Artificial Intelligence Uncovers Striking Structural Parallels Between Ancient Eurasian and African Scripts

Researchers at San Diego State University have utilized artificial intelligence to identify significant structural similarities between the ancient Ethiopic alphabet and several scripts from the Caucasus region, including Armenian, Georgian, and Caucasian Albanian. By training a machine learning model on over 28,000 images of Ethiopic characters, the team bypassed the subjectivity of traditional visual inspections. The AI focused exclusively on the geometric properties of the letters, such as angles, curves, and lines, without any access to historical, cultural, or linguistic context. The results revealed that the Armenian alphabet shares the highest degree of structural resemblance to Ethiopic, a finding that stands out when compared to the Latin alphabet, which showed minimal similarity.

This computational approach provides a reproducible, mathematical framework for analyzing the evolution of writing systems. The findings are particularly compelling because they align with historical timelines suggesting potential contact between these cultures. During the early fifth century, when the Armenian alphabet was developed by Mesrop Mashtots, travelers from the Horn of Africa were known to frequent regions like Syria and Egypt. Because the AI reached its conclusions based solely on visual data, the convergence of computational geometry and historical records offers new evidence for potential cross-cultural influence in the development of ancient writing. This study demonstrates the growing utility of machine learning in providing objective insights into complex historical mysteries.