What if AI lets us do for intellectual endeavors what DNA did for taxonomy? We categorize thought by its visible traits: law, math, philosophy, programming, history. But those may be like classifying species by appearance. If seemingly unrelated tasks consistently activate the same internal routes in neural networks, we may discover a deeper taxonomy based on the underlying operations of thought itself. Statutory interpretation might turn out to be computationally closer to debugging software than to other kinds of legal reasoning, just as DNA revealed unexpected relationships between species that looked nothing alike. AI may not just learn our categories. It teach us how we misdrew them from the start.