Rewriting History with Machine Learning
When artificial intelligence is fed billions of words from medieval manuscripts, religious treatises, and forgotten chronicles, it begins to act like a hyper‑intelligent magnifying glass. Historian and theologian Yusuf Çelik of Vrije Universiteit Amsterdam has turned this concept into practice, using large‑scale language models to scan centuries‑old texts for patterns, names, and anecdotes that have long been eclipsed by dominant narratives.
From Symbols to Surprising Stories
One of the most striking revelations emerged around the David Star, a symbol traditionally linked to Judaism. Çelik’s AI‑driven search uncovered its presence in Islamic tradition as the “Seal of Solomon.” The discovery sparked lively discussion among viewers, many of whom were astonished to learn that a piece of jewelry they wore daily carried a cross‑cultural legacy. The episode highlighted how symbols can bridge communities, even when modern politics paints them as divisive.
A Canine Tale That Defied Expectations
Perhaps the most unexpected narrative involved a devoted dog whose owner wished to inter the animal in a human cemetery. This anecdote, buried in obscure legal commentaries, challenged Çelik’s own assumptions about medieval attitudes toward animals. It reminded readers that personal affection and ritual practices were far more nuanced than the stark, utilitarian pictures often painted by mainstream historiography.
Women, Coins, and Inter‑Faith Relations
Beyond the eye‑catching anecdotes, the AI analysis also amplified quieter voices: outspoken women in early Islamic societies, subtle economic exchanges reflected in coinage, and periods when Jewish‑Islamic relations were collaborative rather than conflict‑ridden. While these findings were less sensational, they provided a richer, multilayered view of the past, confirming that history is a tapestry woven from countless threads.
Why These Stories Vanished
Çelik posits that the disappearance of many narratives is not a matter of deliberate erasure but of selective transmission. Texts that were copied more frequently gained authority, while others faded into the margins of libraries and private collections. AI, by aggregating the full corpus of surviving words, restores visibility to those marginal voices, allowing scholars and the public alike to hear a broader chorus.
Implications for the Future of Historical Research
The project demonstrates that AI can serve as a powerful ally for historians, offering a systematic way to sift through massive corpora that would be impossible to read manually. It does not replace traditional scholarship; rather, it points researchers toward promising leads, be they obscure legal rulings, forgotten poetry, or everyday anecdotes that illuminate the lived experience of ordinary people.
As more archives become digitized and AI models grow more sophisticated, the potential to rewrite sections of our collective memory expands dramatically. The hope is that a more inclusive, data‑driven approach will foster a deeper appreciation for the diversity of past societies and encourage contemporary audiences to question the narratives they have inherited.