DOI: To Be Assigned
John Swygert
June 23, 2026
Abstract
Universities and colleges across the United States hold vast bodies of archived scholarly, scientific, cultural, biographical, agricultural, historical, literary, and technical material that remain physically preserved but practically inaccessible. These records often include unpublished papers, correspondence, field notes, institutional reports, photographs, manuscripts, laboratory records, lecture notes, faculty collections, student research, oral histories, estate donations, and regional scientific observations. Many such collections are known only to specialized archivists, local researchers, descendants, or institutional insiders. In the age of artificial intelligence, this condition is no longer merely inconvenient; it is a national knowledge failure. This paper proposes a federally funded AI-Assisted Collegiate Archive Digitization Program through which accredited colleges and universities may apply for funding, technical assistance, ethical guidance, and publication infrastructure to digitize, transcribe, metadata-tag, preserve, and publicly release dormant archival collections with proper attribution and institutional credit. The program should be administered either through an expanded partnership among the National Archives and Records Administration, the National Endowment for the Humanities, the Institute of Museum and Library Services, the Library of Congress, and the National Science Foundation, or through a newly created National Office for AI Cultural and Scientific Memory. The purpose is not to replace archivists, historians, librarians, scholars, or families, but to give them tools and funding powerful enough to meet the size of the national problem. A recent example from European medieval manuscript transcription demonstrates that specialized AI systems can unlock tens of thousands of historical documents at a speed unimaginable under manual methods alone. The United States should treat its dormant university archives as a strategic memory reserve: a body of hidden knowledge that belongs not only to institutions, but to science, history, descendants, communities, and future generations.
Introduction
Every generation leaves behind more than it publishes. Behind every famous book, scientific paper, agricultural report, invention, lecture, excavation, laboratory notebook, field journal, and private letter is a much larger shadow archive: the surrounding material that was saved, boxed, cataloged, donated, inherited, misplaced, or stored in university collections without ever becoming fully accessible to the public.
This is especially true in colleges and universities. American academic institutions hold the working memory of thousands of scholars, scientists, teachers, explorers, clergy, doctors, agricultural experts, artists, engineers, public servants, and regional historians. These collections may be preserved in the legal and physical sense, but preservation without discoverability is only partial preservation. A box in a university archive may be safe from weather, fire, and casual destruction, yet remain functionally invisible to the descendant, independent researcher, disabled scholar, small-town historian, student, inventor, or scientist who cannot travel across the country to inspect it.
This paper argues for a federal program to help colleges and universities digitize and publish dormant archival records using artificial intelligence under strict standards of attribution, accuracy, transparency, preservation, and public access.
The argument is both technological and moral. It is technological because AI has reached a point where handwriting recognition, optical character recognition, metadata generation, document clustering, translation support, entity recognition, and archival search can radically accelerate the work of making historical material usable. It is moral because the memories of prior generations are disappearing from living human custody. The Greatest Generation is nearly gone. The Baby Boomer generation is aging rapidly. Their parents, teachers, mentors, and institutions left behind records that are often scattered, undigitized, and unknown. If those records are not made accessible now, the nation may lose not only paper, but context.
The purpose of such a program should not be nostalgia. It should be national knowledge recovery.
The Problem: Preserved But Inaccessible
Many collegiate archives are not neglected in the crude sense. They may be housed by skilled archivists and librarians who care deeply about preservation. The problem is scale. Archivists are underfunded. Collections are enormous. Digitization is expensive. Metadata takes time. Rights review can be complicated. Handwriting may be difficult. Scientific records may require subject-matter knowledge. Older photographs may lack identification. Donor files may be split across institutions. Small colleges may hold regionally important material but lack the budget to process it. Major universities may have millions of pages waiting behind higher-priority projects.
The result is a strange contradiction: the records exist, but access does not.
For families and descendants, this can be devastating. A person may know that a great-grandparent, professor, scientist, author, minister, civil servant, or public intellectual left papers to a university, yet still have no practical way to inspect them. The archive may require travel, permission, appointment scheduling, handling limitations, photography rules, fees, or specialized research time. For a healthy and funded researcher, those obstacles may be manageable. For a disabled person, elderly descendant, independent scholar, rural researcher, working-class family, or student without travel resources, they can be absolute barriers.
The same barrier harms science. Dormant records may contain agricultural observations, geological surveys, correspondence with known historical figures, regional climate notes, early medical observations, unpublished manuscripts, local histories, maps, field sketches, archaeological references, plant studies, institutional memory, and interdisciplinary connections never indexed by modern databases. These records may not individually appear urgent to a university budget committee, but collectively they form a hidden national library of observation.
What is not searchable is not truly available.
What is not digitized is not equally accessible.
What is not attributed is not properly remembered.
A Representative Case: R.J.H. DeLoach And The Hidden Wealth Of Collegiate Holdings
A representative example may be found in the archival footprint of R.J.H. DeLoach, an American scholar and agricultural figure whose work and correspondence are associated with more than one university collection. His life intersected with major intellectual, agricultural, and cultural currents of his time, including friendships or associations with figures connected to the Vagabonds circle of Thomas Edison, Henry Ford, Harvey Firestone, and John Burroughs.
The specific importance of DeLoach is not merely family pride or biographical curiosity. His case illustrates a broader national pattern. A historically significant person may leave behind material divided among institutions in different states. Some records may be in Georgia, others in Chicago or elsewhere. A descendant may know the material matters but be unable to travel, gain access, photograph, transcribe, and contextualize it personally. The materials may contain letters, unpublished writing, scientific work, institutional history, or connections to larger historical movements. Yet until those records are digitized, indexed, and published with proper credit, they remain effectively locked behind geography and institutional capacity.
There are thousands of similar cases across the United States.
The nation’s archives do not consist only of presidents, generals, famous inventors, and bestselling authors. They also consist of professors, field researchers, extension agents, pastors, botanists, regional scientists, school founders, translators, local physicians, civil engineers, women who kept community records, librarians, mapmakers, amateur naturalists, veterans, musicians, and families who donated papers believing someone would someday use them.
That “someday” now requires a system.
Why Artificial Intelligence Changes The Archive Question
Digitization once meant scanning. Scanning was necessary, but it did not solve the full problem. A scanned page is an image. It may preserve the look of a document, but it does not automatically make the text searchable, extract names, connect dates, identify subjects, cluster related materials, or translate older language forms. The next phase of archival access requires machine-readable transformation.
Artificial intelligence makes this possible, but only if used carefully.
Recent developments in historical document processing show that specialized systems can transcribe large bodies of difficult handwritten or early textual material. In one reported case, the CoMMa project used specialized AI to transcribe tens of thousands of medieval manuscripts in Old French and Latin within months. The importance of this example is not merely speed. It is method. The project emphasized character-level recognition, open-source tools, and fidelity to the original text rather than modernizing or “correcting” the documents into something historically false.
This principle should guide American collegiate archive digitization. AI should not be used to rewrite the past. It should be used to expose the past, mark uncertainty, preserve original form, and make human review more efficient.
The proper archival AI model is not:
“Let the machine decide what the document meant.”
The proper model is:
“Let the machine help reveal what is there, while preserving the original, marking uncertainty, and keeping human accountability.”
AI can assist with:
- Optical character recognition for typed documents.
- Handwritten text recognition for letters, journals, notebooks, and field notes.
- Metadata extraction, including names, dates, places, institutions, topics, and document types.
- Language identification and translation support.
- Document clustering across collections.
- Duplicate detection.
- Image enhancement for faded pages.
- Search indexing.
- Accessibility tools for disabled users, including screen-reader-compatible text.
- Public-facing summaries clearly labeled as AI-assisted and human-reviewed.
- Scholarly citation formatting.
- Rights and privacy triage, flagging material that may require human review before release.
- Cross-institutional matching, where related collections exist at multiple universities.
These tools do not eliminate archival labor. They make it possible to apply archival labor where it matters most: review, judgment, correction, description, rights decisions, contextualization, and preservation ethics.
Existing Federal Models And Their Limits
The United States already has federal institutions that support preservation, access, and publication of historical records. The National Endowment for the Humanities funds humanities projects across colleges, universities, libraries, museums, research institutions, and nonprofits. The National Historical Publications and Records Commission, housed within the National Archives and Records Administration, has supported projects to digitize nationally significant records and publish collaborative digital editions. The Institute of Museum and Library Services supports libraries and museums, including projects involving access, preservation, and institutional capacity.
These models prove that the federal government already recognizes preservation and access as public goods. However, they are not sufficient for the AI era.
The existing grant landscape is fragmented. Many programs are competitive, limited, discipline-specific, capped in scale, or not designed for mass collegiate participation. Some programs support digitization, others preservation, others editions, others museum or library work. What is missing is a national, unified, AI-assisted program designed specifically to help colleges and universities identify, prioritize, digitize, transcribe, attribute, and publish dormant archival holdings at scale.
The country does not merely need more grants.
It needs an archival memory agenda.
Proposed Program: The Federal Collegiate Archive Digitization And Access Program
This paper proposes the creation of the Federal Collegiate Archive Digitization and Access Program, abbreviated here as FCADAP.
The program would provide direct federal funding, technical tools, training, standards, and publication infrastructure to accredited colleges and universities for the digitization and public release of archival collections.
Its central mission would be:
To recover, preserve, digitize, transcribe, attribute, index, and publish dormant collegiate archival records of scientific, cultural, historical, educational, technical, and civic significance for the benefit of the American public and future generations.
The program should be built around eight principles.
First, public access. Materials digitized with federal funds should be made freely accessible online unless restricted by privacy, donor agreements, copyright, cultural sensitivity, national security, or preservation concerns.
Second, proper credit. Every released item should clearly credit the original creator when known, the holding institution, the collection name, donors where appropriate, funding sources, archivists, editors, and AI-assisted processing where used.
Third, original preservation. Digitization should never replace physical preservation. The original material remains the authoritative object.
Fourth, AI transparency. Any AI-generated transcription, metadata, translation, or summary should be labeled as such. Confidence levels and human review status should be visible.
Fifth, human review. AI should accelerate processing, but archivists and subject-matter reviewers should remain responsible for release decisions, corrections, and contextual framing.
Sixth, distributed participation. The program should serve major universities, small colleges, historically Black colleges and universities, tribal colleges, Hispanic-serving institutions, land-grant universities, religious colleges, community colleges with unique regional holdings, and specialized institutes.
Seventh, cross-institutional connection. The program should help identify related collections across institutions so that divided archives can be digitally reunited without requiring physical transfer.
Eighth, long-term preservation. Files, metadata, transcripts, and access platforms should use open standards and sustainable storage plans.
Administrative Structure
There are two plausible administrative paths.
The first is an interagency model. Under this approach, the program would be administered through a formal partnership among the National Archives and Records Administration, the National Endowment for the Humanities, the Institute of Museum and Library Services, the Library of Congress, and the National Science Foundation. Each agency would contribute expertise. NARA would provide archival standards. NEH would support humanities significance and scholarly access. IMLS would support library and museum implementation. The Library of Congress would provide national bibliographic and digital preservation expertise. NSF would support AI infrastructure, technical research, and computational standards where scientific collections are involved.
The second is the creation of a new federal office: the National Office for AI Cultural and Scientific Memory. This office would not replace existing agencies. Instead, it would coordinate AI-assisted preservation and access efforts across them. Its mandate would be to ensure that the national AI agenda includes not only defense, commerce, automation, and productivity, but also memory.
A nation that invests in AI to predict markets, automate business, and generate entertainment should also invest in AI to preserve the knowledge that made the nation possible.
Grant Categories
The program should include several grant categories.
Planning Grants
Planning grants would help institutions survey collections, identify candidates for digitization, estimate rights issues, prepare metadata, and develop workflows. These would be especially important for smaller institutions that know they hold valuable material but lack the staff to prepare a full digitization proposal.
Pilot Digitization Grants
Pilot grants would fund smaller initial projects, such as a single faculty collection, regional scientific archive, correspondence series, photographic collection, or field notebook set. These projects would test institutional workflow and produce public examples.
Major Collection Grants
Major grants would fund large-scale digitization of nationally or regionally significant collections. These may include multi-year projects involving hundreds of boxes, thousands of images, or interdisciplinary holdings.
Cross-Institutional Reconstruction Grants
Many important figures have papers divided across multiple universities. This category would support digital reunification: not physically merging collections, but creating linked digital portals where related materials can be discovered together.
AI Tooling And Training Grants
These grants would fund training for archivists, librarians, students, and faculty in AI-assisted transcription, metadata review, quality control, rights triage, and digital publication.
Student Archive Corps Grants
A national Student Archive Corps would employ undergraduate and graduate students to assist with scanning, metadata review, transcription correction, local history research, and public outreach. This would create jobs, train future archivists, and connect students to their own institutional history.
Family And Descendant Access Grants
Some projects should include a mechanism for descendants or community stakeholders to request prioritization or remote access. A family member who knows of a significant but inaccessible collection should have a way to ask the holding institution to consider digitization under the program.
Technical Standards
The program should require technical standards that balance access, accuracy, and long-term preservation.
Each digitized item should include:
- A high-resolution preservation image.
- A web-accessible image.
- A machine-readable transcript when possible.
- A human-readable description.
- Creator attribution.
- Holding institution credit.
- Collection and box/folder metadata.
- Rights status.
- AI-processing disclosure.
- Human-review status.
- Persistent identifier.
- Recommended citation.
- Download options where legally permitted.
- Accessibility compliance.
AI transcriptions should be stored in a way that preserves uncertainty. Unclear words, damaged text, abbreviations, and illegible passages should be marked rather than silently corrected. This is crucial. A bad AI system can damage history by making false confidence look authoritative. A good AI system makes uncertainty visible.
The program should therefore require multiple layers of text:
- Original image.
- Raw AI transcription.
- Corrected transcription if human-reviewed.
- Normalized reading version only if clearly labeled.
- Editorial notes where needed.
This layered model protects both accuracy and usability.
Rights, Privacy, And Ethical Limits
Not every archived document should be immediately published. A serious federal program must address rights and privacy.
Some collections may include living persons’ private information, student records, medical information, restricted donor material, culturally sensitive materials, unpublished copyrighted works, Indigenous knowledge, confidential correspondence, or legally protected records. AI can help flag such materials, but humans must decide release rules.
The program should require rights review before publication. It should also encourage tiered access:
- Open public access.
- Researcher access after registration.
- Restricted access requiring permission.
- Metadata-only public listing.
- Closed records until a future date.
The existence of a record can often be made discoverable even when the full document cannot be published. This matters because hidden collections remain invisible to scholarship. A metadata-only listing may still help a researcher locate relevant material and request permission.
Proper Credit And Attribution
The phrase “proper credit” must be central to the program.
Digitization can accidentally strip documents from their context. A letter copied online without collection information becomes an orphan. A photograph without creator, donor, institution, and date loses scholarly value. A transcription without uncertainty markers may mislead. A downloaded image without citation guidance may circulate without credit.
Every item published through the program should include a visible credit line.
A recommended format might be:
Creator, Title or Description, Date if known, Collection Name, Holding Institution, Digitized through the Federal Collegiate Archive Digitization and Access Program, Persistent Identifier.
Where descendants, donors, or estates are relevant, the record should include donor or provenance information when appropriate and legally allowed.
The program should also require citation tools so students, scholars, journalists, descendants, and public users can properly cite archival material.
AI itself should not receive authorship credit. AI should be credited as a tool, not as a creator. The human creator, collection, institution, archivist, editor, and funding program must remain visible.
Why This Matters For Science And Technology
The archival problem is often framed as a humanities issue. That framing is too narrow.
Dormant collegiate archives may contain scientific and technical information valuable to modern research. Agricultural colleges may hold crop observations, pest records, soil studies, climate notes, seed experiments, livestock data, extension correspondence, forestry records, and regional environmental observations. Medical schools may hold historical public health records. Engineering schools may hold bridge, water, mining, rail, or energy documentation. Geology departments may hold field notebooks. Anthropology and archaeology departments may hold site notes, photographs, maps, and correspondence. Music departments may hold oral histories and regional recordings. Education departments may hold school reform records. Religious colleges may hold missionary reports, language materials, and community histories.
The future may need old data.
Climate science, land use, agriculture, epidemiology, genealogy, urban planning, cultural history, linguistics, environmental restoration, and the history of technology may all benefit from records now sitting in boxes.
The assumption that old material is obsolete is one of the great errors of modernity. Old records may preserve baselines. They show what rivers, farms, forests, communities, diseases, institutions, and languages looked like before later transformations. They may reveal abandoned methods, forgotten failures, early warnings, or overlooked insights.
AI makes old records newly searchable. That means old records can become new data.
The Generational Emergency
The United States is also facing a generational memory emergency.
The people who can identify faces in photographs are dying.
The people who know why a professor’s papers matter are retiring.
The people who remember what was in a laboratory, department, farm, chapel, office, or local movement are aging.
The people who can explain family provenance are passing away.
This is not sentimental. It is evidentiary. Context dies with people. Once the person who can identify the handwriting, the location, the nickname, the unmarked photograph, or the institutional relationship is gone, future researchers may have only fragments.
The nation should act while living memory can still assist archival recovery. AI can help process the material, but living humans are needed to explain it. Therefore, the proposed program should include an oral-history and context-capture component. When collections are digitized, institutions should be encouraged to interview descendants, retired faculty, alumni, archivists, local historians, and community members who can provide context.
The box matters.
The story of the box matters too.
Public Benefit
The public benefit of this program would be enormous.
Students would gain access to primary sources from their own institutions.
Families would recover lost ancestry and intellectual inheritance.
Independent scholars would no longer be excluded by travel costs.
Disabled researchers would gain remote access.
Small towns would recover local history.
Scientists would gain old observations and data.
Universities would receive proper credit for their holdings.
Archivists would receive tools and staffing.
AI research would gain high-value, ethical public-interest use cases.
The nation would recover part of its hidden memory.
This program would also strengthen trust in universities. Many citizens see universities as distant or elite institutions. Public archival access would show universities as guardians of shared memory. A land-grant university digitizing agricultural correspondence, a historically Black college digitizing civil rights papers, a small religious college digitizing missionary language records, or a regional university digitizing coal, rail, or forestry archives would all demonstrate public value beyond tuition and degrees.
Legislative Proposal
Congress should authorize a Federal Collegiate Archive Digitization and Access Act.
The Act should:
- Establish a national AI-assisted collegiate archive digitization program.
- Authorize grants to accredited colleges and universities.
- Permit partnerships with libraries, museums, tribal archives, state archives, historical societies, and nonprofit repositories.
- Require free public access to federally funded digitized materials unless legally or ethically restricted.
- Require proper attribution, institutional credit, and persistent identifiers.
- Require AI transparency and human-review labeling.
- Fund technical assistance for small and under-resourced institutions.
- Create a national searchable portal linking participating collections.
- Support student employment and training through a Student Archive Corps.
- Support cross-institutional digital reunification of divided collections.
- Require preservation-quality files and sustainable storage plans.
- Encourage oral-history and descendant-context capture.
- Protect privacy, rights, donor restrictions, and culturally sensitive material.
- Require annual public reporting on pages digitized, collections opened, institutions served, and public usage.
- Provide special funding consideration for endangered collections, under-resourced institutions, and materials of national scientific or cultural significance.
The program should begin with pilot funding and then scale. A reasonable first phase could support planning grants and pilot projects at a diverse set of institutions. Later phases could fund major digitization centers, shared AI infrastructure, and a national portal.
Objections And Responses
Objection One: The Cost Would Be Too High.
The cost of not preserving and digitizing these records is higher. Once records decay, context dies, or collections remain unused for another generation, the nation loses knowledge permanently. Federal spending on archival digitization is small compared with the value of recovered scientific, cultural, educational, and historical material.
Objection Two: AI Will Make Mistakes.
Yes. That is why the program must require transparency, confidence levels, original images, human review, and uncertainty markings. AI error is dangerous only when hidden. Human-only processing also contains errors and delays. The solution is not to reject AI, but to use it under archival discipline.
Objection Three: Universities Should Pay For Their Own Archives.
Many universities already do, but the national value of these collections exceeds individual institutional budgets. Federal support is justified when records have public, scientific, cultural, or historical value. The interstate and intergenerational nature of the material makes this a national issue.
Objection Four: Copyright And Privacy Make Publication Too Complicated.
Some publication will be complicated. That is not a reason to abandon the field. The program can support rights review, tiered access, metadata-only records, delayed release, and permission workflows.
Objection Five: There Are Already Grants For This.
There are related grants, but not a unified national AI-assisted collegiate archive program at the necessary scale. Existing programs should be treated as foundations to build upon, not excuses for inaction.
Conclusion
A nation is not only what it builds. It is what it remembers, what it preserves, what it makes accessible, and what it credits properly.
Across the United States, colleges and universities hold immense dormant archives. Some contain the papers of major figures. Others contain the records of people who were never famous but nevertheless observed, built, taught, measured, wrote, served, and preserved knowledge. These materials are not dead. They are sleeping.
Artificial intelligence has given the country a new instrument for waking them.
The question is whether the United States will use AI only to accelerate commerce and entertainment, or whether it will also use AI to recover the hidden memory of its own people, institutions, sciences, and communities.
A federally funded AI-assisted collegiate archive digitization program would honor the dead, serve the living, and prepare knowledge for the unborn. It would help descendants recover family memory. It would help scholars discover neglected evidence. It would help colleges share what they have preserved. It would help science recover old observations. It would help society remember itself.
The records are already there.
The tools now exist.
The missing element is national will.
References
Futura Team. (2026, June 23). How AI just brought 32,000 medieval manuscripts back to life. Futura-Sciences.
Hayun, D., & Confino, H. (2025, November 26). Vast trove of medieval Jewish records opened up by AI. Reuters.
National Archives and Records Administration. National Historical Publications and Records Commission. Digitizing Historical Records Grant Announcement.
National Archives and Records Administration. National Historical Publications and Records Commission. Publishing Historical Records in Collaborative Digital Editions.
National Endowment for the Humanities. Grants.
Philips, J. P., & Tabrizi, N. (2020). Historical Document Processing: Historical Document Processing: A Survey of Techniques, Tools, and Trends. arXiv:2002.06300. DOI: 10.48550/arXiv.2002.06300.