UBISOFT / GOOGLE
Applying Machine Learning to Ancient Egyptian Hieroglyphs
The Challenge
Could machine learning solve a problem that had challenged researchers for centuries?
Ancient Egyptian hieroglyphs represent one of history's most complex written languages, requiring years of specialist study to interpret accurately. While advances in machine learning suggested new possibilities, no practical research platform existed to test whether the technology could genuinely support translation.
Ubisoft approached the challenge with an ambitious vision: combine machine learning, academic expertise and open collaboration to explore whether computers could assist Egyptologists without replacing their judgement.
To succeed, the project needed more than an algorithm. It required a complete research ecosystem capable of bringing together datasets, researchers, machine learning models and public engagement within a single platform.
Our Approach
Combining machine learning, academic collaboration and software engineering to make the impossible testable.
Rather than beginning with technology alone, Psycle worked alongside Ubisoft, Google and leading Egyptologists to design a collaborative research programme that would allow machine learning to be tested against genuine academic workflows.
By combining software engineering, machine learning infrastructure and user-centred design, we created a platform where researchers could contribute knowledge, validate results and continually improve the quality of the underlying models through real-world collaboration.
This approach transformed an academic experiment into an open research initiative, combining technology, community participation and scholarly expertise to explore one of history's oldest unsolved linguistic challenges.
The Solution
An open machine learning research platform built to support experts, collaboration and discovery.
Psycle designed and developed the Hieroglyphics Initiative as an open machine learning research platform, bringing together browser-based research tools, machine learning services and collaborative workflows into a single digital environment.
The platform enabled Egyptologists to analyse hieroglyphic inscriptions, review machine-generated suggestions and contribute expert knowledge that continually strengthened the quality of the research dataset.
Alongside the research platform, Psycle designed and delivered the complete public-facing ecosystem for the initiative, including the campaign identity, website, launch video, academic presentations, community platform and supporting digital assets, helping communicate complex machine learning research to both specialist and public audiences.
Technology included
Working closely with Ubisoft, Google and leading Egyptologists, Psycle transformed an ambitious research concept into a complete machine learning ecosystem that combined software engineering, academic collaboration, public engagement and digital experience, enabling the research to move beyond a single project and become a shared resource for the wider academic community.
The Outcome
A pioneering machine learning project demonstrating how technology can accelerate academic research.
The Hieroglyphics Initiative demonstrated that machine learning could become a valuable research assistant, helping experts analyse complex inscriptions while leaving scholarly interpretation firmly in human hands.
The project brought together Ubisoft, Google and an international community of Egyptologists through a shared research platform, demonstrating how collaborative technology can accelerate academic discovery.
Beyond the technology itself, the initiative became an internationally recognised example of machine learning applied to digital humanities, with presentations at academic conferences, its launch at the British Museum and inclusion within Google's own technology showcases.
KEY RESULTS
- 80,000+ training samples generated for machine learning
- 800+ hieroglyphs analysed and classified
- Open machine learning research platform released
- Featured at Google Next and international research events
Case Studies
