A groundbreaking study has used artificial intelligence to uncover hidden writing patterns in the Bible, offering new insights into who may have written some of its most debated passages.
By combining modern tools like AI, math, and linguistics, researchers have created a fresh approach to one of the most enduring mysteries in biblical scholarship: authorship.
The international research team, led by Shira Faigenbaum-Golovin, a mathematics professor at Duke University, included experts in archaeology, theology, statistics, and computer science.
Their goal was to trace the likely authors of various parts of the Hebrew Bible, focusing especially on the first nine books, known collectively as the Enneateuch.
Instead of relying on traditional machine learning, which needs large amounts of data, the researchers developed a custom method that could work with the Bible’s often very short and heavily edited passages.
They compared how frequently certain words or word roots appeared in different chapters and analyzed the patterns of sentences to distinguish between different writing styles.
The model was first tested on 50 chapters whose authorship is generally agreed upon by scholars.
It successfully identified three distinct writing traditions or scribal styles: the book of Deuteronomy, the historical books from Joshua to Kings (known as the Deuteronomistic History), and the priestly sections of the Torah.
The model then moved on to chapters whose origins are more contested and offered evidence-backed suggestions for their likely authorship.
One surprising discovery came from examining the Ark Narrative found in the Books of Samuel.
Although the two sections—1 Samuel and 2 Samuel—seem to be part of the same story, the AI analysis showed that they were likely written by different groups.
The section in 2 Samuel showed clear similarities to the Deuteronomistic History, while the portion in 1 Samuel didn’t fit with any of the three known styles, raising new questions about its origins.
What sets this model apart is its transparency. Not only does it assign chapters to likely writing traditions, it also explains which words and phrases influenced its decision—providing a level of clarity that’s often missing in AI-driven research.
Faigenbaum-Golovin has been on this path for over a decade. Her collaboration with archaeologist Israel Finkelstein began in 2010 when they analyzed ancient Hebrew inscriptions found on pottery, which helped date parts of the Old Testament. That early success laid the foundation for the team that produced this latest Bible study, which also includes researchers from Israel, France, and Switzerland.
The results confirmed some scholarly beliefs, such as the close connection between Deuteronomy and the historical books. But the model also revealed surprising distinctions—even in the use of common words like “no,” “which,” or “king”—that helped identify different author groups.
Now, the team plans to apply the same approach to other ancient texts, including the Dead Sea Scrolls. For Faigenbaum-Golovin, the work highlights the unexpected harmony between science and the humanities.
“It’s a surprising symbiosis,” she said. “And we’re just getting started.”
