by 
Using University of Michigan AI, more than 55,000 pages of Board of Regents proceedings have been rebuilt into searchable datasets, giving researchers new ways to examine decisions dating back to 1837.
ANN ARBOR, Mich. — Nearly two centuries of University of Michigan history are getting a technological overhaul.
The University of Michigan Library announced Aug. 13 that it has reprocessed more than 55,000 pages from 71 volumes of Board of Regents proceedings, using artificial intelligence and improved optical character recognition to make the records easier to search and study.
The records cover 1837 through 2023, documenting 186 years of decisions at one of Michigan’s largest public institutions.
University of Michigan AI Project Rebuilds Old Records
The project did not digitize the Regents proceedings for the first time. U-M Library began putting digital versions online more than 20 years ago.
The new project, launched in 2025 with the Institutional Data Initiative at Harvard Law School Library, restructures that existing digital collection.
More than 55,000 scanned pages have been divided into 1,755 individual meetings. The pages received improved text recognition and information about their layouts.
Researchers can also search metadata identifying meeting dates, attendees and subjects such as university leadership, degree programs, new schools and colleges, construction projects, fundraising and strategic planning.
That changes how historians can approach the collection.
Instead of working through thousands of scanned pages looking for a particular subject, researchers can narrow their search before consulting the original documents.
AI Helps Researchers Search 186 Years of History
Artificial intelligence was used to generate summaries of individual meetings and broader summaries covering each decade.
Those summaries provide a starting point, not a replacement for the historical record.
AI can misunderstand older terminology, overlook context or reduce complicated debates to overly simple descriptions. Historians and journalists still need to verify information against the original proceedings.
But with 186 years of records, AI can help researchers determine where to look.
The collection begins in 1837, the year Michigan became a state. Regents proceedings since then have documented decisions involving academic programs, university leadership, buildings, finances and institutional policy.
That creates opportunities to examine how university leaders responded as Michigan and the nation changed around them.
Researchers could search the records for periods surrounding the Civil War, two world wars, the Great Depression, changes in civil rights and major shifts in Michigan’s economy.
Historical Records Become Research Data
The project goes beyond improving a website.
U-M and its partners have produced three datasets from the Regents records. The datasets, technical documentation and processing tools are being made openly available to researchers. Individual meeting records are also being released in EPUB format using modern accessibility standards.
That means scholars can analyze the proceedings computationally rather than reading them only as individual documents.
Researchers might track how frequently certain subjects appeared, examine changes in university priorities or study how institutional language changed over decades.
The Regents collection is also part of a broader modernization effort at U-M Library. A new digital repository under development is scheduled for completion in 2027 and is intended to improve access to the library’s digital collections.
For historians, the significance is straightforward.
Artificial intelligence has not changed what University of Michigan leaders decided during the past 186 years.
It has made 55,000 pages documenting those decisions considerably easier to find, search and question.
How is the University of Michigan using AI with historical records?
The University of Michigan Library and Harvard Law School Library’s Institutional Data Initiative used newer AI technology to enhance the university’s existing digital Board of Regents collection. The project includes improved text recognition, metadata and AI-generated summaries.
How many University of Michigan Regents records were processed?
The project processed more than 55,000 scanned pages from 71 volumes and organized them into 1,755 individual meetings.
How far back do University of Michigan Regents records go?
The Board of Regents has documented important governance decisions since 1837, the year Michigan became a state. The AI-enhanced project covers 186 years of university history.
Did AI create the University of Michigan historical records?
No. The historical records already existed, and U-M Library began making digital versions available more than two decades ago. The new project reprocessed the existing collection using improved OCR, organization, metadata and AI-generated summaries.
Can researchers use the University of Michigan datasets?
Yes. U-M says the project’s datasets and related research materials are being made openly available to scholars and researchers. The original Regents digital collection is dedicated to the public domain under a Creative Commons CC0 license.
Related Posts
