Historians

19-3093.00
Median wage $76,750/yr3,450 employed (US)Rank #248 of 923 scored · top 27% by substitution

Research, analyze, record, and interpret the past as recorded in sources, such as government and institutional records, newspapers and other periodicals, photographs, interviews, films, electronic media, and unpublished manuscripts, such as personal diaries and letters.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution36
Exposure32
Augmentation69

Substitution — the headline: capability discounted by cost, barriers and adoption.

Exposure — technical capability alone, regardless of whether anyone deploys it.

Augmentation — how much AI assists without replacing. High here + moderate substitution = a changing job, not a disappearing one.

Tasks on the substitution scale

21 rated tasks, binned by substitution score.

Position among all scored occupations

Distribution of 923 occupation scores; the marker is this occupation.

Tasks with substitution ≥ 70

5%

Run 1.0.0-draft.1 · computed 2026-08-05 · rater panel: claude-sonnet-5, claude-haiku-4-5-20251001 · intervals span rater disagreement.

Why this score

The five weighted dimensions of the composite, averaged across this occupation's tasks (importance-weighted, panel mean). Exact weights and formulas: /api/v1/methodology.

Task automatabilityw 35%34

panel mean rating 2.4/5 → substitution pressure 34/100

Technical feasibility todayw 20%30

panel mean rating 2.2/5 → substitution pressure 30/100

Cost vs. human wagew 15%38

panel mean rating 2.5/5 → substitution pressure 38/100

Adoption barriersw 20%inverted — strong barriers lower the score49

panel mean rating 3.0/5 (barrier strength) → substitution pressure 49/100

Sector adoption velocityw 10%27

panel mean rating 2.1/5 → substitution pressure 27/100

Task breakdown (21 tasks)

Substitution pressure per task, weighted by O*NET importance in the composite. Expand a task for the full rater audit trail — every rating, every model, every rationale.

Translate or request translation of reference materials.

82

CI 76–87 · exposure 83 · augmentation 88 · importance 2.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Translation via AI is widely piloted and adopted in information-intensive sectors (academia, libraries, archives), but many institutions still maintain gatekeeping on critical materials; adoption is moderate rather than near-total.
Sector adoption velocityclaude-sonnet-54/5Academic and research settings have rapidly adopted machine translation tools as everyday aids, reflecting broader fast adoption in knowledge-work/information sectors.
Augmentation potentialclaude-haiku-4-5-202510014/5AI translation tools directly augment historian productivity by rapidly producing readable drafts of reference materials, allowing historians to focus on interpretation and synthesis rather than manual translation work.
Augmentation potentialclaude-sonnet-55/5AI translation dramatically speeds up historians' ability to access foreign-language sources while they retain judgment over interpretation, context, and accuracy verification.
Task automatabilityclaude-haiku-4-5-202510014/5Current AI systems (GPT-4, Claude, specialized translation engines) can reliably translate reference materials across major languages with 50%+ time saving, though domain-specific or archaic texts may require human review and light post-editing.
Task automatabilityclaude-sonnet-55/5Machine translation of reference materials, including many historical languages, is a well-established capability that meets or exceeds the 50% time-saving bar for most use cases.
Adoption barriersclaude-haiku-4-5-202510012/5While some institutions prefer human translators and quality verification is standard practice, there is no legal requirement that a licensed translator must perform the translation itself, creating minimal hard barriers to substitution.
Adoption barriersclaude-sonnet-52/5No licensing requirement for translating reference materials for research purposes; some friction exists if certified/legal translation is needed, but historians' research use rarely requires this.
Cost vs. human wageclaude-haiku-4-5-202510015/5AI translation (API inference + integration) costs pennies per task, while hiring professional human translators costs tens to hundreds of dollars per page; order-of-magnitude gap favors AI.
Cost vs. human wageclaude-sonnet-55/5AI translation costs pennies per document versus professional human translator fees, an order of magnitude or more cheaper for bulk reference material.
Technical feasibility todayclaude-haiku-4-5-202510014/5Mature commercial translation products (Google Translate, DeepL, Claude) perform this task reliably in production for modern-language materials; niche historical or very specialized sources see higher error rates but the general task is deployable at scale.
Technical feasibility todayclaude-sonnet-54/5Deployed products like Google Translate, DeepL, and LLM-based translators are used in production widely, though accuracy on archaic language, dialects, or poor-quality scans still requires human review.

Organize information for publication and for other means of dissemination, such as via storage media or the Internet.

63

CI 59–67 · exposure 58 · augmentation 88 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Museums, universities, and digital humanities projects are increasingly adopting tools for content management and metadata organization, though adoption is still uneven. Some institutions use AI-assisted workflows while others rely on traditional manual methods, reflecting middling penetration in a sector with varied digitization readiness.
Sector adoption velocityclaude-sonnet-53/5Academic and publishing sectors show moderate AI adoption for content organization and drafting, with pilots and partial integration common but full-scale production deployment uneven across humanities institutions.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can substantially improve a historian's productivity by rapidly structuring research, suggesting organizational hierarchies, generating metadata, and adapting content for different dissemination formats while the historian retains final authority over historical accuracy and presentation choices.
Augmentation potentialclaude-sonnet-55/5AI tools significantly speed up organizing notes, structuring drafts, formatting citations, and preparing content for various dissemination formats, greatly boosting historian productivity while they retain interpretive and editorial control.
Task automatabilityclaude-haiku-4-5-202510013/5AI can assist significantly with organizing, structuring, and formatting historical information for publication—tasks like converting research notes into outlined documents, suggesting organizational schemas, and preparing metadata. However, the task requires judgment about what information is historically significant and appropriate for each dissemination channel, which typically requires human expertise and decision-making.
Task automatabilityclaude-sonnet-54/5Organizing, structuring, and formatting historical information for publication or digital dissemination is largely a text/data-processing task that current LLMs and tools handle well, including drafting outlines, formatting citations, and preparing content for web/CMS platforms.atosphere requires human review for accuracy but the bulk of organizational labor can be offloaded.
Adoption barriersclaude-haiku-4-5-202510012/5Few regulatory or licensing barriers prevent the use of AI to organize historical information for publication. The main friction comes from historians' preferences to retain control over organization decisions and institutional requirements for human review of content before public dissemination.
Adoption barriersclaude-sonnet-52/5No licensing requirement governs this specific organizational task, though institutional/editorial standards and academic norms create some friction requiring human oversight for accuracy and scholarly integrity.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-assisted organization of documents and media is inexpensive compared to paying a historian or skilled research assistant to perform manual organization and formatting work. The cost of inference and integration is modest relative to the loaded wage of a professional organizer or historian doing this task.
Cost vs. human wageclaude-sonnet-54/5AI-assisted organization and formatting tools are inexpensive compared to the hourly cost of a trained historian, especially for repetitive structuring and formatting tasks.
Technical feasibility todayclaude-haiku-4-5-202510013/5Products exist that organize and structure text content (content management systems, generative AI tools for outlining and formatting), and they perform these functions reliably in some contexts. However, they lack the contextual understanding of historical significance and nuance needed to organize complex research for scholarly publication without substantial human oversight.
Technical feasibility todayclaude-sonnet-53/5Products like reference managers, AI writing assistants, and CMS tools with AI features exist and are used in publishing workflows, but full end-to-end organization of scholarly historical content still requires significant human curation and fact-checking.

Edit historical society publications.

58

CI 51–65 · exposure 58 · augmentation 75 · importance 2.9/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Academic and historical publishing sectors move conservatively on AI integration; most historical societies are small, non-profit, digitization-challenged organizations with slow technology adoption and strong attachment to human expertise.
Sector adoption velocityclaude-sonnet-52/5Historical societies and academic/nonprofit publishing are typically small, under-resourced, and slow-adopting relative to tech-forward sectors.
Augmentation potentialclaude-haiku-4-5-202510014/5AI significantly assists human editors by automating routine copyediting, formatting, and citation checking, allowing historians to focus on substantive historical accuracy and interpretation rather than mechanical tasks.
Augmentation potentialclaude-sonnet-54/5AI tools substantially speed up copyediting, formatting, and drafting suggestions for historians while they retain control over factual and interpretive accuracy.
Task automatabilityclaude-haiku-4-5-202510014/5AI can handle much of the copyediting, fact-checking against common databases, citation formatting, and structural corrections, though nuanced historical interpretation and judgment about historical significance still require human review. This combination likely achieves >50% time savings for standardized publication workflows.
Task automatabilityclaude-sonnet-53/5AI can draft, copyedit, and check style/grammar for publications, but historical accuracy verification, source evaluation, and nuanced editorial judgment still require human review, capping full end-to-end automation.'
Adoption barriersclaude-haiku-4-5-202510013/5Historical societies often prioritize reputation and accuracy, creating organizational friction and reluctance to fully delegate editing to AI; professional editorial standards and peer review traditions also create meaningful oversight requirements rather than hard legal barriers.
Adoption barriersclaude-sonnet-52/5No licensing requirement for editing historical society publications, though reputational/institutional expectations of scholarly rigor create moderate friction to full automation.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI inference and integration costs for copyediting and fact-checking are substantially lower than paying a human editor per hour, especially for routine passes, though final review by qualified historians remains necessary.
Cost vs. human wageclaude-sonnet-54/5AI copyediting and drafting assistance is far cheaper per word than a historian's time for routine editing tasks, though human oversight for accuracy adds some cost back.
Technical feasibility todayclaude-haiku-4-5-202510013/5Mature AI writing assistants and fact-checkers exist and are deployed, but they require careful oversight to avoid introducing errors into historical content and perform inconsistently on specialized historical references and archival context.
Technical feasibility todayclaude-sonnet-53/5Grammar/style/copyediting tools (Grammarly, ChatGPT, Word AI) are widely deployed and reliable, but no product reliably handles historical fact-checking or scholarly editorial judgment at production quality.

Present historical accounts in terms of individuals or social, ethnic, political, economic, or geographic groupings.

53

CI 39–67 · exposure 45 · augmentation 88 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Museums, educational platforms, and digital humanities projects are piloting AI-assisted narrative generation and curatorial work, but widespread production adoption remains limited. The cultural and epistemological weight of historical authority slows displacement even as tools improve.
Sector adoption velocityclaude-sonnet-52/5Academic history and humanities scholarship remain slow adopters of AI tools relative to fields like finance or software, with AI mainly used for research assistance rather than core interpretive work.
Augmentation potentialclaude-haiku-4-5-202510015/5AI significantly augments historians by rapidly drafting alternative framings, organizing large source corpora by social/ethnic/political dimensions, and generating preliminary narratives for expert editing and refinement. This remains high-value assistance even where human historians retain final interpretive authority.
Augmentation potentialclaude-sonnet-54/5AI can significantly assist historians by summarizing sources, drafting outlines, suggesting comparative frameworks, and helping organize accounts across different social or geographic groupings.
Task automatabilityclaude-haiku-4-5-202510014/5AI systems can readily synthesize historical narratives, organize accounts by demographic/geographic groupings, and generate coherent, well-structured presentations of historical data from source material. However, the task still requires human judgment on framing, emphasis, and interpretive depth, preventing a full 5-rating despite meeting the 50% time-savings threshold.
Task automatabilityclaude-sonnet-52/5AI can draft narrative summaries and synthesize sources, but constructing an original, well-argued historical account requiring interpretive judgment and novel synthesis is not fully automatable to equal quality today.a
Adoption barriersclaude-haiku-4-5-202510012/5Few legal or regulatory barriers exist to automating narrative presentation. Academic and professional credibility norms create friction—audiences may expect human authorship or verification—but no licensing requirement or legal mandate compels human historians to sign off on the output.
Adoption barriersclaude-sonnet-52/5No formal licensing is required to write history, though academic and publishing norms demand rigorous scholarship and attribution, creating moderate professional/reputational barriers to pure AI output.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI inference and integration costs are now orders of magnitude cheaper than a historian's fully-loaded labor (salary, benefits, overhead), though human review and editorial oversight still add meaningful cost. The ratio heavily favors AI for routine narrative synthesis.
Cost vs. human wageclaude-sonnet-53/5AI drafting is cheap per word, but the need for extensive fact-checking, archival verification, and expert revision narrows the cost advantage compared to a trained historian's output.
Technical feasibility todayclaude-haiku-4-5-202510013/5LLMs and document-synthesis tools can produce competent historical narratives and organize accounts by specified groupings in production settings, but material gaps remain in source verification, factual accuracy under scrutiny, and handling of contested historical interpretations. Products work well for straightforward synthesis but not reliably for nuanced scholarly work.
Technical feasibility todayclaude-sonnet-52/5LLM-based writing assistants can produce plausible historical narratives but are prone to fabrication, oversimplification, and lack of rigorous sourcing, so they are not reliably deployed for professional historical scholarship.

Research and prepare manuscripts in support of public programming and the development of exhibits at historic sites, museums, libraries, and archives.

47

CI 30–65 · exposure 45 · augmentation 88 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Museums, libraries, and archives are traditionally slow to adopt new technologies, and many lack dedicated IT infrastructure for AI integration. While some large institutions are piloting AI-assisted research and curation, widespread production deployment remains limited; the sector ranks below information and finance industries in adoption speed.
Sector adoption velocityclaude-sonnet-52/5Museums, libraries, and archives are historically slow-adopting, under-resourced sectors with limited AI integration beyond experimental pilots.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically augments historian productivity: automated literature discovery, real-time fact-checking against sources, rapid manuscript drafting, and exhibit narrative generation all keep the human historian in the loop while freeing cognitive capacity for interpretation, curation, and scholarly judgment. This is a near-textbook augmentation case.
Augmentation potentialclaude-sonnet-54/5AI is genuinely useful for literature review synthesis, drafting outlines, and generating first-pass text that historians then verify and refine, meaningfully speeding up manuscript preparation.
Task automatabilityclaude-haiku-4-5-202510014/5Current AI systems can automatically perform substantial portions of this task: literature search and synthesis, manuscript drafting from primary sources, exhibit narrative generation, and archival metadata organization. A human historian would still need to verify interpretations, make curatorial judgments, and ensure accuracy, but AI can handle 60–70% of the underlying work at quality comparable to junior researchers, meeting the ≥50% time-saving bar.
Task automatabilityclaude-sonnet-52/5AI can draft text and synthesize research summaries, but original archival research, source verification, and curatorial judgment for exhibit narratives require deep human expertise that current AI cannot reliably replace end-to-end.
Adoption barriersclaude-haiku-4-5-202510012/5Limited hard barriers exist: there is no licensing requirement for historians to conduct research or write manuscripts, though institutional curatorial review and scholarly norms create some friction. Adoption barriers are primarily organizational (preference for human expertise, institutional inertia) rather than legal or regulatory.
Adoption barriersclaude-sonnet-53/5No licensing requirement exists, but institutional standards for scholarly accuracy, provenance, and public trust in museums/archives create meaningful friction against unsupervised AI-generated content.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI inference and oversight costs are substantially lower than employing a full-time historian or research associate ($50k–$80k annually in loaded cost). Even accounting for human review and integration overhead, AI-augmented research pipelines cost 3–5× less per manuscript or exhibit section than hiring dedicated staff.
Cost vs. human wageclaude-sonnet-52/5While AI drafting is cheap, the human archival research, fact-checking, and curatorial oversight still dominate cost, keeping overall savings modest compared to a human historian's full workflow.
Technical feasibility todayclaude-haiku-4-5-202510013/5Deployed products (GPT-4, Claude, specialized research tools, and RAG systems) can draft and organize content reliably for portions of this task, but error rates on historical accuracy and source interpretation remain material. No single production system fully automates the end-to-end manuscript and exhibit development; most museums and archives rely on AI as assistive tools rather than standalone performers.
Technical feasibility todayclaude-sonnet-52/5AI writing tools are used for drafting assistance in museum and archival contexts, but no deployed product independently researches and produces exhibit-ready manuscripts at production scale.

Research the history of a particular country or region, or of a specific time period.

42

CI 25–59 · exposure 38 · augmentation 75 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Historical research occurs primarily in academic, museum, and heritage sectors with relatively slow technology adoption. Most historians still rely on traditional research methods and archives; AI-assisted research remains marginal in production despite pilot projects.
Sector adoption velocityclaude-sonnet-53/5Academia and publishing are moderate adopters of AI tools for research assistance, with growing use of AI search and summarization but still cautious, uneven institutional uptake in humanities scholarship.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully assist by finding relevant sources, translating documents, generating outlines, or flagging chronological inconsistencies, raising a historian's productivity without replacing their interpretive role. This augmentation is meaningful but not transformative for the discipline.
Augmentation potentialclaude-sonnet-55/5AI substantially speeds up literature discovery, summarization of secondary sources, translation, and preliminary synthesis, making it a strong productivity multiplier while the historian retains interpretive control.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist with locating and summarizing primary/secondary sources, but historical research critically depends on human judgment about interpretation, significance, and synthesis across conflicting accounts. End-to-end automation with 50% time savings at equal quality is not achievable today.
Task automatabilityclaude-sonnet-53/5AI can rapidly synthesize secondary sources, summarize known historical narratives, and surface leads, but genuine historical research requires archival work, primary source evaluation, and original interpretation that current AI cannot reliably perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Academic and institutional norms strongly prefer human historians for original research; peer review, publication standards, and reputation mechanisms require human accountability. Funding bodies and employers expect credentialed experts, creating friction against full automation.
Adoption barriersclaude-sonnet-52/5No licensing requirement governs historical research, but institutional expectations of scholarly rigor, citation accuracy, and peer review create moderate friction against pure AI output being accepted as authoritative.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI tools for document retrieval and summarization are cheap, but the core value—expert historical judgment and synthesis—still requires a trained historian. The all-in cost (AI + human oversight) remains comparable to or higher than a historian working directly.
Cost vs. human wageclaude-sonnet-54/5For the literature-review and synthesis portions of research, AI-assisted querying is dramatically cheaper than paying a historian's hourly rate, though archival and primary-source work still requires human cost.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably performs original historical research independently. AI tools can draft literature reviews or summarize documents, but historians in production do not rely on AI to conduct research autonomously; human verification and judgment remain essential.
Technical feasibility todayclaude-sonnet-53/5Deployed LLM-based research assistants and search tools reliably summarize and retrieve historical information, but they still hallucinate facts, misattribute sources, and cannot access many archives, so professional-grade reliability is inconsistent.

Collect detailed information on individuals for use in biographies.

39

CI 25–52 · exposure 38 · augmentation 75 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Historians and academic institutions adopt AI slowly; the field prioritizes methodological rigor and source authentication over efficiency. Information work in academia lags commercial sectors in automation adoption.
Sector adoption velocityclaude-sonnet-52/5Historical research and academic biography writing remain a low-digitization, slow-adopting niche compared to fast-moving sectors like finance or software.
Augmentation potentialclaude-haiku-4-5-202510014/5AI substantially assists biographical research through database search, citation finding, text summarization, and preliminary organization of materials. Historians actively use AI tools today to accelerate information collection while maintaining critical evaluation and final curation.
Augmentation potentialclaude-sonnet-54/5AI substantially speeds up preliminary research, cross-referencing records, and organizing information, letting historians focus on interpretation and verification.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist with information gathering (e.g., text extraction, summarization from documents) but cannot reliably conduct the nuanced, contextual research required for biographical detail collection. The task demands judgment about source credibility, relevance, and completeness—decisions requiring human expertise and verification.
Task automatabilityclaude-sonnet-53/5AI can quickly aggregate publicly available information and search archives, but verifying accuracy, sourcing primary documents, and synthesizing nuanced biographical detail still requires substantial human effort.'
Adoption barriersclaude-haiku-4-5-202510014/5Biographical research and publication carry significant liability exposure for accuracy and attribution. Historians face reputational and professional accountability for sources and claims; legal and ethical standards governing biographical work create friction against full automation.
Adoption barriersclaude-sonnet-52/5No licensing requirement exists, though historians' professional standards for source verification and academic credibility create some friction against pure AI substitution.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI research assistants remain expensive relative to junior researchers, and biographical work requires expert oversight that negates cost savings. The integration and verification costs are substantial for production-quality biography research.
Cost vs. human wageclaude-sonnet-53/5AI search and summarization tools are cheap for initial information gathering, but the need for human verification and archival research keeps overall costs comparable to skilled human labor for rigorous work.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI tools can search databases and aggregate text, no deployed product reliably performs end-to-end biographical research collection independently. Current systems lack the contextual judgment and cross-source validation that historians require; they serve as research aids, not autonomous collectors.
Technical feasibility todayclaude-sonnet-53/5Deployed AI research and search tools can compile background information reliably, but they frequently hallucinate details or miss archival/primary sources critical to serious biographical work.

Conserve and preserve manuscripts, records, and other artifacts.

38

CI 5–71 · exposure 41 · augmentation 50 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Larger cultural institutions and government archives have begun pilot programs with digital preservation AI, but adoption remains uneven and largely in early stages. Smaller historical societies and libraries lag significantly, and many still rely on manual methods.
Sector adoption velocityclaude-sonnet-51/5Archival and conservation work in museums, libraries, and cultural institutions is a low-digitization, physically-oriented field with minimal AI adoption for the hands-on preservation work itself.
Augmentation potentialclaude-haiku-4-5-202510014/5AI significantly assists historians and archivists by automating routine scanning, enhancing image quality, generating metadata, and flagging condition issues, freeing experts to focus on conservation decisions and interpretation. These tools demonstrably raise productivity while keeping the human expert in control.
Augmentation potentialclaude-sonnet-52/5AI can assist with documentation, condition assessment via image analysis, or research on preservation techniques, but offers little help with the physical conservation act itself.
Task automatabilityclaude-haiku-4-5-202510015/5Digital preservation, metadata tagging, condition assessment, and archival cataloging can be largely automated with current AI systems (OCR, computer vision, document classification), achieving substantial time savings. However, certain expert conservation decisions still require human judgment, so end-to-end 50%-time-saving thresholds are readily met for the routine preservation and documentation components.
Task automatabilityclaude-sonnet-51/5Physical conservation of manuscripts and artifacts requires hands-on manual skill, chemical treatment, and material handling that current AI systems cannot perform; this is a physical craft task, not an information task.
Adoption barriersclaude-haiku-4-5-202510013/5Conservation and preservation of artifacts of cultural or legal significance often have regulatory and institutional oversight requirements, and there is material organizational friction around trusting AI with irreplaceable materials. However, no strict licensing barrier prevents automation; institutions can and do implement AI tools alongside human curators.
Adoption barriersclaude-sonnet-54/5Conservation of valuable historical artifacts typically requires trained, often certified conservators due to liability, irreversible damage risk, and institutional standards protecting rare materials.
Cost vs. human wageclaude-haiku-4-5-202510014/5Automated digital scanning, OCR, and AI-assisted cataloging are substantially cheaper than manual digitization and archival labor once systems are implemented, though initial setup and integration costs are significant. Ongoing operational costs for AI-assisted preservation typically run well below the loaded cost of specialized archivists.
Cost vs. human wageclaude-sonnet-51/5AI cannot substitute for the physical labor and specialized manual expertise involved, so there is no cost comparison to make—the human remains the only viable option.
Technical feasibility todayclaude-haiku-4-5-202510013/5AI products for digital preservation (OCR, image enhancement, metadata extraction) and condition assessment exist and are in use in some institutions, but error rates on complex or damaged materials remain non-trivial, and integration with archival workflows requires customization. Production deployment is inconsistent across the sector.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs physical preservation or conservation work; this remains entirely a domain of specialized human conservators using manual and chemical techniques.

Gather historical data from sources such as archives, court records, diaries, news files, and photographs, as well as from books, pamphlets, and periodicals.

36

CI 34–39 · exposure 25 · augmentation 75 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Academic and museum sectors adopt digitization and search tools slowly; most archival work remains manual or semi-automated, and resistance to replacing historian expertise with automated pipelines is strong. Broader adoption remains in pilot phases.
Sector adoption velocityclaude-sonnet-52/5Digital humanities and archival science are adopting AI tools slowly; academic and library sectors are moderate-to-slow adopters compared to fast-moving tech/finance sectors.
Augmentation potentialclaude-haiku-4-5-202510014/5AI-powered transcription, OCR, tagging, and search tools substantially assist historians in locating and organizing sources, allowing them to focus on interpretation and synthesis. These augmentations materially accelerate source gathering while historians retain full analytical control.
Augmentation potentialclaude-sonnet-54/5AI substantially speeds up searching digitized archives, OCR/transcription of documents, and summarizing large text corpora, meaningfully aiding historians while they retain interpretive and verification roles.
Task automatabilityclaude-haiku-4-5-202510012/5Gathering data from structured sources like digitized archives and databases can be partially automated, but historians must interpret context, significance, and reliability—tasks requiring human judgment that current AI struggles with at scale. Across heterogeneous sources (photographs, diaries, court records), document discovery and relevance assessment remain largely manual.
Task automatabilityclaude-sonnet-52/5AI can search digitized text and summarize documents, but locating, accessing, and verifying provenance of archival materials (many undigitized, handwritten, or physically dispersed) still requires substantial human effort and judgment.
Adoption barriersclaude-haiku-4-5-202510013/5Archives often have access restrictions, authentication requirements, and institutional policies governing data use; historians rely on specialized domain knowledge and institutional credentials to access materials. These frictions slow but do not prevent automation.
Adoption barriersclaude-sonnet-52/5No licensing requirement, but archives often require in-person access, permissions, or physical handling protocols, and scholarly norms demand human verification of primary sources.
Cost vs. human wageclaude-haiku-4-5-202510013/5OCR and basic text extraction are cheap compared to manual transcription, but integrating results across many source types and verifying accuracy still requires substantial human oversight, keeping all-in costs roughly on par with human research assistants.
Cost vs. human wageclaude-sonnet-53/5For digitized text search and summarization AI is cheaper, but physical archive visits, handwriting transcription, and verification still require paid human labor, keeping overall cost comparable rather than dramatically lower.
Technical feasibility todayclaude-haiku-4-5-202510012/5Tools exist for OCR, text extraction, and limited archival search, but no deployed system reliably gathers and contextualizes historical data across disparate source types with historian-level quality. Most production systems handle narrow, pre-curated collections rather than open historical discovery.
Technical feasibility todayclaude-sonnet-52/5Products like AI-assisted archival search and OCR/transcription tools exist and are used by researchers, but they handle only digitized subsets and struggle with handwriting, context, and reliability for rigorous historical sourcing.

Trace historical development in a particular field, such as social, cultural, political, or diplomatic history.

34

CI 30–39 · exposure 25 · augmentation 63 · importance 3.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5History is a low-digitization, humanities-heavy field with slow AI adoption. Historians remain skeptical of automated interpretation; institutions prize human expertise and face little commercial pressure to automate. Adoption remains limited to pilots and supplementary tools, not production displacement.
Sector adoption velocityclaude-sonnet-52/5Academic history is a low-digitization, slow-adopting field with cultural resistance to AI-generated scholarship, though some tools are used for literature review assistance.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist historians by accelerating source discovery, generating initial summaries of archival materials, and identifying temporal or topical patterns across documents. However, the historian must validate, contextualize, and synthesize these outputs, so productivity gains are moderate rather than transformative.
Augmentation potentialclaude-sonnet-54/5AI is genuinely useful for summarizing secondary literature, suggesting research directions, drafting outlines, and organizing timelines, substantially aiding historians while they retain interpretive control.
Task automatabilityclaude-haiku-4-5-202510012/5Tracing historical development requires synthesizing fragmented, heterogeneous sources; identifying causal chains; and making interpretive judgments about significance. While AI can retrieve and summarize sources, it cannot reliably construct a coherent historical narrative with defensible causal reasoning or original insight at the level required for scholarly work. A human historian must remain the primary agent.
Task automatabilityclaude-sonnet-52/5AI can synthesize existing secondary sources and draft narrative summaries, but rigorous historical tracing requires original archival research, primary source verification, and interpretive judgment that current AI cannot reliably perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510013/5Academic and professional norms strongly favor human expertise and original scholarship; institutional prestige and peer review create friction against outsourcing core analysis. However, there are no hard legal or licensing barriers; institutions can adopt AI-assisted tools if willing to accept quality tradeoffs and reputational risk.
Adoption barriersclaude-sonnet-52/5No licensing requirement exists, but academic credibility, peer review norms, and citation/plagiarism standards create institutional friction against pure AI authorship.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI inference and integration costs for source retrieval and analysis are modest, but oversight is high because historians must verify sources, evaluate interpretations, and synthesize arguments. The all-in cost remains comparable to or higher than hiring a research assistant or junior historian for the same output quality.
Cost vs. human wageclaude-sonnet-53/5AI drafting is cheap per word, but the necessary fact-checking, archival verification, and domain expertise to make output usable still requires significant paid historian time, narrowing the cost advantage.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably performs end-to-end historical narrative construction and causal analysis. Current AI systems can assist with source retrieval and summarization, but they lack the contextual depth, source criticism, and interpretive judgment needed for rigorous historical work. Production use is limited to narrow, well-structured tasks within the broader workflow.
Technical feasibility todayclaude-sonnet-52/5LLM products can produce plausible-sounding historical narratives but are prone to fabricating citations and details, and no deployed product is trusted for scholarly historical research without heavy verification.

Organize data, and analyze and interpret its authenticity and relative significance.

33

CI 30–35 · exposure 25 · augmentation 63 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Historical institutions and academia are relatively slow adopters of automation; most archives and history departments use AI only for data preprocessing and search augmentation, not for authenticity assessment or significance judgment. Adoption remains limited to assisting human experts rather than replacing their core interpretive role.
Sector adoption velocityclaude-sonnet-52/5Academic history and humanities research adopt AI tools slowly and selectively, mostly for auxiliary tasks like transcription or literature search rather than core interpretive work.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist historians by organizing large datasets, flagging patterns, suggesting related sources, and automating transcription and metadata tagging, raising efficiency on preparatory work. However, augmentation is strongest for data management; interpretation of authenticity and significance remains primarily human-driven.
Augmentation potentialclaude-sonnet-54/5AI tools significantly help with organizing large datasets, searching archives, transcribing documents, and drafting preliminary analyses, meaningfully boosting historian productivity even though final judgment remains human.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist with data organization and pattern recognition, the core task—analyzing authenticity and judging relative significance of historical evidence—requires human expertise, contextual judgment, and domain knowledge that current systems lack. Current AI cannot reliably assess source credibility or make nuanced significance judgments at the quality expected in historical scholarship.
Task automatabilityclaude-sonnet-52/5AI can help organize and summarize data, but authenticating sources and judging their historical significance requires nuanced domain expertise and contextual judgment that current systems cannot reliably replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510013/5Academic and institutional standards require human historians to validate findings and stand behind interpretations; there is no legal licensing barrier but strong professional and reputational friction against full automation. Museums, archives, and universities maintain institutional preferences for human expert judgment on significance.
Adoption barriersclaude-sonnet-52/5There's no formal licensing requirement, but professional and academic norms demand human expert judgment and accountability for authenticity claims, creating moderate institutional friction.
Cost vs. human wageclaude-haiku-4-5-202510012/5Integration of AI tools for data management may save some overhead, but the labor cost of human historians to review, correct, and oversee AI outputs often matches or exceeds the cost of direct human analysis. Oversight remains substantial because errors in authenticity assessment carry high reputational cost.
Cost vs. human wageclaude-sonnet-52/5While AI-assisted search and organization can reduce some labor costs, the interpretive and verification work still requires expert human oversight, keeping overall costs comparable to or only modestly below human-only work.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably performs end-to-end historical analysis and interpretation of authenticity at production scale. AI tools can organize datasets and flag patterns, but deployed systems lack the nuanced judgment needed to validate sources or weigh significance in ways historians would accept for publication or archival work.
Technical feasibility todayclaude-sonnet-52/5Some research and productivity tools assist with document search, transcription, and summarization, but no deployed product reliably performs authenticity verification or significance analysis at professional historian standards.

Coordinate activities of workers engaged in cataloging and filing materials.

29

CI 20–39 · exposure 28 · augmentation 63 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Archives, libraries, and museums—where this task concentrates—show slow digital transformation and conservative adoption of autonomous management systems. Workforce coordination remains one of the most human-centric functions even in digitized institutions.
Sector adoption velocityclaude-sonnet-52/5Archival and historical institutions are typically slow adopters of AI compared to finance or tech, with pilots for digitization tools more common than adoption of AI-driven workflow management.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist coordinators by suggesting task prioritization, generating cataloging schedules, or flagging bottlenecks, which raises human coordinator productivity moderately. The coordinator remains essential but can better manage larger or more complex teams with AI-assisted workflow insight.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by automating metadata tagging, suggesting classification schemes, and tracking progress, freeing the human coordinator to focus on quality control and worker guidance.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist with categorization and filing logic, coordinating *workers* requires real-time task allocation, conflict resolution, and adaptive planning based on human performance and interpersonal dynamics—capabilities current AI systems lack at production quality. Partial automation of cataloging workflows is possible, but end-to-end worker coordination remains predominantly human.
Task automatabilityclaude-sonnet-53/5The cataloging/filing scheme design and worker coordination requires managerial judgment, but AI can automate much of the actual categorization and metadata tagging work being coordinated, offering partial but not full time savings on the supervisory task itself.'
Adoption barriersclaude-haiku-4-5-202510014/5Coordination roles often carry accountability for team performance and outputs; institutional practice strongly favors human supervisory authority for worker management, and cultural/organizational friction around autonomous task assignment to staff is high. Liability and trust barriers are substantial.
Adoption barriersclaude-sonnet-53/5No licensing requirement, but organizational friction exists since supervising staff involves interpersonal judgment, performance evaluation, and accountability that organizations resist delegating to AI.
Cost vs. human wageclaude-haiku-4-5-202510012/5Basic task assignment and monitoring could use cheap workflow software, but the supervision, conflict resolution, and adaptive scheduling that distinguish effective coordination remain cheaper to handle with human coordinators than to build robust AI oversight. All-in costs remain roughly comparable or favor human coordination.
Cost vs. human wageclaude-sonnet-52/5Coordinating human workers still requires a human manager for accountability and interpersonal direction, so AI can reduce some administrative overhead but not replace the managerial cost wholesale.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably performs autonomous worker coordination at scale; existing project management and workflow tools require heavy human oversight and do not dynamically allocate or supervise human cataloging teams. Coordination demands contextual judgment and personnel management that current systems cannot operationalize.
Technical feasibility todayclaude-sonnet-52/5Document management and workflow tools exist, but AI-driven coordination of human workers' cataloging activities specifically is not a mature deployed product category; most tools assist individual cataloging tasks rather than managing teams.

Prepare publications and exhibits, or review those prepared by others, to ensure their historical accuracy.

28

CI 25–30 · exposure 25 · augmentation 63 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Historical institutions and academic publishers adopt AI cautiously. Current use is limited to preliminary drafting support and minor consistency checking, not core accuracy review. The sector values human expertise and is slow to automate judgment-intensive scholarly tasks.
Sector adoption velocityclaude-sonnet-52/5Museums, academia, and publishing (sectors employing historians) are relatively slow, under-resourced adopters of AI compared to finance or tech.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist historians by flagging potential inconsistencies, suggesting cross-references, and accelerating fact-checking workflows, but the historian remains responsible for interpretation and final judgment. This supportive role raises productivity on routine verification tasks without replacing scholarly authority.
Augmentation potentialclaude-sonnet-54/5AI tools are genuinely useful for drafting, cross-referencing sources, and catching inconsistencies, substantially aiding historians while they retain final judgment.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist with fact-checking and identifying potential historical inconsistencies, but the task fundamentally requires human judgment about interpretation, significance, and nuance in historical narratives. Current systems cannot reliably verify complex contextual accuracy or ensure scholarly rigor across diverse historical domains at the 50% time-saving threshold.
Task automatabilityclaude-sonnet-52/5AI can draft text and flag some factual inconsistencies, but verifying historical accuracy requires deep contextual judgment, primary-source interpretation, and originality that current systems cannot reliably perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Historians and institutions face strong professional and reputational barriers to replacing human review: peer-review standards, museum/publication ethics, institutional accountability for accuracy, and the expectation that publications carry human scholarly authority. Errors in historical interpretation carry significant institutional and professional risk.
Adoption barriersclaude-sonnet-53/5No licensing requirement, but professional/academic norms, institutional reputational risk, and curatorial standards create meaningful friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI inference costs for thorough historical review (multiple passes, integration with archival databases, human oversight of flagged items) remain comparable to or exceed the cost of junior historian labor, especially when accounting for oversight and liability.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply generate drafts, but human expert review remains essential to catch factual errors and nuanced interpretation, keeping overall cost savings modest once oversight is included.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI language models can perform basic fact-retrieval and consistency-checking, no deployed product reliably performs comprehensive historical accuracy review at the level required by professional historians. Products exist for document analysis but with material error rates and limited contextual understanding of historiography.
Technical feasibility todayclaude-sonnet-52/5AI writing and research assistants are used to support drafting, but no deployed product independently ensures historical accuracy of publications or exhibits at production scale.

Conduct historical research as a basis for the identification, conservation, and reconstruction of historic places and materials.

28

CI 25–30 · exposure 25 · augmentation 63 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Historical research and conservation operate in public institutions, universities, and heritage organizations—sectors with slow digital transformation and high value placed on specialized expertise; adoption of AI for core research tasks remains minimal despite digitization of archives.
Sector adoption velocityclaude-sonnet-52/5Academic and heritage/cultural sectors are historically slow AI adopters compared to finance or tech, with pilots for digital archives emerging but production-scale AI-driven historical research still rare.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist historians by automating document digitization, enabling full-text search of archives, suggesting related sources, and flagging anomalies in datasets, thereby accelerating research phases while the historian retains analytical and interpretive authority.
Augmentation potentialclaude-sonnet-54/5AI substantially aids in searching digitized archives, translating documents, summarizing sources, and generating research leads, meaningfully boosting historian productivity while they retain interpretive control.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist with literature discovery, digitized document analysis, and fact-gathering from databases, but historical research requires contextual judgment, source evaluation, and expertise in archival work that cannot be fully automated. The task's core—identification and conservation decisions—remains deeply dependent on expert interpretation.
Task automatabilityclaude-sonnet-52/5AI can accelerate literature search, document summarization, and initial archival triage, but the core work of synthesizing primary sources, contextual judgment, and physical/material verification requires human expertise that current AI cannot fully replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Historic conservation is regulated by heritage authorities and often requires licensed or credentialed specialists; legal and ethical liability for incorrect identification or reconstruction falls on human experts, and institutional/governmental trust requires human sign-off on major decisions.
Adoption barriersclaude-sonnet-53/5No licensing requirement for historians per se, but conservation/reconstruction decisions often require professional certification, peer review, or regulatory compliance (e.g., heritage protection laws), creating moderate institutional friction.
Cost vs. human wageclaude-haiku-4-5-202510012/5While AI tools (subscriptions, APIs) are relatively cheap, the labor cost of a trained historian is moderate, and AI still requires significant expert oversight and validation for research-critical tasks, making the cost advantage minimal or absent when integration and verification are factored in.
Cost vs. human wageclaude-sonnet-52/5AI can cut costs for literature review and text search, but the labor-intensive archival visits, expert interpretation, and material analysis remain human-cost-dominated, keeping overall savings modest relative to full task cost.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI tools exist for document OCR, text search, and information retrieval, but no deployed system reliably performs end-to-end historical research with the scholarly rigor and accuracy required for conservation and reconstruction decisions. Products lack the domain expertise and error tolerance needed for high-stakes heritage work.
Technical feasibility todayclaude-sonnet-52/5AI research tools (search, summarization, OCR of archives) are deployed and used by historians as aids, but no product reliably conducts full historical research including provenance judgment, archival fieldwork, or conservation-relevant material assessment.

Conduct historical research, and publish or present findings and theories.

28

CI 25–30 · exposure 25 · augmentation 63 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Academic history remains a low-automation sector with strong cultural norms favoring human expertise and slow digital transformation. While some historians use AI for auxiliary tasks, production adoption of AI-driven research or writing remains minimal and controversial.
Sector adoption velocityclaude-sonnet-52/5Academia and humanities research remain slow adopters of AI tools for core scholarly work, with usage concentrated in writing assistance and search rather than substantive research automation.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist historians by accelerating literature searches, organizing primary documents, detecting patterns in large text corpora, and suggesting thematic connections. However, the historian must remain the primary evaluator and theory-builder, so augmentation is helpful but not transformative of the core task.
Augmentation potentialclaude-sonnet-54/5AI meaningfully assists historians with literature search, translation, transcription of archival documents, summarization, and drafting, improving efficiency while the historian retains interpretive control.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist with literature review, document digitization, and preliminary data organization, but historical research fundamentally requires human judgment in source interpretation, theory formation, and establishing novel historical arguments. The creative and evaluative core of this task—deciding what is significant and why—remains beyond current AI capabilities.
Task automatabilityclaude-sonnet-52/5Original archival research, source interpretation, and theory formation require deep contextual judgment and synthesis that current AI cannot reliably perform end-to-end; AI can assist with search and drafting but not the core scholarly work.
Adoption barriersclaude-haiku-4-5-202510014/5Academic publishing and historical authority carry strong institutional and reputational barriers. Historians' credibility depends on their name and expertise; institutions require human authorship and accountability. Peer review and disciplinary norms create friction against fully automated or AI-primary historical claims.
Adoption barriersclaude-sonnet-53/5No licensing requirement, but academic peer review, institutional credentialing, and scholarly credibility norms create real friction against AI-authored findings being accepted as valid scholarship.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI infrastructure for historiographic work (document processing, literature mining) remains relatively expensive per insight generated, and requires significant human oversight and validation. The cost of errors in historical interpretation (reputational, academic) is high, offsetting modest computational savings.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply handle some literature searching and drafting, but human verification, archival access, and interpretive rigor still dominate cost, keeping overall savings modest relative to full task cost.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI tools exist for text analysis and document processing, no deployed product reliably conducts independent historical research or generates publishable historical arguments. Products like document search and plagiarism detection exist, but they do not perform the core task of synthesizing evidence into novel historical claims.
Technical feasibility todayclaude-sonnet-52/5AI tools (search assistants, summarization, translation) exist and are used in literature reviews, but no deployed product independently conducts historical research or generates defensible original findings at production quality.

Determine which topics to research, or pursue research topics specified by clients or employers.

28

CI 25–30 · exposure 25 · augmentation 75 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Academic institutions and historical research organizations have adopted AI literature tools slowly and cautiously; topic determination remains a human-centered, deliberative process in most history departments and archives, with minimal production-level displacement.
Sector adoption velocityclaude-sonnet-52/5Academic and archival history is a slow-digitizing field with limited production AI deployment for high-level scholarly decision-making, though individual researchers experiment with AI tools.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can meaningfully augment historians' topic discovery by rapidly surveying published research, identifying gaps, and suggesting under-explored angles; historians using LLMs for literature synthesis and brainstorming can explore more potential topics faster while retaining final selection authority.
Augmentation potentialclaude-sonnet-54/5AI can effectively help historians survey existing literature, identify underexplored angles, and generate topic ideas, meaningfully speeding up the ideation phase while the historian retains final judgment.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist in identifying relevant research topics and generating topic suggestions from existing literature, but cannot independently determine which topics merit scholarly investigation—this requires domain expertise, novelty assessment, and alignment with funding/institutional priorities that exceed current AI capabilities.
Task automatabilityclaude-sonnet-52/5Choosing a research direction depends on scholarly judgment, novelty assessment, funding priorities, and client relationships that AI cannot independently originate or negotiate; AI can suggest topics but not determine them end-to-end with equal quality at scale.
Adoption barriersclaude-haiku-4-5-202510014/5Topic selection in academic and professional contexts is a gatekeeping function closely tied to expertise, institutional credibility, and funding bodies' priorities; clients and employers expect historians to exercise professional judgment and take responsibility for research direction, creating high organizational and reputational barriers to full substitution.
Adoption barriersclaude-sonnet-53/5No formal licensing requirement, but professional norms, client trust, institutional funding processes, and reputational stakes create meaningful friction against ceding this decision to AI.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI-assisted topic research tools exist but are typically priced as research subscriptions or modules; the integrated cost remains comparable to or higher than paying a historian for preliminary topic scoping, especially given oversight requirements.
Cost vs. human wageclaude-sonnet-52/5AI brainstorming is cheap, but the actual decision requires expert vetting, client interaction, and domain judgment that still demands significant paid human time, limiting overall cost savings.
Technical feasibility todayclaude-haiku-4-5-202510012/5While LLMs can suggest topics and summarize research gaps, no deployed product reliably performs the full task of determining what historians should research; systems exist for literature analysis but lack the contextual judgment and disciplinary authority required for genuine topic selection.
Technical feasibility todayclaude-sonnet-52/5AI tools can surface gaps in literature or brainstorm topics, but no deployed product reliably performs topic selection as a trusted final decision in professional historical research.

Recommend actions related to historical art, such as which items to add to a collection or which items to display in an exhibit.

28

CI 25–30 · exposure 25 · augmentation 63 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Adoption in museums and historical institutions remains limited; most projects are pilots or internal decision-support tools rather than autonomous systems. Cultural institutions move conservatively on acquisitions and exhibitions, and professional curators remain central to institutional identity.
Sector adoption velocityclaude-sonnet-52/5Museums and cultural heritage institutions are a slow-adopting sector with limited digitization and few AI deployment cases, in contrast to fast-moving sectors like finance or tech.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully assist historians by flagging relevant works, cross-referencing metadata, identifying thematic connections, and generating candidate lists for review. However, the assistant role is bounded—final judgment on value, provenance, and fit remains squarely with the human expert.
Augmentation potentialclaude-sonnet-54/5AI tools can assist historians significantly by aggregating provenance data, comparable works, trends, and background research to inform curatorial decisions, meaningfully speeding up the research phase of the task.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in identifying artworks, suggesting items based on curatorial metadata, and analyzing thematic fit, the task requires domain expertise, aesthetic judgment, and institutional context that current systems cannot reliably replicate end-to-end. AI cannot independently evaluate historical significance or collection strategy at the quality level expected of a professional historian.
Task automatabilityclaude-sonnet-52/5AI can surface information and suggest options but curatorial recommendations require nuanced judgment about provenance, narrative, audience, and institutional mission that current systems cannot reliably replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Museums and institutions typically employ credentialed curators whose professional judgment and institutional authority are legally and ethically central to stewardship decisions. Professional standards, insurance, donor relations, and scholarly reputation create strong organizational and reputational barriers to outsourcing recommendations to AI without expert sign-off.
Adoption barriersclaude-sonnet-53/5No formal licensing requirement exists, but strong institutional, reputational, and donor-relationship factors as well as the need for scholarly authority create meaningful friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5Integration costs for AI systems (model training on institutional collections, custom integration, ongoing curation oversight) combined with the need for expert human review mean the all-in cost per recommendation remains higher than or comparable to having a historian directly evaluate candidates.
Cost vs. human wageclaude-sonnet-52/5While AI querying/summarization is cheap, the actual value-add of expert curatorial judgment means human historians remain necessary, so cost savings are limited relative to the specialized expertise required.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably performs curatorial recommendation at production scale; systems exist for art classification and metadata search, but they lack the nuanced judgment and accountability required for actual acquisition or exhibition decisions. Museums do not substitute curator expertise with AI recommendations.
Technical feasibility todayclaude-sonnet-52/5There are no mature deployed products that autonomously make acquisition or exhibit-selection recommendations in museums; AI is at best used as a research aid, not a decision-maker in production workflows.

Teach and conduct research in colleges, universities, museums, and other research agencies and schools.

22

CI 19–25 · exposure 25 · augmentation 75 · importance 3.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Academic institutions are slow-moving, credential-bound sectors with strong norms favoring faculty and established peer review. Adoption of AI for routine research or teaching replacement remains minimal despite digitization; most AI use in academia remains assistive.
Sector adoption velocityclaude-sonnet-52/5Higher education and research institutions have been slow and cautious in adopting AI for core teaching and scholarship, with pilots for administrative tasks more common than for core duties.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can substantially enhance historian productivity through automated transcription of primary sources, rapid literature mapping, cross-referencing of archives, and outline/draft generation, allowing historians to focus on interpretation and argumentation. These tools are already in use in research workflows.
Augmentation potentialclaude-sonnet-54/5AI substantially aids literature reviews, drafting, translation, and course material generation, meaningfully boosting historians' productivity while they retain analytical and pedagogical control.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist with literature reviews, source analysis, and draft preparation, the core activities—original historical interpretation, classroom teaching presence, and mentoring—require human creativity and contextual judgment. The task involves complex synthesis that AI cannot reliably perform end-to-end at equal quality.
Task automatabilityclaude-sonnet-52/5Teaching and original historical research require sustained human judgment, mentorship, and archival investigation that current AI cannot perform end-to-end; only sub-components (lecture prep, literature summarization) meet the time-saving bar.
Adoption barriersclaude-haiku-4-5-202510014/5Universities and research institutions have strong hiring, credentialing, and tenure structures that legally and organizationally require human faculty. Teaching requires institutional authorization; research publication carries editorial and peer-review gatekeeping that privileges human expertise and institutional affiliation.
Adoption barriersclaude-sonnet-54/5Teaching in accredited institutions requires credentialed faculty, and original scholarly research demands attributable human authorship and academic accountability, creating strong institutional and professional barriers.
Cost vs. human wageclaude-haiku-4-5-202510011/5The loaded cost of a historian—salary, benefits, institutional overhead—far exceeds any plausible cost of AI assistance. AI cannot substitute for the human researcher/educator at comparable quality, making the ratio heavily unfavorable for automation.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply assist with drafting or summarizing, but full replacement of teaching and research responsibilities would still require costly human oversight and credentialed expertise, keeping overall costs comparable to human labor.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI system can autonomously teach classes or conduct publishable historical research. Products exist for research support (literature mapping, document analysis) but fall short of reliable end-to-end performance on the full task scope in production settings.
Technical feasibility todayclaude-sonnet-52/5AI tools exist for research assistance and course material generation, but no deployed product independently conducts historical research or teaches courses reliably at scale.

Advise or consult with individuals and institutions regarding issues such as the historical authenticity of materials or the customs of a specific historical period.

22

CI 14–30 · exposure 17 · augmentation 63 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Historical consulting is concentrated in academia, museums, and specialized archives—sectors with high regard for human expertise, slow digitization, and limited pressure to automate professional judgment. Adoption of AI for advisory roles remains negligible.
Sector adoption velocityclaude-sonnet-52/5Academic and museum/archival sectors show slow, uneven AI adoption relative to fast-moving digital industries, with most use limited to research support rather than consultation delivery.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by rapidly retrieving comparative historical sources, summarizing archival materials, and flagging potential authenticity concerns for the historian to evaluate. This augmentation improves research speed but does not transform the core advisory task, which remains human-centered.
Augmentation potentialclaude-sonnet-54/5AI tools can rapidly surface historical records, cross-reference sources, and draft contextual summaries, significantly aiding a historian's research and consultation process while the human retains final judgment.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires deep contextual judgment about historical authenticity and period-specific customs, demanding nuanced interpretation of evidence and comparative knowledge across periods. Current AI systems cannot reliably perform end-to-end consultation that would meet the 50% time-saving threshold while maintaining the evidentiary rigor historians require.
Task automatabilityclaude-sonnet-52/5AI can provide background research and preliminary assessments of historical context, but authoritative advisory consultation requiring nuanced judgment, reputation, and accountability cannot be fully automated end-to-end today.'
Adoption barriersclaude-haiku-4-5-202510014/5Institutions and individuals seeking historical consultation typically require accountability and professional credentials; a human historian's reputation and judgment are key assets. Legal and reputational liability for incorrect authentication or advice creates strong friction against full AI substitution.
Adoption barriersclaude-sonnet-53/5No licensing requirement generally governs historians' consulting work, but institutional trust, reputational liability, and client preference for credentialed experts create moderate friction against pure AI substitution.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI assistance might reduce research overhead, but the expert human historian must still validate conclusions and take professional responsibility for advice. The integration and oversight costs remain substantial relative to what would be displaced from the historian's work.
Cost vs. human wageclaude-sonnet-52/5While AI queries are cheap, the expert judgment and liability inherent in professional historical consultation still require human oversight, keeping effective cost per reliable output relatively high.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can retrieve historical facts and draft preliminary analyses, no deployed product reliably performs authentic historical consultation at production quality. Historians would need to heavily vet AI outputs for accuracy, anachronism, and proper sourcing—defeating the productivity gain.
Technical feasibility todayclaude-sonnet-52/5No deployed product performs authoritative historical authentication or consultation reliably in production; existing LLMs are used informally as research aids, not as certified consultants.

Interview people to gather information about historical events and to record oral histories.

22

CI 14–30 · exposure 17 · augmentation 63 · importance 3.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Historians work in academic and cultural institutions with slow tech adoption and strong professional norms valuing human-centered scholarship; there is no meaningful market pressure or evidence of production AI replacing interviews in this sector.
Sector adoption velocityclaude-sonnet-52/5Historians and archival institutions are slow adopters of AI for primary research methods, though transcription tools are increasingly used as auxiliary aids.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist historians by transcribing interviews, summarizing transcripts, suggesting follow-up questions, and organizing archival metadata, but the human historian remains essential for the interview itself and interpretive judgment.
Augmentation potentialclaude-sonnet-54/5AI substantially assists with transcription, question preparation, thematic coding, and summarization of oral history recordings, boosting historian productivity while the human still conducts the interview.
Task automatabilityclaude-haiku-4-5-202510011/5Conducting interviews requires genuine human engagement, relationship-building, and adaptive questioning that responds to unpredictable conversational dynamics. Current AI systems cannot reliably conduct authentic interviews with the nuance and sensitivity required to elicit meaningful historical narratives from human subjects.
Task automatabilityclaude-sonnet-52/5AI can help draft interview questions and transcribe/summarize recordings, but conducting a live human interview requiring rapport, follow-up judgment, and adaptive probing cannot be fully offloaded to current systems.
Adoption barriersclaude-haiku-4-5-202510014/5Strong barriers exist: informed consent and ethical interview practices require human discretion and accountability; oral history is a regulated/credentialed field in many academic and archival contexts; subjects typically expect and prefer human interviewers for trust and authenticity.
Adoption barriersclaude-sonnet-53/5No licensing requirement, but interview subjects typically expect human interaction, trust-building, and ethical handling of sensitive personal narratives, creating moderate friction.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI transcription and analysis tools have modest costs, but the core interview task still requires human labor; automation would primarily shift work rather than dramatically reduce per-unit cost relative to skilled historian wages.
Cost vs. human wageclaude-sonnet-52/5AI transcription/summarization is cheap, but the core interviewing labor still requires a human interviewer, so overall cost savings are limited to peripheral tasks.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can transcribe and assist with interview management, no deployed product reliably conducts the full interview process end-to-end with human subjects today. Some tools aid preparation and post-processing, but the live interpersonal interaction remains fundamentally human.
Technical feasibility todayclaude-sonnet-52/5Transcription and summarization tools are mature, but no deployed product autonomously conducts nuanced oral history interviews with strangers reliably today.

Speak to various groups, organizations, and clubs to promote the aims and activities of historical societies.

9

CI 5–13 · exposure 0 · augmentation 38 · importance 3.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Historical societies and community organizations operate in sectors with low AI adoption for core mission delivery; they rely on volunteer and professional networks where in-person speaking remains central and is not being displaced.
Sector adoption velocityclaude-sonnet-52/5Historical societies and cultural nonprofits are typically slow adopters of AI, especially for public-facing relational tasks like speaking engagements.
Augmentation potentialclaude-haiku-4-5-202510012/5AI can assist by drafting speech outlines, suggesting talking points, or identifying relevant historical details, but augmentation is limited since the human historian must deliver and adapt the message in real time.
Augmentation potentialclaude-sonnet-53/5AI can help draft speeches, talking points, slides, and audience research, meaningfully aiding preparation even though delivery remains human.
Task automatabilityclaude-haiku-4-5-202510011/5Speaking to live groups and organizations requires real-time audience engagement, adaptive communication, and charismatic presence that current AI systems cannot deliver end-to-end. While AI can draft talking points, the core task of in-person or live speaking to diverse audiences remains firmly human-dependent.
Task automatabilityclaude-sonnet-51/5Live public speaking to promote an organization requires physical presence, personal rapport, and real-time audience engagement that current AI cannot replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Organizations and clubs specifically seek a human historian to speak, promote engagement, and represent the society. Customer preference for human authority figures, combined with the expectation of expert presence, creates substantial barriers to AI substitution.
Adoption barriersclaude-sonnet-53/5No licensing requirement exists, but strong preference for human presence, credibility, and relationship-building at community events creates real friction against substitution.
Cost vs. human wageclaude-haiku-4-5-202510011/5The task requires a human speaker whose presence and expertise command fees; AI cannot substitute here at any cost advantage since the task inherently demands human representation and live interaction.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute delivering the same in-person persuasive speaking service, so cost comparison favors the human by default.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably performs live public speaking or group presentations as a substitute for a human speaker. AI can assist with speech writing but cannot authentically represent a historical society or engage audiences in real time.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs live promotional speaking engagements to clubs and groups on behalf of an organization; this remains firmly human-delivered.

Related occupations — Life, Physical & Social Science

How to read this

A high substitution score does not mean this job disappears — it means a large share of its current tasks face replacement pressure, so the mix of tasks is likely to change. High augmentation alongside substitution typically means the occupation reorganizes around the protected tasks. Wide confidence intervals mean the rater panel disagreed: treat those scores as open questions, not verdicts.

What would change this score

New model capabilities (automatability, feasibility), falling inference costs (cost ratio), regulation and licensing shifts (barriers), and measured sector adoption (velocity) all re-enter at every index release. Each release is recomputed, versioned and kept queryable — scores are claims with a date on them, not permanent labels.