Artificial intelligence is revolutionising the study of ancient history by deciphering damaged texts and inscriptions from civilizations like Ancient Rome and Mesopotamia. AI tools can now fill in missing words, date inscriptions, and even read texts from carbonised scrolls, offering unprecedented insights into the past.
Unlocking Ancient Secrets with AI
AI is proving to be a powerful ally for historians and archaeologists, capable of analysing vast amounts of fragmented data that would overwhelm human researchers. These advanced algorithms can identify patterns in damaged texts, predict missing words, and even determine the origin and age of inscriptions, thereby filling crucial gaps in our understanding of ancient societies.
Key Takeaways
AI can decipher damaged ancient Greek and Roman inscriptions, often outperforming human experts in accuracy and speed.
Machine learning models are being used to read texts from carbonised papyrus scrolls, such as those found in Herculaneum.
AI tools can help date and geographically locate inscriptions, providing a more comprehensive historical context.
These technologies are seen as collaborative tools, augmenting rather than replacing the expertise of human scholars.
Deciphering Damaged Inscriptions
Thousands of ancient inscriptions, found on stone, ceramic, and metal, are damaged or eroded, making them difficult to read. AI systems, like one developed by DeepMind called Pythia, have been trained on massive datasets of inscriptions. Pythia can suggest missing words or characters, with human experts then selecting the most plausible options. This approach has shown a significant reduction in errors compared to human analysis alone.
Reading Carbonised Scrolls
The catastrophic eruption of Mount Vesuvius in 79 CE carbonised hundreds of papyrus scrolls in Herculaneum. These scrolls, resembling lumps of coal, have been notoriously difficult to unroll and read. However, advancements in AI, combined with techniques like X-ray computed tomography (CT) scans, have enabled the retrieval of text from these fragile documents. A recent breakthrough saw a student win a prize for reading the first text from a carbonised scroll, opening the door to potentially hundreds more.
AI as a Historical Assistant
Tools like Aeneas, developed by researchers including those from Nottingham University and Google DeepMind, are designed to assist historians. Aeneas can analyse Roman inscriptions, identify relevant historical parallels, and help date them.
While AI can accelerate the process of piecing together historical narratives, human scholars remain essential for interpreting the AI's output, ensuring accuracy, and providing nuanced understanding. The collaboration between AI and human expertise promises to transform the field of classical studies.
