Historians
Scrub through 152years of this role's history, from when it first emerged, through every wave of technology that reshaped it, to the cited projections for where it's heading next.
The Institute of Historical Research at the University of London publishes "Doing Historical Research in the Age of Generative AI" (blog.history.ac.uk, May 2025), documenting that AI tools fabricate archival citations, invent historical events, and misquote primary sources with high fluency -- and that historians cannot safely delegate fact-checking to AI. The same year, a Lancet preprint (May 2026) documents a sixfold increase in AI-hallucinated citations in academic papers between 2023 and 2025. For a discipline whose entire scholarly credibility rests on accurate citation of verifiable primary sources, these findings are directly load-bearing: the AI hallucination failure mode amplifies, rather than reduces, the value of the trained historian who can distinguish a real archival document from a plausible fabrication.
The tools that defined the work
Select an era to see how it reshaped the work.
Archival seminar and Quellenkritik (Ranke's source-critical method, 1825)
Leopold von Ranke's historical seminar at the University of Berlin, begun in 1825, established the foundational technology of the historical profession: systematic Quellenkritik (source criticism). Students were trained to locate, authenticate, transcribe, and critically evaluate primary documents rather than simply reading previous authors. This was not a physical technology but a methodological one -- the seminar replaced the dilettante's library with the archive as the historian's workshop. Herbert Baxter Adams imported it to Johns Hopkins in 1880, where he trained the first generation of American professional historians in the same documentary method. The technique remains the core of historical training today, over two centuries after Ranke first taught it.
Work toolChanging equipment American Historical Review and peer-reviewed journal infrastructure (AHA era, 1884)
The founding of the American Historical Association in 1884 and the launch of the American Historical Review in 1895 established the peer-reviewed journal as the historian's primary communication technology. This was the mechanism by which the profession governed itself: claims had to survive expert review, and careers were made or ended by publication in recognized journals. The journal infrastructure created the citation economy that defines historical scholarship to this day -- and which now collides with AI's hallucination failure mode, since a discipline built on verified citations cannot tolerate fabricated ones.
Work toolChanging equipment Microfilm reader and photocopy machine (mass archival reproduction, 1960s)
The microfilm reader, widely adopted in research libraries through the 1950s and 1960s, was the first technology to allow historians to work with reproductions of archival documents held in distant repositories. Before microfilm, research required physical travel to the archive and hand-transcription of documents. Microfilm collections -- Foreign Office records, newspaper archives, papal registers, plantation ledgers -- allowed a scholar in Iowa to read documents held in London, Rome, or Havana. The photocopier (Xerox 914, 1959) amplified this: historians could produce working copies of archival documents for annotation. These technologies did not change what historians did, but they dramatically changed the geographic and financial constraints under which they did it.
Effect on the workMicrofilm and photocopying reduced the cost of archival research substantially, allowing historians at non-elite institutions to undertake large-scale primary source projects that would previously have required a research fellowship at a European archive. This contributed to the expansion of the profession in the 1960s and the democratization of historical research beyond the Ivy League.
Work toolChanging equipment CD-ROM databases and digitised newspaper archives (JSTOR 1995, ProQuest Historical Newspapers)
JSTOR launched in 1995 with the stated goal of providing electronic access to academic journals, beginning with backruns of 10 journals and expanding rapidly. ProQuest's Historical Newspapers product digitised the full runs of major US newspapers (New York Times from 1851, Chicago Tribune, Atlanta Constitution) and made keyword search across millions of pages available at research libraries. For historians, these tools transformed the literature review phase of research: finding every scholarly article citing a specific person, event, or concept went from a weeks-long manual index search to a minutes-long database query. The digitisation of newspapers was even more transformative for social and political historians, enabling research questions that required scanning thousands of issues.
Effect on the workDatabase search and digitised archives are estimated to have cut the literature review phase of a major historical project by 60-80% compared to the card-catalogue and printed index era. This productivity gain was not reflected in employment growth -- it freed time for deeper archival work rather than expanding the profession.
Work toolChanging equipment Digital archives and open-access primary sources (Google Books, Internet Archive, HathiTrust, FRASER)
Google Books launched its mass digitisation program in 2004 in partnership with major research libraries; HathiTrust Digital Library was established in 2008 to archive and provide access to the resulting corpus; the Internet Archive's Wayback Machine and digital text collections provided a parallel open-access repository. The Federal Reserve Bank of St. Louis's FRASER archive digitised decades of federal economic reports, BLS bulletins, and Congressional hearings. For historians, these platforms moved a substantial fraction of the pre-1923 (public domain) printed record online and searchable -- including newspapers, government documents, periodicals, and published diaries -- while FRASER opened up twentieth-century primary sources in economic and labor history that had previously required archival travel. The digital archive era did not replace the physical archive but dramatically expanded the domain of work that could be done remotely.
Work toolChanging equipment AI handwritten text recognition + LLM research tools (Transkribus, JSTOR AI, NotebookLM, Ithaca)
Transkribus (READ-COOP SCE), active in historical research communities since around 2018 and substantially upgraded in 2022-2025, applies trained handwritten text recognition (HTR) models to archival manuscript corpora -- turning weeks of paleographic transcription into hours of expert review and correction at a Character Error Rate of 5-10% on trained models. JSTOR's AI Research Tool, launched July 2025, enables semantic search across 14,000+ institutional subscriptions without the hallucination risk of general-purpose LLMs. Google's NotebookLM allows historians to upload digitised primary source collections and query them with source-linked summaries. DeepMind's Ithaca model (Nature, 2022) achieves 62% accuracy on ancient Greek inscription restoration versus 25% for unaided expert epigraphers. These tools represent the first generation of AI that materially augments archival productivity rather than simply speeding up search. The structural constraint remains: AI tools fabricate archival citations with high fluency (Stat News/Lancet, May 2026: AI-hallucinated citations in academic papers increased sixfold between 2023 and 2025), which means the historian's source-verification role is amplified rather than replaced.
Effect on the workThe Institute of Historical Research (May 2025) explicitly documents that AI tools fabricate archival citations, invent historical events, and misquote primary sources -- and that historians cannot delegate fact-checking to AI. The professional consensus is that AI augments productivity in transcription, literature triage, and corpus synthesis while structurally amplifying the value of the human expert who can authenticate, contextualize, and verify what the AI produces.
AI audit toolsPattern detection
What credible sources project
Scrub the slider past now to anchor each scenario on the scrubber. The spread is the range of futures credible sources project for this role.
What's shifting in the work right now
The historical view above shows how this role has moved. This is the present-day detail: which AI tools are picking up which tasks, where the edge still is, and the natural directions this work can grow.
What's changing in your day
Three parts of your work where AI is already doing real lifting, and what stays yours.
AI is sitting alongside you hereConduct systematic literature review and secondary source triage for book-length research projects and grant proposals — using JSTOR AI Research Tool for semantic discovery of secondary literature across 14,000+ participating institutions (no hallucinated citations
Conduct systematic literature review and secondary source triage for book-length research projects and grant proposals — using JSTOR AI Research Tool for semantic discovery of secondary literature across 14,000+ participating institutions (no hallucinated citations; searches real JSTOR corpus only), and Elicit for structured extraction of argument, methodology, and evidence type from 30-50 papers simultaneously — then independently assessing every AI-surfaced source for scholarly credibility, verifying that citations are real and accurately represent the cited work, and constructing a historiographical argument from verified secondary evidence.[10],[11]
Use JSTOR AI Research Tool to run a semantic search on, say, "Atlantic slavery and capitalism 1820-1860" — it will surface relevant articles from the JSTOR corpus by conceptual relevance rather than keyword matching, and generate short summaries that let you judge whether a paper is worth reading before opening the full text. Available at no added cost at 14,000+ JSTOR-participating institutions since July 2025; was prominently featured at AHA Annual Meeting January 2025. Use Elicit to extract structured data (central argument, primary vs. secondary evidence used, methodology, time period) from a batch of 40 articles you are deciding whether to cite — it compresses a three-hour bibliography sweep into 30 minutes of review. Critical safeguard: every AI-surfaced citation must be independently verified as a real publication before you cite it. The Stat News / Lancet analysis (May 2026) documents a sixfold increase in AI-hallucinated academic citations between 2023 and 2025. JSTOR AI is safer than general-purpose LLMs for this task because it only retrieves actual JSTOR content — but even there, generated summaries require verification against the full text before quoting.
AI is sitting alongside you herePrepare historical reports, organizational histories, and policy-relevant historical analyses for government agencies, federal departments, think tanks, and organizations with institutional memory mandates — using AI tools to draft initial structure and background sections from supplied notes and verified data, then applying expert historical judgment to verify every factual claim, anchor every assertion to documented primary evidence, and ensure the narrative accurately represents the historical record without the anachronisms, inventions, or hallucinated citations that AI tools routinely introduce.
Prepare historical reports, organizational histories, and policy-relevant historical analyses for government agencies, federal departments, think tanks, and organizations with institutional memory mandates — using AI tools to draft initial structure and background sections from supplied notes and verified data, then applying expert historical judgment to verify every factual claim, anchor every assertion to documented primary evidence, and ensure the narrative accurately represents the historical record without the anachronisms, inventions, or hallucinated citations that AI tools routinely introduce.[6],[5]
Federal government historians (National Archives, State Department, Department of Defense, NASA, and agency historical offices) produce the authoritative institutional record — accuracy is a legal and policy requirement, not a stylistic preference. AI can draft background sections and section structures from your notes and outlines, and can accelerate the writing phase for routine descriptive sections. The non-delegatable work: every factual claim in a government historical report must be cited to a verified, accessible primary source; every date, name, and organizational fact must be independently confirmed; and the interpretive framing of contested historical events must reflect the actual state of the evidence, not an AI's training data. The Stat News / Lancet research (May 2026) shows that AI hallucinated citations are now appearing in academic papers at a rate of 1 in 277 — in a government historical report, a fabricated source is not just a scholarly embarrassment; it is a records-integrity failure. Use AI for structure and drafting velocity; retain independent verification of every empirical claim.
AI is sitting alongside you hereBuild and manage digital history projects and public-facing online archives — using Omeka or Omeka S for collection organization, metadata generation, and digital exhibit design
Build and manage digital history projects and public-facing online archives — using Omeka or Omeka S for collection organization, metadata generation, and digital exhibit design; applying AI-assisted Dublin Core metadata tagging for batch-uploaded historical photographs, documents, and objects; designing interpretive narratives for digital exhibitions that translate archival research for general public audiences without the jargon of academic historiography; ensuring that all metadata accurately represents the provenance and content of original source materials.[12],[6]
Omeka (Roy Rosenzweig Center for History and New Media) is the dominant platform for public historians building digital collection websites, virtual archives, and online exhibitions — deployed by state humanities councils, historical societies, university special collections, and local history organizations. AI tools accelerate two specific Omeka workflows: (1) batch metadata generation — use ChatGPT to draft Dublin Core descriptions for a collection of 200 digitised Civil War photographs given a template and a set of supplied facts (date range, donor provenance, geographic region), then review and correct each entry against the actual image; this is significantly faster than writing 200 descriptions from scratch; (2) interpretive narrative drafting — use ChatGPT or NotebookLM to generate a first-draft exhibition essay from notes you have written, then rewrite it for voice, accuracy, and audience appropriateness. The critical constraint: every factual claim in the metadata and exhibit copy must be independently verified against primary sources before publication — AI-generated exhibit labels that hallucinate facts about historical objects or people are a professional and institutional reputation risk.
Where this role is heading
Natural next steps for someone with your foundation: not exits, evolutions.
History Teachers, Postsecondary
Historians with active research programs and publication records — particularly those holding PhDs and affiliated with academic institutions or research centers — are the natural pipeline for postsecondary history faculty positions. The core competencies (archival research, scholarly writing, historiographical argument) are identical; the adjustment is adding teaching and student advising to a research identity rather than substituting one for the other. The CRI delta is slightly negative because postsecondary history teaching has its own AI exposure in the lecture prep, bibliography generation, and essay assessment domains — but the human-advantage moat from archival expertise and primary-source interpretation transfers directly. The academic job market for tenure-track historians is highly constrained (MLA / AHA job reports document continued decline), so this pivot is less accessible than its Low difficulty implies — the barrier is supply-side, not skill-based.
- · Course design and syllabus construction: building a 14-week undergraduate or graduate course with primary-source assignments, AI-resilient assessments, and scaffolded writing instruction
- · Teaching methods for historical thinking: facilitating document analysis exercises, Socratic seminars, oral history projects, and paleography labs that build evidence-based interpretation skills
- · Student advising at undergraduate and graduate levels: guiding thesis topic selection, archive identification, IRB compliance for oral history, and academic career strategy
- · AI-resilient assessment design: replacing unobserved take-home essays with primary-source critique workshops, Transkribus paleography labs, in-class analytical writing on unseen document excerpts
- · Academic departmental governance: curriculum committee participation, promotion and tenure review, faculty hiring processes, and AHA disciplinary compliance expectations
See the same long-arc view for your own profession.
Browse the directory by industry, or search by title or SOC code. New roles ship every few weeks. Every profile cites every claim.
Browse all roles