What isn't reasonable about this approach? I think most people have got GeoAI backwards, creating derived imagery via AI is a terrible idea. Building better tools than we have with AI, using existing geoscience principles is absolutely the right approach.