New AI Tool Promises Major Leap in Understanding Ancient History

A groundbreaking artificial intelligence tool could dramatically enhance how historians interpret ancient human history, researchers say.

While AI has previously been used to restore sections of damaged Roman scrolls, a new system named Aeneas goes even further. This tool can reconstruct missing words in Roman inscriptions found on monuments and artifacts, as well as estimate their age and geographic origin.

However, given that AI can misinterpret even contemporary texts, some experts have expressed concerns that over-reliance on such systems might lead to inaccuracies in historical research.

Despite these concerns, Professor Dame Mary Beard of the University of Cambridge has hailed Aeneas as potentially “transformative” for historical studies. She notes that unlike traditional methods—reliant on scholars’ memory, judgment, and reference databases—Aeneas opens new possibilities for decoding ancient texts at scale.

Developed by Dr. Thea Sommerschield (Nottingham University) and Dr. Yannis Assael (Google DeepMind), Aeneas draws from a vast archive of 176,000 Roman inscriptions, many with accompanying images. The system analyzes language, grammar, and context by comparing incomplete inscriptions to known historical examples, known as “parallels.”

“Inscriptions are among the oldest written records we have,” said Dr. Sommerschield. “They’re critical primary sources, but most are degraded, incomplete, or undated—like solving a massive jigsaw puzzle with most pieces missing.”

The process of interpreting inscriptions has traditionally been slow and labor-intensive. Aeneas automates much of this work in seconds, offering historians quick access to potentially relevant parallels they might otherwise miss.

The tool was recently tested on a renowned Roman inscription at the Temple of Augustus in Ankara, Turkey. Known as the Res Gestae Divi Augusti, it details the life of Emperor Augustus and is central to Roman historical studies. Aeneas was able to narrow its likely date to two timeframes: 10–20 CE and 10–1 BCE—matching the current scholarly consensus.

In trials involving 23 historians, those working alongside Aeneas produced more accurate interpretations than either the AI or historians working alone. Dr. Sommerschield highlighted that the system not only improved accuracy but also introduced new insights previously overlooked.

Still, the developers stress that Aeneas is a support tool—not a replacement for human expertise.

“AI won’t always get it right,” said Dr. Assael. “It’s up to historians to interpret the AI’s suggestions and determine what makes the most sense based on broader evidence.”

Aeneas builds on earlier work where AI was used to digitally “unwrap” and read a charred scroll from Herculaneum using X-ray imaging and machine learning, demonstrating how emerging technologies can help recover long-lost texts.

By accelerating analysis and revealing previously unnoticed patterns, Aeneas could redefine how we explore the ancient world—faster, and perhaps even smarter, than ever before.

ALSO READ : Scholarships Available for International Students at Zhejiang University’s Haining Campus