AI and ML in Testing

Shadow Testing: The Secret AI Adoption Already Happening in Your Supply Chain

Adam Mackay Head of AI Research / QA-Systems, UK

Shift-Left and Continuous Functional Testing with System Simulation

Dr James Hui System Simulation Product Manager / Wind River, UK

The role AI/ML and big data play in creating an operations to design/engineering safety loop

Gary Brown Principal Safety Engineer & AI Referent / Airbus Commercial SAS

Accelerating Airworthiness: Automation's Role in Faster Verification for DO-178C Certification

Paul Butcher UK Programme Manager and Head of Dynamic Analysis / AdaCore, UK

AI and ML are emerging in aerospace and avionics, raising complex testing and certification challenges, presenting significant verification and validation challenges. AI use cases are broadly categorized into narrow/specialized AI (e.g., object detection) and generative models, which pose greater unpredictability due to non-deterministic outputs. Current architectures favour non-safety-critical, passive AI/ML systems operating in parallel with primary avionics functions—supporting anomaly detection, predictive maintenance, and decision support. In light of limited data availability, what methods are most appropriate for testing and validating? How do you approach verification and validation of non-deterministic AI systems within the constraints of deterministic safety standards like ED-324/DO-178C? Where will EUROCAE technical standards (WG114) support the development of systems and the certification of aeronautical systems implementing AI-technologies?