Russia develops AI that recognizes ancient Egyptian hieroglyphs

Russian scientists have developed an AI technology that can translate texts written in ancient Egyptian hieroglyphs. The model was trained on the most complete collection of ancient Egyptian texts with translations. This was reported by RIA Novosti, citing the AIRI Artificial Intelligence Institute.

"Scientists from the AIRI Institute, ISP RAS, and ITMO University have developed a system that can accurately recognize and translate hieroglyphic texts. The primary data source for training the model was the Thesaurus Linguae Aegyptiae—the most complete collection of ancient Egyptian texts with translations. The model's accuracy was assessed in two stages. First, automatic metrics were applied, and then the texts were evaluated by professional Egyptologists from the Higher School of Economics," the statement reads.

This development can be used in augmented reality mobile apps—visitors visiting a museum or seeing an ancient Egyptian artifact can point a camera at it and understand its inscription. AI also helps researchers improve the process of deciphering and interpreting new materials, and it facilitates the learning process for Egyptology students just beginning to learn the language.

Ancient Egyptian hieroglyphs are more complex to recognize than the standard alphabet: AI must understand hundreds of unique symbols, some of which are visually similar.

"The project is the result of the work of an interdisciplinary team of machine learning and Egyptology specialists. We have already presented the first scientific paper on the development of the method at the SIGGRAPH 2025 international conference. Access to the system is currently open upon request to interested specialists," commented Ilya Makarov, PhD, leading researcher at the AIRI Institute and project leader.