ROB SCHMITZ, HOST:
During World War II, the Nazis looted more than 600,000 works of art, mostly owned by Jews. In the decades since, there's been an effort to get these works back to their owners, but it's not easy, and roughly a fifth of them have still not been returned. Now, a new AI chatbot might be making that complicated process a little simpler. NPR's Chloe Veltman reports.
CHLOE VELTMAN, BYLINE: Some Jewish families and institutions know that art may have been taken from them during World War II, but reclaiming it is often a long and fraught process.
JOEL GREENBERG: There's no central repository, no database that's telling you where that art is.
VELTMAN: Joel Greenberg is the founder of Art Ashes. The nonprofit helps families track down and recover Nazi-looted art. Greenberg says, on top of locating the missing art, the rightful owners need to prove it was stolen from them in the first place.
GREENBERG: And then to recreate the historical record of what actually happened to the family and the family's art is also very difficult 'cause those records, if they exist, are spread among different repositories around the world.
VELTMAN: There are dozens of archives containing records of these missing artworks in a variety of languages and formats.
GREENBERG: If you could create a system that could scan all the databases that have the Nazi-looted art records, that would be very, very useful.
VELTMAN: A team at Santa Clara University in Silicon Valley is aspiring to do just that.
MICHAEL SANTORO: We are applying AI tools to make it easier to make the kinds of connections that seasoned researchers would have to spend a long time having to do manually.
VELTMAN: This is Michael Santoro. The business school professor says his team's new chatbot, the AI Provenance Assistant, is testing how artificial intelligence could eventually make it easier for people to search archives of stolen art. Right now, Santoro says the tool is only scraping records from one database, that of the Jeu de Paume museum in Paris. The Nazis warehoused tens of thousands of looted artworks and other cultural objects at the museum during the war and kept elaborate records, records that are now available for anyone to search online. But Santoro says his AI tool is more intuitive because it does not require users to deal with drop-down menus and input keywords into boxes.
SANTORO: By enabling ordinary language queries to be made of art databases.
VELTMAN: By ordinary language, Santoro means you can type a question using everyday words into the chatbot's interface online. For example, what artworks belong to Hugo Simon? And the bot will bring up a list of more than 70 belonging to the late Jewish banker and art collector, including stolen works by Picasso and Canaletto.
CARLA SHAPREAU: This project shows promise.
VELTMAN: Carla Shapreau teaches cultural property law at the University of California, Berkeley. She says tools like the AI Provenance Assistant will be more useful when they encompass many databases, not just one. Shapreau also says these chatbots should link to digitized copies of the original archival records, such as confiscation and claim files, since the answers provided by AI and the underlying database are susceptible to errors.
SHAPREAU: It's all speculative if you can't look at the evidence, and the evidence is in the primary sources.
VELTMAN: Some people who specialize in tracking down looted artworks from World War II, like Marc Masurovsky, are less optimistic about AI's potential to streamline his field of expertise.
MARC MASUROVSKY: People find looted art without resorting to AI.
VELTMAN: Masurovsky is a cofounder of the Holocaust Art Restitution Project. He says many archives remain undigitized and are subject to data privacy laws. This drastically limits the pool of data an AI tool can scrape.
MASUROVSKY: Even if they are published on the internet, how are you going to interpret all that alphabet soup? These are really complex cases.
VELTMAN: Meanwhile, the team behind the AI Provenance Assistant at Santa Clara University say they plan to expand their tool to include more databases. They also hope, down the road, to apply AI to finding and repatriating African, Asian and pre-Colombian art as well. Chloe Veltman, NPR News. Transcript provided by NPR, Copyright NPR.
NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.
