AI Is Opening the Archive Doors
When I visited Bruges in the late 1970s, someone casually mentioned that there were still rooms of historical papers that hadn’t really been examined. Whether that was literally true or simply reflected how much work remained to be done, the comment stuck with me for nearly fifty years.
At the time, it wasn’t difficult to understand why.
Historians naturally focused on the biggest discoveries. If you found another work by Jan van Eyck or Simon Bening, everyone wanted to know. If a document rewrote the history of kings or wars, it was immediately important. But what about the everyday records? Merchant account books. Legal disputes. Apprenticeship contracts. Personal letters. Household inventories. They often waited patiently on the shelves because there simply weren’t enough trained people—or enough lifetimes—to read everything.
Now, for the first time in history, that may be changing.
Artificial intelligence is learning to read historical handwriting. Not perfectly, and certainly not without human experts checking its work, but well enough to produce a first transcription in minutes instead of weeks. Suddenly thousands of manuscripts can become searchable instead of remaining images that only a specialist can decipher.
Think about what that means.
Instead of spending months turning page after page looking for one reference to Bruges or Ghent, a researcher may soon be able to search an entire collection in seconds. Instead of wondering whether a woman’s name appears only once, we may discover she appears dozens of times in records that no one had connected before.
As someone fascinated by material culture, this excites me more than almost anything happening in historical research today.
I’ve often said that history isn’t just about kings, queens, and famous artists. It’s also about the anonymous craftsman who mixed pigments, the merchant who ordered parchment, the apprentice who carried supplies across town, and perhaps even the itinerant scribe who earned a living by writing for others.
Those are the people who rarely make the history books.
They are also the people most likely to be hiding in those forgotten documents.
For years I’ve written about my fictional persona, Jehanne Bening. She lives alongside famous artists, but most of her life happens in the ordinary spaces between great events. She buys supplies. She travels. She accepts commissions. She meets merchants. She worries about finding work.
The more I learn, the more I realize that AI may someday uncover real women whose lives looked remarkably like hers.
Perhaps somewhere in an overlooked account book is the payment to a woman for copying a manuscript. Perhaps a merchant’s letter mentions an illuminator traveling farther than anyone expected. Perhaps a guild record preserves the name of someone invisible for five hundred years.
Those discoveries won’t replace careful scholarship. They will require even more expert historians to verify, interpret, and place them into context.
But AI can help open the door.
As an SCA member, I’ve always believed that understanding the ordinary lives of people in the Middle Ages and Renaissance brings us closer to the past than memorizing another list of kings. Every object, every scrap of paper, every handwritten note tells a story.
For decades, many of those stories have simply been too difficult to find.
Now, for the first time, we have a tool that may help us hear those forgotten voices again.
I suspect we’re standing at the beginning of a golden age for historical research. Not because AI will replace historians, but because it will help historians ask questions that were impossible only a few years ago.
Who knows? Somewhere in an archive in Bruges—or Ghent, or Venice, or Nuremberg—there may already be a document waiting to tell us something no one has known since the sixteenth century.
And that, to me, is every bit as exciting as discovering a lost masterpiece.
Image information: the photographer was Jean-Christophe BENOIST. It is posted under the CC BY-SA 3.0 http://creativecommons.org/licenses/by-sa/3.0/, via Wikimedia Commons