There is something quietly subversive about the Action Learning commitment to questioning over advising.
In most professional contexts, expertise is performed through answers. The expert knows things. They provide information, recommendations, solutions. Questions are what you ask before you know enough to contribute. Once you know enough, you advise.
Action Learning inverts this. The most valuable contribution a Set member can make is a good question. And a good question is one that opens up new thinking in the presenter: that surfaces an assumption, reveals a connection, or creates productive discomfort.
For ethical AI, this has profound implications. The AI ethics field often operates in expert mode: producing frameworks, guidelines, principles, and toolkits. These are valuable. But they can also create a false sense that ethical AI is a solved problem, requiring only the correct application of the right framework.
The honest reality is that ethical AI is an ongoing inquiry. We do not yet know how to build truly fair AI systems. We do not yet know how to create meaningful accountability for AI decisions. We do not yet know how to involve affected communities in AI governance in ways that are genuinely empowering rather than performative. These are open questions.
Action Learning’s commitment to questioning is, in this sense, an epistemic honesty. It acknowledges that the most important work in ethical AI is not applying known solutions but developing new understanding of genuinely novel challenges.
The question “what are you not yet asking about this situation?” is among the most powerful in Action Learning’s repertoire. For ethical AI practitioners, it is a discipline worth developing: regularly asking what questions are not being asked, whose voices are absent from the inquiry, what assumptions are so embedded that they are invisible.
This is the spirit of Ethos Sophia’s tagline: one mind cannot see it all.
The questioning practice of Action Learning is a methodological expression of that conviction.
Reflection question: What question about your organisation’s AI practices are you currently not asking? What is making that question difficult to ask?