Deep Past Initiative Partners with Kaggle on Prize Challenge to Translate 4,000-Year-Old Tablets

DPI and Kaggle have launched a $50k AI challenge to decode 4,000-year-old business records, a major computational frontier for Machine Learning

NEW HAVEN, CT, UNITED STATES, December 16, 2025 /EINPresswire.com/ -- In a joint effort by experts in ancient history and machine learning, the Deep Past Initiative is partnering with Kaggle to host a competition that can solve the question, can AI decode 4,000-year-old business records? A first of its kind, the Deep Past Challenge competition runs from December 16, 2025, to March 23, 2026, with a $50,000 prize pool for the most successful model capable of translating millennia old clay tablets from Akkadian into English.

The Deep Past Challenge seeks to develop new machine learning tools that can decipher and read clay tablets inscribed in the world’s first system of writing, the cuneiform script. There is no harder script than cuneiform for a computer to decipher. More complex than Chinese Hanzi and more ambiguous than Egyptian hieroglyphs, teaching machines to translate a three-dimensional, multi-valent form of writing is a major technological milestone. According to co-founder Dr. Gojko Barjamovic:

“If we can teach a computer how to read and translate cuneiform, there is no other language or script in the world it will not be able to master. Reading cuneiform is an ultimate computational frontier.”

Before the emperors of Rome or philosophers of Athens, merchants from the city of Assur built a trade network connecting the world from China to Europe. Their voices are preserved in four-thousand-year-old contracts, letters, and loans, written in clay, bearing witness to a world of commerce, conflict, and kinship. The Deep Past Challenge aims to resurrect the world of Assur and its trade network, flush with headstrong sons, pious fathers, diplomatic daughters, and vengeful relatives. For co-founder Dr. Agnete Lassen, Assur’s history is uniquely suited for such a task:

“These tablets are raw glimpses into everyday life, never intended to stand the test of time. But now they are a unique lens through which we get to experience a world long forgotten through the eyes of people whose voices are often sidelined by the filter of history, such as women, children, and entrepreneurs.”

This challenge will push the limits of current AI capabilities, using ancient records as a test-case for machine learning. Co-founder Dr. Ruchir Agarwal, sees this opportunity to democratize one of the largest written datasets to survive from the ancient world:

“Throughout human history, writing has been our primary means of preserving knowledge, yet vast portions of humanity’s recorded heritage remain inaccessible. Our vision is to expand access to humanity’s written past by making primary sources from across civilizations more accessible to researchers, educators, and the world alike.”

The Deep Past Initiative is a non-profit organization designed to develop and deploy modern technological innovation towards the study of the ancient past. It joins leading data scientists, ancient historians and experts in natural language processing and network analysis to decipher and publish ancient data in easily accessible formats for scholars and the general public. Kaggle is a data science competition platform and online community for data scientists and machine learning practitioners under Google LLC. The effort is financially supported by leading algorithmic trading firm XTX Markets.

For more information, visit deeppast.org and follow along on @DeepPastAI on LinkedIn and X.

CONTACT: press@deeppast.org

SOURCE Deep Past Initiative

Andrew Deloucas
The Deep Past Initiative
press@deeppast.org
Visit us on social media:
LinkedIn
X

Legal Disclaimer:

EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.