Artificial intelligence is more prominent than ever, providing information to learners across the globe. However, when AI systems fail to provide in-depth knowledge of ancient cultures, it hinders the ability to preserve and understand history.
Megan Leight, art history professor, is leading the Artful Algorithms project that could potentially change how AI learns and preserves ancient languages essential to understanding our world.
Supported by a collaboration grant from the WVU Humanities Center, the project explores the impact of AI on ancient Maya glyph decipherment.
Leight said the idea for the project came from a disheartening lack of Maya script recognition from the most advanced AI systems, limiting research and education.
The project is a cross-University collaboration between the Art History program at the College of Creative Arts and the Computer Science program at the Benjamin M. Statler College of Engineering.
Prashnna Gyawali, assistant professor of computer science, serves as the co-principal investigator.
Students from both programs worked on the initial dataset this spring and will be hired this summer to help create a database capable of recognizing and segmenting Maya glyphs. The end goal is to design an AI framework that can translate the glyph blocks in ancient Maya art.
“The explosion of data in fields such as natural image processing has demonstrated the remarkable capabilities of these models in tasks like object classification, segmentation and recognition. However, a similar depth of annotation for Maya written images, crucial for training effective AI models, is notably absent,” Gyawali explained in the project overview.
Deciphering ancient text is crucial to understanding how civilizations lived, and the Artful Algorithms project aims to advance understanding of ancient Maya culture by making glyph decipherment more accessible.
“By leveraging AI innovations to critically examine the potential and challenges of using AI for the decipherment of Maya script, we aim to enhance understanding of the underpinnings of ancient Maya civilization,” Leight said.
Once complete, the aim is to make the dataset publicly available to help others in their work teaching AI systems. Leight and Gyawali will present the project at an academic conference and anticipate publication in a peer-reviewed journal.
“We believe this type of AI programming is the future for language decipherment — building generative AI models to learn and assess ancient texts and images,” Leight said. “The Vesuvius Challenge popularized using AI to view text on a burned scroll from Herculaneum in February 2024. Maya glyphs are the next frontier for AI and language decipherment, and this project is our first foray into this challenging interdisciplinary field relying on humanities scholars working alongside data scientists and researchers.”