AI Model Apollo Aids Deciphering of Damaged Ancient Greek Texts

The Austrian Academy of Sciences is releasing Apollo, an artificial intelligence model developed with AI lab Mistral and technology firm Sail Reply. The model is designed to assist researchers in restoring and interpreting ancient Greek papyri, many of which are severely damaged.

Apollo has been trained on approximately 600 million historical Greek words. It can fill in missing words and phrases with the most statistically probable options, significantly speeding up a previously laborious process that required deep expertise. The model will be accessible to researchers via a chatbot interface.

According to researchers, Apollo can help identify relevant papyrus fragments more quickly and uncover new research avenues. While the model is unlikely to broadly alter understanding of the ancient world, it may reveal new details about ancient life and substantiate existing scholarly assumptions.

Developers believe the same technique could be applied to other ancient languages, such as Latin or Egyptian, or other disciplines benefiting from large-scale data distillation. The risk of historical record contamination is addressed by Apollo offering alternative completions for scholars to select from.