A University of Warwick epigraphy expert has collaborated with Google DeepMind to evaluate 'Aeneas', an AI model that reimagines Roman inscriptions.
Co-authoring a paper published in the world's leading multidisciplinary science journal Nature today, Alison Cooley, Professor of Classics and Ancient History at the University of Warwick, has played a key verification role in developing the first artificial intelligence (AI) model for contextualising ancient inscriptions.
Aeneas, named after the wandering hero of Graeco-Roman mythology, is designed to help historians better interpret, attribute and restore fragmentary texts.
Writing was everywhere in the Roman world, etched onto everything from imperial monuments to everyday objects. From political graffiti, love poems and epitaphs to business transactions, birthday invitations and magical spells, inscriptions offer modern historians rich insights into the diversity of everyday life across the Roman world.
Often, these ancient texts are fragmentary, weathered or deliberately defaced, and restoring, dating, and placing them is nearly impossible without contextual information. Historians have traditionally relied on their expertise to identify 'parallels' - texts that share similarities in wording, syntax, or provenance.
Aeneas greatly accelerates this complex and time-consuming work. It reasons across thousands of Latin inscriptions, retrieving textual and contextual parallels in seconds, allowing historians to interpret and build upon the model's findings.
As a global expert in epigraphy - the study and interpretation of ancient inscriptions - Professor Cooley investigated the 'real world' integrity of Aeneas' analysis.
Professor Cooley explains: "I was delighted to test out Aeneas' capabilities via a detailed examination of the renowned Roman inscription, the Res Gestae Divi Augusti. A self-penned, self-congratulatory lifetime achievement summary by the Emperor Augustus, the Res Gestae is a much-studied text, so using it as a trial for Aeneas was a high bar. Not only is the inscription stuffed with imperial exaggerations, irrelevant dates, and false geographical flags, there's also significant academic disagreement over its date.
"I found that Aeneas was able to contextualise all the inscription's ambiguous dating and provenance features accurately. It picked up on clues of spelling and vocabulary, as well as the linguistic nuances indicating subtle political ideology and imperial attribution. Aeneas identified other texts showing these same characteristics of language - even texts found thousands of miles apart - and used this pattern of findings to estimate chronology.
"Interestingly, Aeneas hedged its bets. In doing so, it exactly reflected the current difference in scholars' opinions, giving two probable date ranges rather than a single prediction."
Professor Cooley concludes that this groundbreaking research demonstrates the advantages of collaboration between ancient historians and modern technologists, saying, "State-of-the-art generative models are now helping to turn epigraphy from a specialist discipline into a cutting-edge field of historical enquiry."
Aeneas, co-developed by Google DeepMind and the University of Nottingham, is part of a wider effort to explore how generative AI can help historians better identify and interpret parallels at scale. The model can also be adapted to other ancient languages, scripts and media, from papyri to coinage, expanding its capabilities to help draw connections across a wider range of historical evidence.
