Are there enough open source aero engineering projects to give the current ai approaches a remotely plausible amount of training data?

Imagine is a good word to use. Before offering a solution, understand the problem first - is "debugging speed" a problem that needs solving, in this case?

Much more likely is that it would hallucinate a plausible sounding but incorrect answer and send intermediate and junior engineers on a wild goose chase