AI Sheds New Light on the True Age of the Dead Sea Scrolls

AI Sheds New Light on Dating the Dead Sea Scrolls

Since their discovery nearly 80 years ago, the Dead Sea Scrolls have offered invaluable insights into Jewish life over 2,000 years ago. These ancient Hebrew and Aramaic texts, preserved on tiny fragments of parchment and animal skin, include some of the oldest Hebrew Bible manuscripts and legal codes from turbulent times.

Despite their value, most scrolls lack precise timestamps, such as references to specific rulers, making it difficult to securely date them. Traditionally, researchers have relied on handwriting analysis and radiocarbon dating to estimate their age.

Introducing AI for Scroll Dating

Now, a new approach using artificial intelligence (AI) offers a fresh perspective. A machine learning model named Enoch cross-references carbon-14 dating with geometric analysis of letter shapes. This method, detailed in a recent PLOS ONE study, matches human paleographers’ age estimates about 80% of the time.

Where Enoch differs—dating about 20% of fragments as older than previously believed—these findings could reshape how scholars view the spread of literacy and ideas in the ancient Near East.

Human Expertise Remains Essential

While AI adds a powerful tool to the dating process, experts emphasize it won’t replace human paleographers. The lead researcher behind Enoch compares AI to a microscope for a biologist—an aid, not a substitute. Human judgment is crucial in developing and interpreting AI models.

Building on Past Machine Learning Applications

Machine learning has been applied before to ancient texts. For example, in 2021, researchers used it to identify multiple scribes behind the Great Isaiah Scroll. Similarly, carbonized scrolls from Herculaneum have been digitally “unwrapped” to reveal 2,000-year-old Latin works.

Extending these techniques to the Dead Sea Scrolls, the AI model helps date fragments from Jerusalem and surrounding areas, spanning the 4th century B.C.E. to the 2nd century C.E.—a period marked by major political upheavals:

• 332 B.C.E.: Alexander the Great’s conquest introduces pagan rule.
• 160 B.C.E.: The Maccabee uprising challenges Greek control.
• 70 C.E.: Roman forces destroy the Jewish temple during the siege of Jerusalem.

Challenging Established Script Chronologies

Prior paleographic analyses linked specific scripts to certain periods. The “Hasmonean” script, linked to the Jewish Hasmonean dynasty, was believed to date roughly from 100 B.C.E. to the 1st century C.E. The “Herodian” script was thought to be standardized during King Herod’s reign around the time of Jesus’s birth.

Enoch’s findings complicate this timeline, suggesting some “Herodian” scripts predate King Herod by up to 50 years. However, it also confirms previous paleographic estimates for some scrolls, such as 4Q114, which contains chapters of Daniel. Enoch placed its origin between 230–160 B.C.E., consistent with the historical context of the Maccabee uprising.

Balancing AI and Traditional Paleography

Experts emphasize both AI and paleography have limits. Handwriting analysis alone cannot provide exact dates without supporting data. AI offers a complementary perspective but requires careful interpretation.

Human handwriting reflects individual variation, making it a complex subject for automated dating. Combining traditional paleographic expertise with AI enhances accuracy and understanding without replacing the nuanced judgment of scholars.

For professionals interested in AI applications in research, exploring machine learning techniques for text analysis offers practical benefits. Comprehensive AI training resources can help deepen skills in this area—visit Complete AI Training for courses on machine learning and its applications.