LONDON: Google’s DeepMind has introduced a groundbreaking artificial intelligence tool, Aeneas, that is transforming how historians decode and restore ancient Roman inscriptions. Designed to assist in identifying damaged or incomplete texts, the tool is already gaining traction among researchers for its high accuracy and historical insight.
Aeneas was developed in collaboration with academics including Dr Thea Sommerschield from the University of Nottingham. It draws on an extensive database of nearly 200,000 Latin inscriptions—containing over 16 million characters—to suggest missing words and determine the likely origin and time period of each piece.
The ancient inscriptions, carved into tombs, monuments, and even everyday artefacts, offer glimpses into daily life, governance, and personal stories from the Roman Empire. However, many are incomplete due to weathering and time. Aeneas analyses these texts using not just linguistic patterns but also historical context from the 7th century BC to the 8th century AD, providing deeper connections than traditional keyword searches.
In practice, the AI has helped reconstruct famous inscriptions like the Res Gestae Divi Augusti and made linkages between texts across Roman provinces. One of its key features is the ability to predict the province of origin and date of an inscription—accurate within roughly 13 years. When tested on inscriptions with known originals, Aeneas demonstrated high reliability.
Historians describe the tool as “transformative,” with a team of 23 researchers finding it useful in 90% of cases. Experts note that Aeneas bridges technology and humanities, significantly enhancing the efficiency and scope of epigraphic research.
With its integration of deep learning and linguistic scholarship, Aeneas represents a leap forward in digital archaeology—offering scholars new ways to explore the Roman world.
