■ HIGH RISK ■ Education
AI is swallowing the cataloging, transcription, and finding-aid drudgery that consumed most archivist hours — but appraisal, context, and deciding what deserves to survive remain human calls. Fewer positions, more curatorial ones.
“Digital archives organize themselves. Your white gloves are decorative now.”
Our AI replacement risk score — how we score jobs
Archivists decide what history gets kept and how anyone finds it later. The day-to-day mixes appraisal (which of these 400 boxes from a defunct company matter?), arrangement and description (building finding aids so researchers can navigate collections), preservation (acid-free folders, climate logs, digitization queues), reference work with researchers, and increasingly the wrangling of born-digital records — email archives, hard drives, institutional systems that nobody documented.
The descriptive middle of that job is exactly what AI does well. Handwritten-text recognition now transcribes historical documents that once required paleography skills; speech-to-text processes oral history backlogs; machine-generated metadata, entity extraction, and auto-classification can produce first-draft finding aids from digitized collections at a scale no staff could match. Archives are chronically backlogged — 'more product, less process' became doctrine because description couldn't keep up — so institutions will happily let models chew through the backlog. Grant-funded cataloging positions, a traditional entry point, are the first to feel it.
What resists is judgment and stewardship. Appraisal is an ethical act: choosing what to keep shapes what the future can know, and institutions won't delegate that to a classifier trained on the past's biases. Donor relations, navigating privacy and copyright in collections, authenticating records, handling culturally sensitive material, and designing digital-preservation strategy for formats that rot in decades — all remain human work, and the born-digital deluge is creating more of it. The profession's problem is arithmetic: automation removes the tasks that justified headcount, concentrating remaining roles among fewer, more senior archivists. Our 61 reflects a field transformed and thinned rather than eliminated.
Automatability: our editorial assessment of current and near-term AI capability
Serious pressure by around 2030. Handwriting recognition and auto-description tools are moving from pilot projects to production workflows in large archives now, and budget-strapped institutions will convert backlog-cataloging positions into software licenses as the tools mature mid-decade. Appraisal-, rights-, and digital-preservation-focused roles hold; the entry-level descriptive jobs that trained past generations of archivists get scarce first.
Viable but tighter. The descriptive labor that filled entry-level jobs is automating, so fewer positions exist per institution — and the field was already competitive. Candidates who combine traditional appraisal training with digital-preservation and data-governance skills remain employable, including outside cultural institutions. Going in expecting a career of arranging and describing boxes is the risky version of the plan.
Increasingly well, yes. Handwritten-text recognition handles many 18th- to 20th-century scripts at accuracy levels that make transcription review, rather than transcription itself, the human job. Difficult hands, damaged documents, and unusual languages still defeat it. For archives, this converts impossible transcription backlogs into feasible review projects — good for access, hard on transcription-based employment.
The parts with consequences: deciding what to acquire and keep, negotiating with donors, weighing privacy and copyright, authenticating disputed records, and setting preservation strategy for digital formats that decay. These involve ethics, relationships, and institutional accountability. An algorithm can describe a collection; it can't take responsibility for destroying one.
Adopt them aggressively and own the quality layer. Archives have decades of backlog — using AI to clear it makes you more valuable, not less, if you're the person designing the workflow and auditing the output. Simultaneously, shift your professional identity toward appraisal, digital preservation, and access strategy, which is where positions will consolidate.