January 3, 2025
AI is uncovering ancient texts, unlocking secrets from the past
AI-driven technologies are revolutionising the study of ancient manuscripts, uncovering texts that have remained hidden for centuries. Breakthroughs, such as decoding Greek text from Herculaneum scrolls and translating Joseon dynasty records, highlight the power of AI in historical research. AI tools like neural networks and transformer models are also being used to r...

The use of artificial intelligence (AI) is transforming the study of ancient manuscripts, unlocking texts that have remained unreadable for centuries. From deciphering burnt Roman scrolls to analysing damaged cuneiform tablets, AI-driven tools are revealing new information about the ancient world. This advancement allows researchers to examine vast archives, identify previously unknown writings, and reconstruct missing text with unprecedented accuracy. Scholars are now equipped with more data than ever before, opening up opportunities to answer historical questions that were once out of reach.

Breakthrough in Herculaneum Scroll Deciphering

As reported by nature.com, AI has successfully revealed significant portions of Greek text. Computer scientists Brent Seales from the University of Kentucky and participants of the competition called Vesuvius Challenge, utilised advanced neural networks to detect ink patterns on the fragile scrolls. As reported by Nature.com, the February 2024 winning entry employed a TimeSformer model, revealing text attributed to an ancient Greek philosophical work. Federica Nicolardi, a papyrologist at the University of Naples, described the discovery as a “historic moment for the field.”

Applications in Other Historical Archives

AI tools are also being applied to diverse historical collections. Researchers in South Korea are employing transformer-based networks to translate the extensive records of the Joseon dynasty, written in Hanja. This system has expedited translations of state records, offering insights into the political and cultural trends of the era. Similarly, initiatives such as the Fragmentarium project at Ludwig-Maximilians University are using neural networks to identify overlapping cuneiform fragments, uncovering lines from the Epic of Gilgamesh and a previously unknown hymn to Babylon.

Challenges and Future Prospects

Concerns regarding accuracy and accessibility remain as AI enables rapid decoding of large text corpora. Experts emphasise the need for interdisciplinary collaboration and open-source data to ensure transparency and replicability. Brent Seales has noted to nature.com that the success of AI in these applications underscores its role as a complement to human expertise, providing data that specialists can interpret and analyse. As AI continues to evolve, researchers anticipate breakthroughs in decoding lost languages and exploring underground libraries, potentially reshaping the understanding of ancient civilisations.