How AI is helping historians better understand our past

So far, the project has yielded surprising results. A pattern found in the data allowed researchers to see that as Europe fractured along religious lines after the Protestant Reformation, scientific knowledge coalesced. Scientific texts printed in places such as the Protestant city of Wittenberg, which became a center of scientific innovation through the work of Reformed scholars, were imitated in centers such as Paris and Venice before spreading across the continent. The Protestant Reformation isn’t exactly an understudied topic, Valleriani says, but a machine-mediated perspective allowed researchers to see something new: “It was absolutely unclear before.” Patterns applied to tables and images started returning similar patterns.

Computers often only recognize contemporary iterations of objects that have a longer history – think iPhones and Teslas, rather than standards and Model Ts.

These tools offer greater possibilities than simply tracking 10,000 tables, Valleriani explains. Instead, they allow researchers to draw conclusions about changing knowledge from patterns in groups of records, even if they’ve actually only looked at a handful of documents. “Looking at two paintings, I can already draw a huge conclusion over 200 years,” he says.

Deep neural networks also play a role in examining even older history. Deciphering inscriptions (known as epigraphy) and restoring damaged examples are tedious tasks, especially when inscribed objects have been moved or lack contextual clues. Specialist historians must make educated guesses. To help, Yannis Assael, a researcher at DeepMind, and Thea Sommerschield, a postdoctoral researcher at Ca’ Foscari University in Venice, have developed a neural network called Ithaca, which can reconstruct missing parts of inscriptions and assign dates and places. to texts. The researchers say the deep learning approach – which involved training on a dataset of over 78,000 registrations – is the first to tackle restoration and attribution together, learning from large amounts of data. data.

So far, according to Assael and Sommerschield, the approach sheds light on decree inscriptions from an important period of classical Athens, which have long been attributed to 446 and 445 BCE, a date that some historians have disputed. As a test, the researchers trained the model on a dataset that did not contain the inscription in question, then asked it to analyze the text of the decrees. This produced a different date. “Ithaca’s predicted mean date for the decrees is 421 BCE, which aligns with the most recent advances in dating and shows how machine learning can contribute to debates around one of the most important in Greek history,” they said by email.

time machines

Other projects propose using machine learning to draw even broader conclusions about the past. This was the motivation behind the Venice Time Machine, one of many “time machines” across Europe that have now been established to reconstruct local history from digitized documents. The Venetian State Archives cover 1,000 years of history spread over 80 kilometers of shelves; the researchers’ goal was to digitize these records, many of which had never been examined by modern historians. They would use deep learning networks to extract information and, by tracing names that appear in the same document through other documents, reconstruct the ties that once held Venetians together.

Frederic Kaplan, president of the Time Machine Organization, says the project has now digitized enough of the city’s administrative documents to capture the texture of the city from centuries past, making it possible to go building by building and identify the families who lived there. at different points in time. “Hundreds of thousands of documents need to be digitized to achieve this form of flexibility,” Kaplan explains. “It’s never been done before.”

Yet when it comes to the ultimate promise of the project — nothing less than a digital simulation of medieval Venice down to the neighborhood level, through networks reconstructed by artificial intelligence — historians like Johannes Preiser-Kapeller, A professor at the Austrian Academy of Sciences who led the study of Byzantine bishops, says the project was unable to deliver because the model cannot figure out which connections are meaningful.