The Role of AI in Deciphering Ancient Texts
Historians have long struggled with the challenge of determining the exact dates of ancient texts, especially those that lack clear chronological markers. One such example is the “Res Gestae Divi Augusti,” a historical account of Emperor Augustus’s achievements. This document, like many others from the ancient Roman period, does not include any explicit dates, making it difficult for scholars to pinpoint its creation.
Enter Aeneas, a generative AI model developed by Google DeepMind. Unlike the mythological figure from Roman legend, this Aeneas is a powerful tool designed to analyze and interpret ancient texts. According to reports from The New York Times, Aeneas was able to estimate the date of the “Res Gestae Divi Augusti” to around 15 CE, shortly after Augustus’s death in 14 CE. This breakthrough highlights the growing potential of artificial intelligence in the field of historical research.
How Aeneas Works
Aeneas functions by mimicking the analytical processes that historians typically use. It retrieves relevant contextual information, identifies isolated texts, and analyzes them to draw conclusions. This approach allows the AI to process vast amounts of data quickly, something that would take human researchers years to accomplish.
In a recent paper published in the science journal Nature, researchers from Google DeepMind and several European universities tested Aeneas’s capabilities. They found that the AI provided valuable insights for historians in most cases, reinforcing its usefulness in the study of ancient texts.
For historians, Aeneas is part of an expanding array of AI tools that are transforming the way they uncover details about the ancient world. These technologies are helping to turn epigraphy — the study of ancient inscriptions — into a more accessible and dynamic field of historical inquiry.
Challenges in Studying Ancient Texts
Analyzing ancient texts is a complex and time-consuming task. Historians often have to search through archives across the globe to find “parallels” — similar texts from the same era or with comparable language. By comparing these documents, researchers can infer what missing parts might say or understand the context in which the text was written.
Thea Sommerschield, a historian at the University of Warwick and coauthor of the study, described the process as “like solving a gigantic jigsaw puzzle.” However, this method requires extensive training and can be extremely tedious.
Results from the Aeneas Study
The evaluation of Aeneas revealed impressive results. According to the paper, the AI provided “useful research starting points in 90 percent of cases, improving their confidence in key tasks by 44 percent.” When combined with human expertise, the outcomes were even better than when either the AI or the historians worked alone.
One concern with AI models is the possibility of “hallucination” — generating false or inaccurate results. However, Aeneas addresses this issue by providing probabilities for its predictions. As noted by Alison Cooley, a professor at Warwick University, Aeneas “hedged its bets,” offering two probable date ranges rather than a single prediction. This approach reflects the current scholarly debate on the dating of the text.
The Future of AI in Historical Research
This development underscores the potential of AI as a valuable tool for subject matter experts. Unlike non-experts, who may encounter unexpected or unsettling results, trained historians can effectively utilize AI to enhance their research.
If used correctly, Aeneas and similar tools could revolutionize the study of ancient history. Researchers may soon gain deeper insights into the past, unlocking new knowledge that has remained hidden for centuries.
As AI continues to evolve, its role in historical research will likely expand, offering exciting possibilities for scholars and enthusiasts alike. The integration of technology with traditional methods promises to make the study of ancient texts more efficient, accurate, and accessible.