EXPOSED
The bulk of a historian's paid day is reading, summarizing, synthesizing secondary literature, drafting narrative text, and preparing exhibit copy, reports, or background briefs — exactly the work large language models already do at usable quality and enormous speed. What resists is physical archival work: locating and handling uncatalogued manuscripts, reading damaged or handwritten primary sources in context, authenticating provenance, and standing behind an interpretive claim under professional or legal scrutiny. There is no license, no signature requirement, and a tiny national headcount (about 3,450), so most employed historians work in government, consulting, museums, or historic preservation rather than pure scholarship — and those roles pay for defensible judgment, not prose volume.
Mixed — a routine tier and a judgment tier. How much of the day-to-day work current AI already does at usable quality.
Some physical or field component. Physical work in unpredictable environments. Robotics lags language models badly.
No licence, no signature requirement. Is a licensed human legally required to sign, and personally liable if it goes wrong?
Some relationship component. Do buyers specifically pay for a human — presence, care, accountability?
Meaningful discretion. Does the role own consequential calls made with incomplete information?
History Teachers, Postsecondary EXPOSED
Has AI actually changed your work?