Before I waste valuable learning/track time on this myself, I thought it might pay to ask. Start with the default car and let the AI run a couple of laps. Record the time. Make an adjustment to, say, rear flap. Let the AI drive a couple of laps and record the time. Make more changes, letting the AI drive laps and continue recording the times. Theoretically because you're not dealing with human faults, you could arrive at a "best build." Since the AI would (I believe) attempt to drive the same line no matter what the setup, the lap-to-lap comparison would be valid. Once you arrive at a "best" build, fire the AI and take over the car yourself. Where's the ignorant fallacy in this? I expect the most common response would be that every car has to be tuned not only to the track but the driver's skill. Maybe it would work as a training tool...if the AI can drive this setup at this time, I should be able to do so, too. It eliminates any question that the car is the problem and puts the onus on the driver. This then leads to a second question. Could you run an offline series where you were the pit manager? You build the car, choose the tires, etc., and let an AI driver race it against the AI competition. I'm not sure that would hold interest for very long, but I just wonder whether it would work at all.