Some diamond industry organizations admit it while others deny it, but there can be little doubt that AI has already started to create waves throughout the diamond industry. AI is already being used in several areas of diamond grading, and that trend is clearly going to continue growing.
The announcement by the GIA in July that it was axing hundreds of jobs can be understood against the background of developments in AI for grading. In retrospect, it should not come as a surprise since it has already been three years since the global diamond industry’s leading diamond grading lab and research organization joined hands with computing giant IBM to develop AI-based diamond grading technology.
The GIA has a huge advantage in adopting AI-based grading systems since its database contains the diamond characteristics of many millions of stones that it has graded over the decades at its labs worldwide. The GIA is estimated to grade some 4 million diamonds annually, which gives an indication of the depth of information that it possesses relating to a huge range of diamonds.
Why Using AI To Grade Diamonds Make Sense
As with any other job, diamond graders make judgments all the time. But, as with every other employee in any sector, they also have different moods and issues affecting their daily performance. Consequently, it's easy to understand why there are sometimes errors in grading reports.
There are differences of opinion, borderline judgements, times when people are short-tempered or under pressure due to personal issues and make mistakes.
As a result, there are uncountable numbers of cases where different labs have given the same diamond a different grade. Not only that, but very frequently the same lab has given a stone a different grade when it has been resubmitted. Different graders, different grades.
Another issue that prevents human graders from providing consistently top-quality results, and which could fairly easily be rectified, says one veteran diamantaire, is the lighting in lab premises. “Traditional diamond graders have a very difficult job. Labs are a sort of twilight zone since they have very dim lighting aimed at reducing reflections as lab staff spend a lot of time looking through microscopes which strain their eyes.”
Given the rise in online sales of both loose diamonds and diamond jewelry over the last 10-15 years and their expected further expansion, the trade has come to rely on diamond grading reports being accurate and consistent. Consistency is vital and that is why labs have developed machinery offering the repeatability that AI offers.
David Block, CEO of Israeli diamond planning and cutting technology company Sarine, believes that in time, AI will eventually be able to grade all the 4Cs. Furthermore, automated grading of all the parameters involved will be far superior to manual grading. AI means repeatable grading without any of the emotions that human graders bring to the job – the same stone will be graded the same way time after time.
This won’t happen overnight since machines need to learn about accuracy based on what has gone before. Although technology as of now knows how to grade color, clarity, fluorescence, symmetry and cut grade consistently, future challenges relate to polish quality, milkiness in stones and other issues. These include color tinges, for which AI currently does not have enough examples.
As a result, human judgment will still be needed despite the expected widespread use of AI, but there can be little doubt that the number of people involved in the trade will decline sharply.
Recruiting Outsiders to Bring a Fresh Perspective
What could make more sense than to use the skills and experience of veteran diamond professionals in developing AI tools for the trade? Ironically, the answer could well be actually to recruit people from outside the industry.
The simple reason is that outsiders can bring a completely different perspective. They are more likely to be able to see the wood for the trees, whereas industry professionals may only be seeing the trees and not the bigger picture.
New technologies, such as artificial intelligence (AI), don’t exist in a vacuum. To utilize them effectively and integrate them into the diamond trade, there must be a human element. But are companies doing enough to tempt people with the right skills into the diamond industry?
Because fresh perspectives are critical in creating innovative disruptive technologies, professionals are being recruited from disruptive sectors such as fintech, e-commerce and AI research. Their non-acquaintance with the industry enables diamond firms to question industry standards. This is one of the steps that labs are taking to enable them to develop advanced technologies.
Other Ways AI is Being Used
Chroma Diamonds, a unit of manufacturer BRD Diamonds based in Antwerp, uses proprietary, patent-pending AI technology to assess fancy-colored diamonds. It’s also created a machine learning system that can predict the color grades for all potential polished stones resulting from a particular parcel of rough.
Meanwhile, lab-grown diamond maker Adamas One bought a stake in AI company NexGenAI Solutions earlier this year. It aims to use the company’s AI technology to optimize the development of its lab-grown diamonds. This will include its manufacturing output volume, coming up with new jewelry designs, and making its marketing more efficient.
“Artificial intelligence is cutting-edge technology that is rightly gaining significant attention and traction both in the current business macroenvironment and in the culture as a whole,” said Adamas One CEO Jay Grdina. “This investment and partnership allow us to strategically fold AI into our business, with the goal to maximize output, efficiency, creativity and value.”
Looking to the Future
What can we expect the diamond industry to look like 10 or 20 years from now?
Firstly, we can predict that the number of humans involved in diamond sorting and grading will decline substantially. The diamond trade has been engaged in searching for ways to reduce costs since the early 2000s, and even before, due to the high levels of competition for polished diamond sales and often wafer-thin profit levels.
This is one of the main reasons, of course, that 90% of the world's diamonds, by volume, are cut and polished in India. Israel, Belgium and the United States saw their numbers of manufacturing workers fall sharply from the 1980s onwards as production moved to the much less expensive diamond centers in Asia. In addition to India, manufacturing also shifted to Vietnam, Laos, Thailand and China.
As the last major generation of diamond sorters and graders retires, it would seem reasonable to assume that Western diamond trading companies will not replace them, particularly given the high cost of training and retaining such staff.
A diamond company’s operations in the future, with almost no human intervention, could very well look like this:
• Diamonds for purchase are placed in an instrument which rapidly checks individual stones which fit the company’s manufacturing criteria – the 4Cs. Meanwhile, diamonds that are not a fit are filtered out to be returned to the seller.
• The diamonds that are a good fit are moved on to the planning stage for cutting and individually given a number for the cutting and polishing required.
• The diamonds are moved on to automatic cutting and polishing.
• After cutting and polishing, the diamonds go through a computerized quality assurance phase.
• The diamonds are then photographed and videoed 360 degrees to create a perfect all the way round image for potential clients.
• At the end of the process, the company either sends them for grading to a lab or uses an e-grading operation which produces an extremely detailed report about myriad elements of the diamond, rather than just the simple 4Cs+ reports that are created today.
• The reports are automatically uploaded to the company’s website and spreadsheets for sending to customers which it knows is looking for each specific type of stone.
• AI programs will also be used to identify new potential clients who will receive the spreadsheets via a range of media, including Instagram, Facebook, Slack, regular email and other avenues.
• The diamonds will come with recommendations for which type of jewelry they are most suited.
• The diamonds will move along to the packaging process with every diamond coming with a report on its weight and other characteristics. Shipping to customers around the world will be automatically triggered.
Final Words
These are just some of the possibilities that AI could bring about. In some ways, they can be regarded as rather mild and that changes that we cannot even imagine will be seen in the not-too-distant future. After all, apart from a relatively small number of people in the tech sector, who was ready for ChatGPT to make such a huge splash?
But this also raises an interesting point: although ChatGPT has received huge publicity and extremely large numbers of people around the world have and continue to sign up to it, its drawbacks have swiftly become apparent. It produces text that is often described as bland and shows its machine origins, rather than the human touch that is needed to persuade people to carry out desired actions.
In other words, the changes that some industry players are certain will come about due to AI may not happen or might simply not be acceptable to the diamond trade.
Abraham Dayan for Rough&Polished
