Fields / City
First stop
Preservation already has a strong hold in this corpus, but almost all of it is physical: Forestry keeps trees standing, Archaeological Conservation keeps excavated objects intact, Museum Conservation keeps paintings and artefacts from decaying.
This field is Preservation applied to the informational form of the same instinct — the record, rather than the object. A library, an archive, and a records office all exist because information that matters — a historical decision, a scientific dataset, a birth certificate, a company's own account of what it did and why — does not survive on its own. Someone has to decide what is worth keeping, describe it well enough that a future person can find it, and hold it against the two things that destroy information over time: physical decay and simple loss of context. The structural pull is Preservation because the entire deliverable of this field, across every one of its roles, is that something outlasts the institution, the format, or the person who made it.
The field is not one job, and the four roles that carry it pull in genuinely different directions while sharing the same root instinct. An archivist appraises, arranges, and describes historical and institutional records — manuscripts, organisational archives, personal papers — so that a researcher decades from now can actually locate what they need inside a collection that might run to thousands of boxes; this is documents and records specifically, distinct from ``'s Archaeological Conservator (who stabilises physical objects recovered from excavation) and from ``'s Collections Manager / Registrar (who tracks and moves museum objects, not paper or digital records). A digital preservation specialist does the same custodial work for material that was born digital or has been digitised, against a threat physical archivists never faced — file formats, storage media, and the software needed to open a file all become obsolete on a timescale of years, not centuries, and someone has to keep migrating, checking, and re-certifying digital collections just to keep them readable. A public librarian runs the point of the field where Preservation meets Spread: a public library's core purpose is not storage but access, getting information, services, and increasingly basic digital and social infrastructure out to the whole population of a community rather than a specialist few. And an academic subject librarian teaches the specific skill of finding and evaluating information within a discipline, and — increasingly — manages the research data that underpins a university department's published work; the role is a specialist information professional embedded in a department, and is explicitly not ``'s Academic Administrator (Dean / Provost), a senior institutional-leadership role with which it shares only a campus, not a job.
Underneath Preservation, Organization and Explanation do most of the field's daily texture. Nothing in this field is preserved by accident — it survives because someone imposed a defensible order on it (an archival finding aid, a library's classification system, an organisation's retention schedule) and because someone made that order intelligible to the people who need to use it. A records and information governance manager sits at the sharpest edge of this: inside any organisation of real size, someone has to decide, and defend, what gets kept and for how long, balancing statutory retention requirements against UK GDPR's data-minimisation principle, which requires records to be destroyed once they are no longer needed — get either side of that wrong and the organisation is in breach of the law.
What nobody tells a student interested in this field is how unevenly its fortunes are currently running. Public libraries are genuinely under sustained financial pressure — revenue funding for UK public libraries has been cut by roughly £243 million a year in real terms since 2010, and CILIP's own Libraries at Risk Monitor is still actively tracking closures and service reductions into 2026 [CILIP / Big Issue joint report; CILIP Libraries at Risk Monitor 2026]. At the same time, the corners of the field built around digital preservation and records governance are growing, driven by forces that have nothing to do with library funding at all: every organisation now legally has to manage personal data under UK GDPR, and every institution with a serious digital collection has discovered that "keep the file" and "keep the file readable" are two different, and increasingly well-paid, problems. A student choosing this field is really choosing between several quite different economic realities that happen to share a professional root.
The entry route runs predominantly through a CILIP-accredited postgraduate qualification for librarianship and information science, or an ARA-accredited (Archives & Records Association) postgraduate qualification for archives and records management — several UK universities offer courses accredited by one or both bodies. A genuine non-graduate route has existed via library-assistant-to-Chartership progression and, for archives specifically, a Level 7 apprenticeship — but that apprenticeship route narrowed materially in 2026: the UK government withdrew funding for new Level 7 Archives and Records Management apprenticeship cohorts starting that year, though existing cohorts are funded through to completion [Archives & Records Association / UK apprenticeship funding policy 2026]. That makes the postgraduate-qualification route the more reliable path into the profession for a student starting now, even though the field has historically prided itself on multiple ways in.
The pull
Ephemeral → Enduring
The pull to make things last
Organization runs alongside it. Underneath Preservation, Organization and Explanation do most of the field's daily texture.
The line forks
Service update
What is newly possible in Libraries, Archives & Records Management, what stays yours, and how AI is changing the way in. In that order, and only if you want it.
3 barriers have dropped
Item-level description was never affordable at the rate collections arrived; the profession's own 'more product, less process' minimal-processing standard was an explicit acknowledgment that most material would only ever get coarse description, producing a 'hidden collections' problem of accessioned but functionally unfindable material.
AI-drafted description -- LLMs and multimodal models generating first-pass catalogue records and finding-aid text from digitised or transcribed material, verified by a trained archivist rather than written from scratch -- shifts labour from drafting to verifying, letting a single archivist move much more previously-undescribed material to a finer grain of description.
A single archivist making a large, previously undescribed backlog collection genuinely findable at a much finer grain of description than minimal processing ever allowed, because AI drafts the description and the archivist verifies it rather than writing every entry from scratch.
A small institution with one archivist clearing years of accessioned-but-undescribed material without needing to hire additional cataloguing staff.
Very likely to keep going this way. How much it helps someone entering the field is less certain: as the tools become standard at well-funded archives, the advantage moves from using them to checking their output reliably.
This speeds up drafting descriptions of archive material; it does not replace deciding what is worth keeping. Every AI draft still needs checking by a trained archivist, because a confidently wrong description hides material again. It helps more with typed material than with difficult old handwriting.
Effective digital preservation required either an in-house Digital Preservation Coalition-trained specialist or an expensive external consultancy, out of reach for small archives and single-person records offices, which risked quietly losing born-digital material to format obsolescence with no early warning.
AI-assisted format-identification and preservation-risk tooling -- automated fixity checking and predictive flagging of at-risk formats and storage media -- lowers the technical floor for a non-specialist to run baseline digital-preservation triage.
A small institution's single archives or records person running baseline digital-preservation risk triage without in-house specialist expertise, so problems surface before they become losses instead of after.
Could go either way. The evidence comes mainly from companies selling the tools, not independent studies. How much it counts in a career is also less certain, as running these checks may become a basic expectation rather than something that stands out.
This only helps spot which digital files are in danger of being lost. Proper long-term preservation, such as complex decisions about moving files to new formats and a sound preservation policy, still needs specialist judgement. It lets a non-specialist know there is a problem, not solve it.
Building and defending a records retention schedule that reconciles statutory retention against UK GDPR's data-minimisation principle required an in-house specialist or an external compliance consultancy, out of reach for smaller organisations that are nonetheless fully bound by UK GDPR.
AI-assisted policy drafting and bulk-records classification against regulatory retention templates makes a defensible baseline records-governance programme newly viable for a single, non-specialist administrator inside a smaller organisation.
A single administrator inside a small organisation running a defensible baseline records-governance programme -- drafting and applying retention classifications against statutory templates -- without commissioning a specialist compliance consultancy.
Could go either way. How big the change gets is also less certain. The evidence is thin and comes mostly from companies selling the tools and industry publications.
AI can draft a plan for how long records are kept, using a general template. A person who understands the organisation's own legal position must still sign it off and defend it under UK data protection law. If it goes wrong, the regulator investigates that person, not the tool.
These are real jobs that exist now and did not exist before the current wave of AI. Tap one to see what it is and how well we know it.
Directs, validates, and corrects AI-drafted archival description and metadata at scale, rather than drafting from scratch -- the human-in-the-loop verification layer of AI-assisted cataloguing workflows.
'Making Hidden Collections Visible' human-AI collaboration case study; Springer AI-assisted archival metadata generation platform · current fact
A digital preservation specialist whose remit explicitly includes evaluating and directing AI-assisted preservation tooling (format-risk prediction, fixity-checking automation) alongside traditional digital-preservation practice.
IFLA 'AI and the Future of Digital Preservation' programming; Preservica's framing of AI-tooling evaluation as a digital-preservation competency · seen in the wild
Extends records-governance practice to the records generated by an organisation's own AI agents -- treating agent decisions and outputs as records subject to retention, access, and audit obligations.
Smarsh 'Government Records Management in 2026' framing of AI agents needing identity/access governance similar to human users · early signal
Coordinates an institution's overall approach to AI adoption across archives, library, and records functions -- tool evaluation, institutional policy, staff AI literacy.
ARA 'AI Preparedness Guidelines for Archivists' 2026 implies institutional operationalisation is underway · early signal
The AI-Assisted Cataloguing archivist and Digital Preservation / AI Tooling Specialist configurations are documented in named case studies and professional-body programming. The AI-Agent Records Governance Officer and AI Preparedness Lead configurations are structurally reasoned extensions of documented concerns and guidance, not yet independently verified as named, hired-for job titles -- read them as emerging patterns worth watching, not established roles.
Bars above the line are the parts of this work that still need a person. Bars below it are what AI can already do. Tap any column to see the actual work behind it.
The honest read. Three strong dimensions and two moderate, distributed so that every one of the field's five archetypes carries at least one strong robustly-human dimension -- protection is spread across the field rather than concentrated in a single archetype, which is part of why this field reads more AI-resistant overall than its deployment-maturity scores alone would suggest.
The field's AI story concentrates in language reasoning and computer vision, both applied to the same underlying task -- making material describable and findable -- while agentic AI is the credible near-term frontier rather than current practice. The clearest signal that this field's AI story is still early is that its most serious institutional pilot (GDS + The National Archives, 2026) explicitly chose to spend its first phase on data quality and governance rather than on deploying a tool.
3 ways into this field, and AI is not doing the same thing to each of them. One is opening up rather than closing.
AI-assisted description drafting is compressing some of the practice-task volume that used to build entry-level fluency (Collapse 1), compounded by the UK government's 2026 withdrawal of funding for new Level 7 Archives and Records Management apprenticeship cohorts, narrowing the field's other genuine non-graduate entry route
UK public library real-terms funding cuts of ~GBP243m/year since 2010 dominate entry compression here; CILIP Libraries at Risk Monitor still tracking closures into 2026; AI's own direct contribution (chatbot absorption of routine transactional load) is real but secondary to the funding story
UK GDPR's universal applicability and expanding digital/born-digital collection volumes create durable, non-discretionary demand structurally unrelated to library funding; Mode 1 independently frames digital preservation as one of the field's most in-demand corners, with advertised roles (GBP40k-54k) well above the sector's trainee baseline
That is everything we currently know about AI in Libraries, Archives & Records Management. It shows where things are moving so you can choose which way in suits you.
The job market
People and jobs. No documented structural shortage comparable to cyber/medicine; not a simple credential-oversupply surplus. UK Information Workforce Data Consortium mapping data shows 40% of the workforce with 20+ years of experience (an ageing-workforce signal independent of AI) and a bimodal pay distribution (30% under GBP22k, 20% over GBP40k), consistent with a field restructuring around a growth pocket (digital preservation/governance) and a contracting pocket (funding-constrained public/heritage services).
The main limit on getting in. Distinct from most fields in this corpus: the dominant binding constraint on entry into the public library and heritage-archive segments is public funding austerity, not AI displacement. Credential_gate (the postgraduate-qualification requirement) is a secondary binding constraint field-wide, sharpened by the 2026 withdrawal of Level 7 apprenticeship funding.
Entry-level jobs. Contracting in the public library and heritage-archive entry segments (funding austerity; 2026 apprenticeship-funding withdrawal); growing in the digital preservation and records/information-governance segment (UK GDPR universal applicability; expanding digital collections). Net field-level trajectory called contracting because the two largest traditional entry routes (public library assistant, archives trainee) are both narrowing even as the smaller, faster-growing DP/governance segment offsets some of this.
Cost of training. Dominant entry route is a postgraduate CILIP- or ARA-accredited qualification; entry-level pay is low (GBP17k-22k trainee-level). Real debt-to-income tension for students weighing the credential investment.
The single biggest determinant of a young person's entry prospects in this field is which corner they enter, not AI deployment in general: public funding austerity is compressing entry in public libraries and heritage archives far more than AI is, while AI is a genuine but secondary factor -- and in the digital preservation/records-governance corner, AI-adjacent demand (GDPR compliance, growing digital collections) is actively expanding entry, not compressing it.
Interchange
People drawn to Libraries, Archives & Records Management are often drawn to these.
End of the line
That is a conversation about this field, not a test. Or keep going on your own. Both doors stay open.