Gübelin Gem Lab is celebrating its 100th anniversary with the introduction of Gemtelligence, an algorithm based entirely on artificial intelligence (AI) and deep learning to determine the country of origin of rubies, sapphires and emeralds.
In cooperation with the Swiss technology innovation centre CSEM, Gübelin developed a platform determining the origin and treatment of coloured gems using machine learning. Gemtelligence recognises heat treatments in rubies and sapphires. In this process, the deep learning architecture relies on the latest scientific concept: a comprehensive AI system for evaluating a wide variety of analytical data. This system uses and validates the analyses which the Gübelin Gem Lab has collected over decades.
Gübelin’s reference stone collection comprises more than 28,000 gemstones from all commercially relevant mines, and together with analyses of tens of thousands of high-grade customers’ stones which the lab has analysed since the 1970s, forms the backbone of Gemtelligence.
The algorithm, based on the latest deep learning architecture, recognizes significant patterns among various data modalities, which, compared to gemmologists, provides heightened precision and robustness. For most of the gemstones examined, this means that both the work-intensive visual (microscopic) examination by human experts as well as the expensive ICPMS laser ablation analytical step can be avoided.
Gemtelligence is being successfully implemented step by step in the three laboratories in Lucerne, Hong Kong and New York. Currently, each stone there is still tested by human experts.
Raphael Gübelin, President of the House of Gübelin, explained: “For the Gem Lab’s 100th anniversary, we are presenting a further innovation with Gemtelligence. With this milestone, we are setting out in a new direction in gemmology and perpetuating our pioneering spirit. My great-uncle Eduard Josef established the discipline of determining the origin of gemstones and developed it into a service. With Gemtelligence we are transferring this task to a new authority to ensure the consistency of results. Gemtelligence allows for higher precision, which leads to deeper trust.”
Daniel Nyfeler, Managing Director of the Gübelin Gem Lab and Project Manager for Gemtelligence, added: “The fascinating aspect of this project is not just the improved precision and consistency of the data evaluation but also the ability to simply dispense with time-consuming and expensive analyses. This is something the entire industry will profit from.”