In a groundbreaking development, a team of researchers from Tel Aviv University and Ariel University has harnessed the power of artificial intelligence to unlock the secrets of ancient Sumerian tablets. Their innovative approach is reshaping our understanding of humanity’s earliest narratives, revealing astonishing insights into our shared past.
In May 2023, computer scientists and Assyriologists, led by Guy Gouthier and Shai Gardner, published a paper detailing their creation of a neural network capable of translating Akkadian—a language used in ancient Babylon and Assyria—into modern English. This remarkable achievement allows the AI to read cuneiform script directly from digital representations of the tablets, which date back to between 2500 BCE and 100 CE. The model has proven effective, scoring comparably to existing commercial translation tools for modern languages.
The implications of this technology are monumental. With an estimated 500,000 to 600,000 cuneiform tablets scattered across museums and collections worldwide, a staggering 90% of these relics have never been fully published or translated. These tablets, excavated during significant archaeological expeditions in the 19th and early 20th centuries, have largely remained untouched, waiting for scholars to decipher their contents. The challenge has been the limited number of Assyriologists—fewer than 500 globally—who can tackle the complexities of these ancient languages.
The AI’s introduction is not about replacing human scholars but rather assisting them. By processing the more routine administrative texts, legal documents, and receipts at astonishing speeds, the technology enables experts to focus on the more intricate literary works, such as epic tales and religious texts, where deeper cultural meanings reside.
Fast forward two years, and the project has expanded significantly. Initiatives like the Cuneiform Digital Library Initiative and the Digital Hammurabi Project at Johns Hopkins University are now utilizing these neural translation tools, allowing tens of thousands of previously untranslated tablets to be processed for the first time. Each month, new texts come to light, and the catalog of readable material is finally growing, matching the pace at which these tablets were originally excavated.
What these texts reveal is more profound than many might expect. Long before the stories entered the Hebrew Bible, the Mesopotamians were already documenting the creation of humanity, the great flood, and the divine negotiations surrounding life and death. These narratives, found in the tablets housed in museums like the British Museum and the Louvre, have been waiting for over a century for someone to read them.
The AI tools are not just revealing famous literary works but also the vast supporting documentation that contextualizes these narratives within the civilizations that produced them. The first city, Eridu, is described in the tablets as the birthplace of kingship, while Nippur served as a religious center, yielding approximately 50,000 recovered tablets, many of which contain vital information about Sumerian beliefs and practices.
One of the most significant texts is the Atrahasis Epic, detailing humanity’s creation and the subsequent divine decisions regarding their fate. This narrative, which dates back to around 1700 BCE, describes how humans were created to relieve the gods of labor. However, as humanity multiplied, their noise disturbed the gods, leading to a series of calamities, including a great flood—a story that echoes in many later traditions, including the biblical account of Noah.
The AI translation tools are not just uncovering these narratives; they are also providing context through administrative records, royal correspondence, and temple inventories. This newfound accessibility allows scholars to understand how these stories fit into the broader tapestry of Mesopotamian civilization.
Despite the excitement surrounding these advancements, the AI does have its limitations. While it excels at translating shorter, administrative texts, accuracy diminishes with longer sentences, particularly in literary contexts. The team behind this technology acknowledges that it should be viewed as a tool to assist human scholars rather than a replacement.
As the AI continues to process the vast collection of cuneiform tablets, the next decade of Mesopotamian studies is poised for transformation. What remains to be seen is how the insights gained from the previously unread 90% of these tablets will reshape our understanding of ancient civilizations and their narratives.
The journey of discovery has only just begun, and as more tablets are translated, the ancient voices of the Sumerians and Babylonians will finally be heard. What new revelations await us in the depths of history?
