Ai Agents Need Permissions, Not Forgiveness

​AI agents are acting on our behalf. They answer our clients, handle our requests and increasingly make decisions for us. But can we trust that they act in our interest?

​Until now, trust has been growing slowly between people and institutions. Agents bring a new kind of relationship, and we are only beginning to understand it. Whose values does an agent inherit, and what happens when it gets things wrong?

​When trust runs short, forgiveness becomes a luxury. A colleague who oversteps can apologise and learn; an agent that oversteps may already have moved money, shared data or made a promise in our name. What may agents do without asking, and what must always need our permission?

​Dr. Maria Fay (SAP & University of Münster) brings the perspective of AI adoption across industries from pharma to manufacturing to consumer products, where governing agents at scale is a daily question. Matthias Plattner (Julius Baer, Swiss Bankers Association) brings the perspective of private banking, where client trust is the core of the business and choosing a model is a decision about values. Both speakers have contributed to the recently published book, “Computers with Salaries and Cemeteries: AI Ethics from Industry to Philosophy to Science Fiction”, and will address some of the ideas featured in the book.

​After their conversation, the discussion is yours: what may an agent do without asking, whose values should it carry, and who decides? Bring your own experience with agents.