plan. It is one concrete step: an action that will produce new information or new experience that the Set can learn from together.

Reflection: the meeting ends with brief collective reflection on the quality of the Set’s work. Were the questions helpful? What kind of questions seemed to open up new thinking? What patterns are emerging?

The Set Adviser

Action Learning Sets often benefit from a facilitator: called a Set Adviser: whose role is to help the Set develop its questioning practice, not to contribute content. A good Set Adviser asks the Set questions about its own process: “What kind of questions have you been asking?” “What are you noticing about how you’re working together?” “What is the Set not talking about?”

Reflection question: What would it feel like to be in a group where people asked you questions about your ethical AI challenges instead of offering advice? What might you discover?

Scroll to Top