Ethical AI challenges are precisely the kind of problems that Action Learning is designed for. Here is why:
They are genuinely unresolved. There is no textbook answer to “how do we balance the efficiency gains of our AI hiring tool against its potential for demographic bias?” There is no algorithm for “should we deploy this AI system when its error rate is acceptable on average but significantly worse for a specific demographic group?” These are problems that require genuine inquiry, honest wrestling, and the development of new understanding: exactly what Action Learning cultivates.
They involve multiple values in tension. Ethical AI consistently requires navigating conflicts between values that all have legitimate claims: efficiency and fairness, transparency and privacy, innovation and precaution, individual benefit and collective risk. These tensions cannot be resolved by applying a single framework. They require the kind of careful, multi-perspectival thinking that Action Learning develops.
They are deeply contextual. The right approach to AI ethics in a hospital is different from the right approach in a financial services firm, which is different again from the right approach in a social media company. Programmed knowledge can provide frameworks, but contextual wisdom: knowing how principles apply in this specific situation: comes from experience, reflection, and the questioning of others who have faced similar challenges in different contexts.
They involve significant personal and organisational risk. Raising ethical concerns about an AI system, slowing a deployment, or recommending that a project be cancelled are not comfortable actions. People who take them face real professional consequences. Action Learning provides a supported space to think through these situations: not to make the risks disappear, but to develop the clarity and courage to act with integrity.
They benefit from diverse perspectives. Action Learning Sets work best when their members bring different backgrounds, roles, and experiences. For ethical AI, this means including diverse technical, operational, community, and disciplinary perspectives. The questions that a community advocate asks about an AI challenge will be different from those a data scientist asks: and both will be necessary.
Reflection question: What ethical AI challenge would you bring to an Action Learning Set if you had access to one? What questions would you most want people to ask you about it?