This joint presentation by Volkswagen Group and Keysight Technologies outlines an approach for Generative AI based creation and validation of synthetic training data for automotive AI systems. We demonstrate how AI generation workflows improve real-world performance, and why rigorous validation is essential: generated data must match domain semantics and characteristics before being used in training. Using Keysight’s AI Software Integrity Builder, Volkswagen analyzes datasets and model behavior to assess data quality, compare distributions, reveal discrepancies, and guide metric based optimization in a closed loop. Join us to learn how this approach enables shifting training data collection from road to AI.

• Insights into the challenges and benefits of using AI generated data in testing and training workflows.
• Methods to verify that AI created data meets strict domain and quality constraints.
• Approaches to assess how synthetic data influences automotive AI model performance.
• Suggested workflows to integrate AI data generation into existing development processes.
• Practical guidance on applying AI driven data creation to enhance reliability and scalability.