(TR25) (INCYBER) Wanted: resilient AI for mission-critical systems

AI is often seen as an advantage in terms of overall resilience because we can use it to automate certain tasks. But when it is used to enhance mission-critical systems, they can become weak points. How do we ensure that AI failures will not compromise mission-critical systems? What are the formulas and best practices for creating resilient AIs that can adapt to unforeseen situations and keep running smoothly even in the event of critical incidents? How do we increase the robustness, reliability, and transparency of AI algorithms to meet resilience requirements in strategic areas such as healthcare, security, and critical infrastructure?