In person & onlineWebinar

Built on trust: Why dependable outcomes need trusted context and clear control

  • * Register (or log in) to the Neural Network to add this session to your agenda or watch the replay

  • Date
    29 September 2026
    Timeframe
    16:00 - 17:00 CEST, Geneva
    Duration
    1 hour

      When AI supports services used by millions, even a small failure rate can have serious human consequences. Generative AI is probabilistic, and a convincing answer is not necessarily correct, appropriate or safe. How can organisations harness its capabilities without passing that uncertainty on to the people they serve? 

      This AI for Good session explores the context and controls needed to make AI more dependable, including a governed data foundation for retrieving trusted information, alongside clear permissions, evidence requirements and rules governing how that information is found, used and acted upon. Concepts such as knowledge graphs, provenance and deterministic controls will be translated into practical questions of meaning, authority and accountability. 

      Using a humanitarian service-information scenario, Philip Miller will demonstrate how current evidence, local knowledge and human review can support dependable outcomes. Participants will gain a practical framework for selecting an initial workflow, evaluating solutions and measuring service quality. Designed for programme leaders, practitioners, policymakers and technical teams, the session makes the case for AI that is useful, governable and worthy of trust.

      Participants will learn to 

      • Explain why model accuracy alone is insufficient for dependable services at scale. 
      • Distinguish trusted context from the controls governing discovery, access and action. 
      • Define evidence, human-review and evaluation requirements for one practical use case. 

       

      Are you sure you want to remove this speaker?