Smithsonian uses AI to connect Revolutionary-era artifacts with historical records

A newspaper advertisement and a museum creamer can tell different parts of the same historical story. Smithsonian researchers are using AI to bring those parts together. New Associated Press reporting details how Revolution Crossroads is finding links across Revolutionary-era collections, while trained historians check the nuances machines may miss.

The initiative grew out of the 250th anniversary of American independence. The Smithsonian and Library of Congress wanted to explore whether AI could reveal more about the communities and everyday lives behind their collections—not simply provide another way to look up well-known historical figures.

Researchers began applying AI models in spring 2026. The work required preparing the underlying material: objects had to be digitized, and some had to be digitized again, to make them more accessible to the systems. That preparation is part of the project, alongside the search for connections.

In its September 4 introduction, the Library of Congress described a partnership bringing together three-dimensional objects, artworks, historic newspapers and archival materials. Its central question was whether machine analysis and pattern recognition could yield insights into human communities. Connecting those different repositories was intended to give others new ways to explore them, too.

One early result makes the approach concrete. Researchers located a silversmith who advertised a creamer in newspapers around 1774. The advertised item was similar to a creamer held by the Smithsonian’s National Museum of American History. The connection joins an object in one collection with evidence of commercial activity in another.

The example is a link to a similar object, not proof that the advertisement described the museum’s particular creamer. AP reported that researchers had drawn connections quickly that previously would have required extensive manual searches across multiple catalogs. The useful result is a research lead spanning collections.

Natalie Buda Smith, the Library of Congress’s director of digital strategy, said this can help researchers who do not know which institution to approach first. Part of a story may sit at the Smithsonian and another part at the library. Finding something in one institution, she emphasized, need not end the search.

“An ideal scenario is that we get to a place where we’re actually adding to the public record,” said Becky Kobberod, the chief digital and innovation officer at the Smithsonian. “Maybe they’re not Thomas Jefferson or George Washington, but they did live in a community and build that community, and that name is now known.”

Becky Kobberod, Smithsonian chief digital and innovation officer, speaking to AP

Kobberod’s team has now identified about 10,000 objects centered on people who lived between 1770 and 1810. The public can search the records online. That gives the initiative a present-day use alongside its larger ambition to recover stories about people whose lives have faded from view over the centuries.

Kobberod said the Smithsonian is not using current frontier models—the leading-edge systems drawing much of the scrutiny around AI. She acknowledged a “healthy fear” of the technology while arguing that the institution should explore the role it can play in its work.

Trained historians review the project’s data. Kobberod pointed to distinctions such as a nickname belonging to a particular person or a phrase that was merely a generic term. Recognizing either can depend on a career spent studying the period, rather than a machine’s ability to find a pattern.