History professors

Three parts of your work where AI is already doing real lifting, and what stays yours.

• Seminar room1876–1950

  Seminar room, archive, and typewriter (German research model era)

  The history professor of the founding era worked with primary sources in physical form: handwritten manuscripts, printed newspapers, government documents, and monographs. The German seminar model, transplanted to Johns Hopkins by Herbert Baxter Adams in 1876, organized the work around a shared table and primary source critique. The typewriter arrived at American universities in the 1880s and gradually replaced longhand for lecture notes, correspondence, and article drafts, but the archive visit remained the irreducible core of historical research. Lectures were delivered from handwritten or typed notes. Student essays were graded by hand.

• Microfilm reader1950–1980

  Microfilm reader and photocopier (postwar archive democratization era)

  Microfilm transformed how historians accessed primary sources. Rather than traveling to distant archives for weeks at a time, scholars could order microfilm reels of newspapers, government records, and manuscript collections to their home institution. The Readex American Newspapers project (1950s) and the microfilming of the National Archives' vast record groups made archives accessible to faculty without travel budgets. The photocopier, arriving in American academic offices in the 1960s, allowed historians to photocopy archival documents for annotation at home rather than transcribing by hand. These two tools compressed the archive-to-publication cycle and expanded the range of primary sources a single scholar could engage.

  Effect on the work

  Microfilm and photocopying effectively increased the productivity of individual history faculty, allowing a scholar at a state university with limited travel funds to access primary sources that would have required extended archive visits in earlier eras. This contributed to the rise of specialized, source-intensive historical monographs as the dominant scholarly form.

• Word processor1980–1995

  Word processor and early digital databases (PC revolution era)

  The IBM PC (1981) and Apple Macintosh (1984) arrived in academic offices through the 1980s, and word processing replaced the typewriter for lecture preparation, article drafting, and correspondence. For history faculty, the impact was most visible in manuscript revision: the ability to restructure a chapter without retyping the whole document fundamentally changed how historians wrote and revised. The first digital databases also appeared: Historical Abstracts went online in the early 1990s, and library OPACs replaced card catalogs. These tools reduced the mechanical overhead of research and writing without yet changing the core archival methods of the discipline.

• Internet1995–2010

  Internet, JSTOR, and digitized archives (digital turn in historical research)

  JSTOR launched in 1995 as a digital archive of back issues of major academic journals, giving historians desktop access to secondary literature that had previously required physical library visits. Google Scholar (2004), Google Books (2004), and the Internet Archive's growing collection of digitized texts began to make large swaths of older scholarship and primary materials searchable from a faculty member's desk. The Library of Congress launched its American Memory project in 1994, the British Library began digitizing newspapers, and the HathiTrust Digital Library launched in 2008. The aggregate effect was a substantial compression of the literature-review phase of historical research and a shift toward source materials that had been digitized, creating a visible selection bias toward the recent, the English-language, and the institutionally well-digitized.

  Effect on the work

  The digital turn made the average history article faster to research and write, but it also raised the implicit methodological standard: a scholar who had not consulted JSTOR, Google Scholar, and major digital archive collections was considered to have done incomplete research. The digital divide between well-funded research universities (early digital subscriptions) and teaching-focused institutions compressed over the 2000s as costs fell.

• Learning Management Systems2010–2022

  Learning Management Systems and online instruction (Canvas, Blackboard, Moodle)

  Canvas launched in 2011 and rapidly displaced Blackboard at many universities through the 2010s, becoming the dominant LMS for history courses. Online and hybrid course delivery expanded throughout the decade, accelerating sharply during the COVID-19 pandemic (March 2020). History courses moved to Zoom, recorded lectures, and asynchronous discussion boards. For history faculty, the LMS era shifted a significant portion of the teaching workload toward course design and digital content production, tasks that required new skills but did not change the core research methods of the discipline. The pandemic's forced experiment with online history teaching documented both what translated well (recorded lectures, asynchronous primary source analysis) and what did not (the live seminar discussion dynamic, which most historians regard as the highest-value learning environment for the discipline).

• AI handwritten text recognition2016–2023

  AI handwritten text recognition: Transkribus (READ-COOP, 2016)

  Transkribus, launched by the READ-COOP cooperative in 2016, applied machine learning to the long-intractable problem of automated handwritten text recognition (HTR) for historical documents. By 2023, hundreds of universities were using Transkribus to generate first-draft transcriptions of handwritten manuscripts, letters, court records, and archival registers. Trained Transkribus models achieve character error rates of 5-10% on period-specific document types. For history faculty, the tool transformed the most time-intensive phase of primary source work: a semester-long paleography project that once required weeks of hand transcription could now be scaffolded by AI, with the scholar's expert correction work concentrated on the errors that matter historically (misread names, anachronistic vocabulary, damaged text). University of Helsinki and Central Connecticut State University both documented pedagogical uses of Transkribus in undergraduate history courses by 2023.

  Effect on the work

  Transkribus has meaningfully accelerated archival transcription for trained historians while simultaneously creating a new pedagogical tool: assigning students to correct AI transcription errors against the original document teaches paleography and source criticism in ways that passive reading cannot. The tool did not reduce history faculty demand but did change the balance of skills taught in graduate methodology seminars.

• Generative AI2022–present

  Generative AI: ChatGPT, JSTOR AI Research Tool, NotebookLM (large language model era)

  ChatGPT's public release in November 2022 arrived in history departments as a dual-edged tool. On one side: a generation of students who could now generate a plausible 1,500-word essay on the causes of World War I in under two minutes, undermining the unobserved take-home essay as a valid assessment instrument. The AHA's 2024 member survey found 68.9% of historians had already redesigned courses to mitigate AI misuse, and 92.6% sought institutional guidance on AI policy. On the other side: JSTOR's AI Research Tool (launched July 2025) offered semantic literature search across 14,000+ participating institutions at no additional cost; NotebookLM allowed historians to synthesize uploaded primary source collections with inline source references; and Transkribus began integrating LLMs for natural language querying of transcribed texts. The historian's source-verification role was amplified, not diminished, by AI's presence.

  Effect on the work

  The AHA's 2025 Academic Jobs Report documents a 13% decline in full-time history faculty from the 2010 peak to 2025, with hundreds of job listings in 2024-25 representing the smallest number since the pandemic, and a decline from the 2008 peak. AI contributed to this structural contraction indirectly: declining history undergraduate enrollment (a decade before generative AI) is the primary driver, but AI's ability to perform many routine lecture-prep, Q&A, and bibliography tasks at zero marginal cost has accelerated the case for reducing history instructional headcount at budget-constrained institutions.