Can Artificial Intelligence Unlock Secrets of the Ancients?
AI is helping historians rewrite history, quite literally. Published in the journal Nature, a new study reports that AI can fill gaps in ancient Greek inscriptions and indicate where and when they are from. The paper notes that “Over the centuries, many inscriptions have been damaged to the point of illegibility, transported far from their original location and their date of writing is steeped in uncertainty.”
According to TRTWorld, the researchers say the new AI system they have developed can not only accurately read ancient Greek inscriptions, but also fill in gaps in the text that have occurred due to damage, and even chronologically and geographically locate them.
Dr Thea Sommerschield is co-author of the study by Ca’ Foscari University of Venice, Harvard University and AI firm DeepMind. Explaining the importance inscriptions hold for historians, she stated that they were evidence of the thought, language, society and history of ancient civilizations as they were written directly by the people of those times, according to the Guardian.
“But most surviving inscriptions have been damaged over the centuries. So their texts are now fragmentary or illegible.” It was also not unusual for inscriptions to be moved from their original location by later rulers and dynasties.
Much of what survives from the ancient world is fragmented and out of context (Dennis Jarvis / CC BY-SA 2.0)
With conventional methods such as radiocarbon dating not working for stone, historians are often unable to glean complete information from such damaged inscriptions. Alternatives are needed to fully understand the texts.
From the Ithaca of Antiquity to the Futuristic Ithaca of AI
Nicknamed “Ithaca” after the Greek island home of Homer’s King Odysseus, the modern Ithaca is a deep neural network architecture AI. It seeks to assist epigraphists fill in crucial gaps in inscriptions from Homer’s time, simultaneously restoring texts as well as pinpointing the time and place they came from.
- Artificial Intelligence Inching Closer to Deciphering Long Lost Languages
- Google AI Decodes Broken Greek Texts Better than Humans
And, as New Atlas reports, the prospects look promising. So far Ithaca has shown a 62 percent success rate in restoring damaged texts, and a 71 percent success rate in identifying their original location along with the ability to accurately date the texts to within 30 years of their creation.
The use of Ithaca alongside research by historians improved on the success rate of Ithaca alone in restoring damaged texts from 62 percent to 72 percent. Historians working unassisted by Ithaca had a success rate of 25 percent, so its astonishing possibilities have quickly become apparent.
“Just as microscopes and telescopes have extended the range of what scientists can do today, Ithaca aims to singularly augment and expand the capabilities to study one of the most significant periods of human history,” Dr Yannis Assael, a co-author of the work, is quoted in the Guardian as saying.
The team believe that the model could be tuned to almost any ancient language, from Latin to Mayan to Cuneiform. Further, it could possibly be capable of reading Greek literary texts written on papyrus.
This has potential for better access to the poetry of Sappho, for example, most of which survives only in fragmentary form. It could even be developed to decide the authorship of texts through linguistic analysis.
Ithaca could be used to unlock the mysteries of Sappho (Georgios Kollidas / Adobe Stock)
Ithaca, the researchers said, has already unlocked some of the secrets of the ancients. When applied to a set of decrees recovered from the Acropolis in Athens, it unexpectedly concluded that one of the documents was different to the rest.
These texts, relating to collection of tributes across the Athenian empire, were long believed to date to 448-7 BC. However, Ithaca concluded that one was actually 30 years younger and was written in 424 BC. This aligned in with other recent evidence.
- AI Bot Will Sniff Out Historic Smells to Recreate Ancient Smellscapes
- Reconstructing Jesus: Using Science to Flesh out the Face of Religion
“Although it might seem like a small difference, this 30-year shift has momentous repercussions for our understanding of the political history of classical Athens, and helps us better align literary sources – such as Thucydides’ account of these years and events – with the epigraphic record,” said Sommerschield, according to the Guardian.
An Imperfect Mind, for the Moment
While scholars are excited at the possibilities the AI opens up, they warn about the need to proceed with extreme caution in using it to interpret the past. Prof. Peter Liddel, an expert in Greek history and epigraphy at the University of Manchester acknowledged that the AI would certainly add to historians’ toolbox and help understand processes like the development of imperialism or the nature of cult activity.
However, like scholars, the AI too has only the available ancient record to go by. “AI is only powerful as a tool to help us ask questions about, and make comparisons to, the existing evidence,” he said, according to the Guardian.
Prof. Melissa Terras, an expert in digital cultural heritage at the University of Edinburgh, is also in equal parts enthusiastic and circumspect about the use of the AI. She said it was important to continue training historians in the traditional methods to interpret the results generated by the software.
Fragmentary tablets from Phaistos in the undeciphered Linear A text (Zde / CC BY-SA 4.0)
But she felt the AI had immense possibilities, given the structured formats of ancient texts that survived only in fragments. “This means they require a lot of cross-referencing for the human brain to solve the puzzle – but this is the type of repetitive calculation that [AI systems such as] deep neural nets excel at,” she stated.
So, while welcoming the new technology, experts warn that results generated by it would be incomplete and less reliable if conclusions are drawn independently of human knowledge. However, it is a good tool to generate data for the human brain to interpret, and who knows what secrets of the ancients it may be able to unlock.
Top image: Ithaca looks likely to offer us a new understanding of the ancient world. Source: Kras99 / Adobe Stock.
By Sahir Pandey
Assael, Y. et al. 2022. Restoring and attributing ancient texts using deep neural networks. Available at: https://www.nature.com/articles/s41586-022-04448-z
Davis, N. 2022. AI could decipher gaps in ancient Greek texts, say researchers. Available at. https://www.theguardian.com/science/2022/mar/09/ai-could-decipher-gaps-in-ancient-greek-texts-say-researchers
Lavars, N. 2022. Google's DeepMind AI fills in the blanks on broken ancient texts. Available at: https://newatlas.com/computers/googles-deepmind-ai-gaps-ancient-texts/
TRT World. 2022. AI can help historians unravel ancient texts. Available at: https://www.trtworld.com/magazine/ai-can-help-historians-unravel-ancient-texts-55451