Google DeepMind, Historians Deploy AI to Reconstruct Ancient Roman Latin Inscriptions

Aeneas, an AI tool developed by researchers from Google DeepMind and the University of Nottingham, is revolutionizing the study of ancient Latin inscriptions by predicting missing text, estimating dates, and locating origins within the Roman Empire. Trained on the largest combined database of Latin texts, Aeneas can identify linguistic parallels across thousands of inscriptions and reconstruct incomplete fragments with high accuracy, significantly aiding historians who previously relied on time-consuming guesswork. The system's capabilities extend to analyzing famous texts like the Res Gestae Divi Augusti, revealing new insights into Roman imperial political discourse. Aeneas employs generative neural networks to provide not only restorations but also ranked lists of similar inscriptions, speeding up epigraphic research and improving accuracy. Historians have found that using Aeneas makes interpreting, attributing, and dating inscriptions faster and more reliable, transforming how ancient history is studied. This breakthrough demonstrates how AI can model historical uncertainty and unlock new understandings of the ancient world.