AI Reveals Dead Sea Scrolls May Be a Century Older Than Previously Believed

AI Analysis Suggests Some Dead Sea Scrolls Are Up to 100 Years Older Than Previously Believed

A new study combining radiocarbon dating with artificial intelligence (AI) indicates that several of the Dead Sea Scrolls may date back as far as 2,300 years ago, pushing their origins up to a century earlier than formerly thought. This adjustment aligns the manuscripts more closely with the lifetimes of their presumed authors.

Discovered in 1947 by Bedouin shepherds in caves near the Dead Sea, these scrolls include thousands of fragments from hundreds of manuscripts, primarily written in Hebrew on parchment and papyrus. Found near Khirbat Qumran in the West Bank, they have shaped modern perspectives on ancient Judaism and early Christianity.

The Dead Sea Scrolls include over 200 manuscripts of what is now recognized as the biblical Old Testament, representing the oldest surviving copies of the Hebrew Bible. These documents offer direct insight into the texts and cultural context from that era.

Dating Challenges and New Techniques

Most scrolls lack explicit dates, so scholars traditionally relied on paleography—the study of ancient handwriting—to estimate their age, dating them roughly between the third century BCE and the second century CE. However, this new research suggests some scrolls should be dated back as far as the late fourth century BCE.

Earlier radiocarbon dating done in the 1990s had limitations, partly because contaminants like castor oil—used in the 1950s to enhance legibility—skewed results toward more recent dates. The updated analysis applied refined radiocarbon techniques to 30 manuscripts, revealing that most are older than previously believed.

How AI Contributes to Manuscript Dating

After establishing new carbon dates, researchers trained an AI system named Enoch, using high-resolution images of the dated scrolls. Enoch successfully predicted the age of manuscripts with an 85% accuracy rate when tested on known samples.

The AI then assessed 135 additional scroll fragments not previously carbon-dated. Experts reviewed these AI-generated estimates as either realistic or unrealistic based on paleographic knowledge, and 79% were deemed realistic. Notably, some texts were estimated to be 50 to 100 years older than earlier evaluations.

For instance, a fragment containing verses from the Book of Daniel was formerly dated to the second century BCE—after the author’s lifetime. The new method places it closer to the author’s actual era. Similarly, a manuscript with Ecclesiastes passages shifted from a paleographic estimate of 175–125 BCE to a range between 300 and 240 BCE based on AI results.

Implications and Future Prospects

AI offers a non-destructive alternative to radiocarbon dating, which requires physically cutting tiny samples from the scrolls. While AI cannot yet fully replace carbon dating, it provides valuable supplementary analysis, especially when physical testing is not feasible.

Scholars outside the study have welcomed this dual-method approach. The combination of AI and improved carbon dating enhances calibration and precision in dating ancient manuscripts. Experts highlight that while AI results are promising, ongoing validation with more samples is necessary to confirm reliability.

One computer science expert noted that this data-driven approach marks a significant advance in manuscript dating. The AI’s performance is expected to improve as it processes additional data, ultimately contributing a valuable tool for researchers in archaeology and ancient texts.

Further Reading

• PLOS One Journal – Source of the published study
• Latest AI Courses for Research Professionals