Google DeepMind has developed an AI tool called Aeneas to assist historians in interpreting ancient Roman inscriptions.

The AI can predict the origin and date of inscriptions and suggest possible words to fill in gaps caused by damage.

Aeneas is trained on a database of nearly 200,000 inscriptions from the 7th century BC to the 8th century AD.

It connects texts through historical relationships rather than relying solely on keyword searches.

The tool can attribute inscriptions to one of 62 Roman provinces and estimate their creation within a margin of 13 years.

Historians testing Aeneas have noted its ability to provide important context, reporting an accuracy rate of 90% in aiding analysis.

For example, it identified linguistic connections between inscriptions on a votive altar in modern-day Mainz, Germany, and an older altar in the region.

Researchers consider Aeneas a significant development for studying inscriptions, which are vital records of ancient life but often damaged or fragmented.

The tool may broaden access to interpretation in this field. However, experts highlight the importance of critically evaluating the AI’s output.

Details of the project have been published in Nature, and the tool is currently accessible to researchers online.

Recent Google developments