The impact of generative AI on the automotive industry: benefit or risk for rally drivers?

27 Tem 13:23

In my final year working with the development team of simulators for a racing car manufacturer, we tested a generative AI to create suspension variants and calibrate the aerodynamics of a rally car prototype. The algorithm delivered several configurations that, in virtual tests, reduced lap times in tight corners by 0.3 to 0.5 seconds without the driver having to change their driving style. What was interesting was that once we applied these proposals to the real car, the drivers noticed the vehicle responded in a more predictable way, allowing them to focus on strategic decision-making instead of fighting against unexpected behavior. However, problems also arose: the AI tended to optimize for very specific track conditions, and when the terrain changed rapidly (such as on wet gravel sections), the generated setup proved less safe, forcing us to impose validation limits and maintain a manual "override" to prevent the car from operating outside approved safety margins. So, I believe generative AI can be a great ally in car preparation, as long as it’s combined with strict human oversight and regulations requiring validation of every solution in real-world scenarios. The key is to use AI as an optimization tool, not as a replacement for the driver’s judgment and the homologation process; in this way, we preserve the essence of human control while leveraging the benefits of intelligent simulation.