Research taste in history: the importance of learning Babylonian math

Allan Chapman, an Oxford historian of science known for his eccentric persona and deep immersion in past scientific minds, died earlier this year, prompting a reflection on how AI is reshaping history education and research. The author argues that while language models excel at memorization, spaced repetition, and brute-force archival work, they remain stuck in a "median perspective" that lacks the "research taste" found in humanistic disciplines. Chapman exemplified the kind of out-of-distribution thinker—wearing tweed, carrying two pocket watches set to GMT and Oxford sidereal time—whose ability to inhabit alien worlds of thought may become both more important and more rare as AI tools spread. The piece contends that traits like eccentricity and creativity will matter more, not less, in an AI-shaped era, because variety rather than uniformity drives intellectual progress. What to watch next: how history departments and humanities funders adapt curricula and hiring to reward idiosyncratic, time-traveler-style scholarship that frontier models cannot replicate.