AI Reconstructs 3,000-Year-Old Babylonian Hymn From 30 Clay Fragments

Researchers from LMU Munich and the University of Baghdad have successfully reconstructed a 3,000-year-old Babylonian hymn using artificial intelligence. This hymn, known as the Hymn of Babylon, had been lost for over 2,000 years and was last studied in 100 BC. The reconstruction was achieved by piecing together 30 clay fragments that had been excavated over the years. The team utilized a specialized AI program to analyze and match text fragments based on combinations of cuneiform signs, a method detailed in a methodology paper from last year.

The researchers employed natural language processing approaches to indicate that fragments belonged to a single text. They worked from the Electronic Babylonian Library Platform, which contains 1,402 manuscripts, using n-gram matching as their primary method of reconstruction. Other methods included vocabulary overlapping and searching for the longest common strings of text. The rediscovered poem was significant enough to be taught as part of Babylon’s curriculum, suggesting that the author was likely a member of Babylon’s priestly class. The poem includes a section describing priests as the “free citizens” of Babylon and celebrates the city’s natural resources, beauty, acceptance of foreigners, and support for the poor.

The hymn reads, “The foreigners among them they do not humiliate. The humble they protect, the weak they support. Under their care, the poor and destitute can thrive. To the orphan they offer succour and favour.” This discovery highlights the increasing use of AI in reconstructing damaged or fragmented texts, a trend that has become indispensable to researchers. While languages like Akkadian and Sumerian are still underrepresented in large language models, efforts are being made to improve computational tools for ancient Near Eastern studies.

The reconstruction of ancient texts using AI has become increasingly common among scholars. In 2023, a 21-year-old student made headlines for developing a machine learning algorithm to decipher ancient Greek letters inside a sealed scroll from Herculaneum. This demonstrates the potential of AI in archaeology, where deciphering cuneiform, the wedge-shaped script pressed into clay with reed styluses, is a task that once took lifetimes. Now, machine learning tools can match symbols, syntax, and context across thousands of broken tablets, speeding up the process exponentially.

By training algorithms to recognize recurring phrases and linguistic patterns, the team created an AI-powered archaeologist capable of sifting through mountains of textual rubble and identifying matching pieces. This project shows the potential of digital tools to bring us closer to our past, allowing us to listen again to voices long silent and find poetry in clay. The rediscovery of the hymn brings the spirit of Babylon back into the light, telling of a city that knew beauty, reverence, and complexity, and of people who believed that to write was to remember—and that to remember was sacred.