AI and the Invisible Library

In a recent article published in Nature, the acclaimed historian and author Jo Marchant explores what might well be one of the most extraordinary technological breakthroughs of our generation. Titled How AI is Unlocking Ancient Texts — and Could Rewrite History, the piece explores how artificial intelligence is revolutionising the way we uncover and interpret the past. This convergence of cutting-edge technology and ancient history represents not just a leap in our capabilities but a profound shift in how we connect with our collective heritage.
This inflection point, by definition, marks a dramatic change in our civilisation’s trajectory. In history, such moments have included the invention of the printing press, the rise of the internet, and the sequencing of the human genome. Today, many experts argue that AI represents a similar paradigm shift. But this isn’t just about automation or optimisation; as Marchant’s article shows, AI is enabling entirely new avenues of exploration, particularly in fields like archaeology, linguistics, and historical research.
One of the most compelling examples Marchant highlights is the ongoing effort to decode the Herculaneum scrolls, which were carbonised during the eruption of Mount Vesuvius in AD 79. Using advanced neural networks and innovative scanning techniques, researchers have begun to uncover Greek texts hidden within these fragile, charred remnants. These texts, previously considered unreadable, may include philosophical works by the Epicurean thinker Philodemus, offering fresh insights into ancient thought on music, pleasure, and sensation.
Projects like the Vesuvius Challenge are at the forefront of this revolution. By leveraging AI models like TimeSformer—designed to analyse spatial and temporal dimensions—and crowdsourcing innovation, researchers have achieved breakthroughs that were unimaginable just a decade ago. Nearly 16 columns of Greek text have been revealed so far, with ongoing efforts to virtually “unwrap” more of the scrolls. According to Brent Seales, a pioneer in virtual unwrapping, the potential scale of discovery could surpass anything seen in the past century.
Meanwhile, AI is also transforming the study of other ancient records, such as the royal archives of Korea’s Joseon Dynasty. Written in Hanja, an archaic script, these vast records detail centuries of governance and cultural life but have remained largely inaccessible due to their complexity and scale. By using multilingual transformer models, researchers have begun translating these texts into modern Korean, offering insights into a historical period rich with detail yet largely unexplored.
Beyond these examples, AI holds promise for tackling mysteries like the undeciphered Linear A script of the Minoan civilisation or revealing texts hidden in medieval book bindings and Egyptian mummy wrappings. These breakthroughs signify the dawning of what Seales calls an “invisible library,” where technology transforms once-lost knowledge into a shared resource for humanity.
A Personal Connection to the Past
For me, the idea of unlocking the invisible library resonates deeply. Since 2007, I’ve been immersed in an ongoing personal research project about Iron Age Europe, specifically the unique and fascinating cultural heritage of Pre-Roman tribes of the Cantabrian Mountains, my ancestral homeland. These ancient peoples left behind a patchwork of archaeological artifacts and oral traditions, but much of their story remains incomplete. In an article I wrote last year, I reflected on how painstaking it is to reconstruct such histories—a process of piecing together fragments to form a coherent narrative.
AI’s ability to process vast amounts of data and fill in gaps with remarkable accuracy feels like a revelation. Imagine applying similar techniques to decipher inscriptions or reconstruct the rituals and daily lives of these tribal ancestors. The potential to bring clarity to their history is not just exciting; it’s transformative.
However, as exciting as these developments are, they also demand careful reflection. The ethical questions surrounding ownership, access, and interpretation of newly uncovered texts are significant. Who owns the knowledge that AI reveals? How do we ensure that cultural artefacts and their histories are treated with respect and sensitivity, particularly when they originate in marginalised or colonised regions? And as we lean on AI to decode the past, how do we avoid losing the human touch—the nuanced interpretation that only comes from lived cultural and historical understanding?
Navigating the Ethical Terrain
Marchant’s article also touches on these ethical complexities, reminding us that while AI can unlock extraordinary possibilities, it must be wielded responsibly. Consider the Herculaneum scrolls: as AI accelerates their decoding, questions arise about how the findings are shared, who has access, and how they fit into the broader narrative of classical studies. Similarly, as Korean royal archives are translated, how do we balance scholarly rigour with public accessibility?
In my own work on the Cantabrian tribes, I’ve grappled with similar questions. Researching the history of a region that has long been overshadowed by dominant narratives requires not just diligence but respect for the cultural context. As AI becomes a tool in such endeavours, we must ensure that its use amplifies, rather than erases, the voices of those whose stories it helps to tell.
Talking About Our Generation
AI’s impact on historical research is a defining moment of our time. By uncovering the invisible library of human knowledge, it bridges the past and the present in ways that challenge and inspire. Projects like these illustrate not just the power of AI but its profound potential to reshape our understanding of who we are and where we come from.
As a passionate learner of interdisciplinary intersections like the worlds of history and technology, I see this moment as more than a technological milestone. It’s an opportunity to reconnect with the past, to rediscover forgotten voices, and to ensure that the stories we uncover are preserved and shared ethically. This is our inflection point—and it’s up to us to make the most of it.