Diamond grading headed towards a shinier future

Ever since GIA was established in 1931, our focus has always been our consumer protection, nonprofit mission – to ensure public trust in gems and jewellery. All our activities – from our independent grading reports to world-class educational programmes, from developing sophisticated instrumentation to our ground-breaking gemmological research – are the result of our singular focus on building consumer confidence and public trust. Sriram Natarajan, Managing Director of GIA IndiaTechnology is increasingly playing a big role in diamond certification and authentication, with grading labs embracing it wholeheartedly to promote accuracy of the grading process, remove inconsistencies therein, and gain trust of consumers, say Suneeta Kaul and Vijetha Rangabashyam

Being a high-value item, the diamond trade comes with several inherent issues – determination of provenance, responsible sourcing, mining, processing, and so forth. In addition to these, the diamond industry also has to grapple with the tricky issue of certification authenticity, which has not been a fully transparent and foolproof process anywhere in the world Inconsistencies in grading reports has been a big problem in the diamond world, and industry stakeholders are hoping the increasing use of technology will protect both diamond sellers and buyers from illegal and unethical practices.

Grading labs, which are the custodians of consumer trust, are constantly reinventing the manner in which grading is done. In order to eliminate anomalies and human error, technology is being hailed as the silver bullet that will deliver the industry from its shortcomings.

Technology and grading labs
All major grading labs are adopting cutting-edge technology in order to make the grading process as foolproof as possible. For Gemological Institute of America (GIA), the world leader in diamond grading, the grading process and technology go hand in hand. Says Sriram Natarajan, Managing Director of GIA India, “GIA has been involved in supplementing visual grading with instrumentation for several decades. Right from introducing the first binocular gemological microscope with darkfield illumination in 1938, to developing the Diamolite and Colorimeter in the 1940s; from starting the laser inscription service in 1983 to being able to distinguish natural diamonds from laboratory-grown diamonds in 1995 -- we have continued to add more technological innovations into the grading process.

GIA was among the first to use blockchain technology in its grading report, working with blockchain technology expert Everledger on projects with Chow Tai Fook and JD.com. In 2020, GIA announced its association with IBM Research for automated clarity grading. GIA and IBM are planning other collaborations combining gemological evaluation with artificial intelligence. GIA will continue to add technology and automation to the grading process, ensuring that any additional technology is based on a deep understanding of the GIA International Diamond Grading System™ and its decades of research, for greater consistency and accuracy.”

Says Tehmasp Printer, Managing Director, International Gemological Institute (IGI), India, “IGI has been quick to innovate, adapt and upgrade technologies in the grading process. Over the last few years, we have been giving our clients the option of choosing a light return report, which uses highly sophisticated machines to calculate the light return value of the stone. This is better understood by the end consumer than just the various diamond measurements and facet angles mentioned in most other laboratory reports. Our biggest project so far, and the one we are working on currently, is the AI/ ML model for grading diamonds.”

Technological advances in diamond identification and light performance are expected to eliminate unethical practices such as inflated grading that have come to the fore over the past decades, which, in turn, will go a long way in fostering consumer trust. Says Donald A. Palmieri, President and Founder, Gem Certification & Assurance Lab (GCAL), “GCAL has been a frontrunner for consumer trust in lab results from its inception. We have strengthened our consumer guarantee twice in just over 20 years of existence with our cash-backed ‘Zero Tolerance 4Cs Consumer Guarantee’. Our guarantee is as much a cultural innovation as it is a technical innovation, providing our graders and technicians the process, software, and advance instrumentation to confidently work everyday with the ongoing goal of accuracy and consistency.

In 2011, we acquired Gemprint Corp, and immediately improved the software, processing time, and storage capabilities, and developed algorithms for instant light performance results during the Gemprinting process of a diamond or gemstone. GCAL has been a leader in chain of custody technology and process development, with the adoption of synthetic DNA (rough) and Gemprint (polished), dating back to 2008 with Rio Tinto ‘Mine of Origin’ and other chain of custody programmes.”

According to Ellen Joncheere, CEO, HRD Antwerp, “HRD Antwerp has always focused on trust and transparency in all its services. The global pandemic, and the political, economic, and social events over the last few years have increased our interests in improvisation of our knowledge. We have become more inquisitive about why things happen and where everythingwe use comes from.”

Use of AI/ML in diamond grading
For GIA, what matters is whether technology is able to help deliver consistent and accurate results. Says Natarajan, “The GIA AI clarity system is built on the globally accepted standards – the GIA International Diamond Grading System– that is recognised by everyone in the diamond value chain. The AI is fuelled by data from tens of millions of diamonds examined by GIA’s expert diamond graders in the Institute’s state-of-theart grading laboratories around the world. Whether driven by AI or skilled experts in a gemmological laboratory, reliable and independent results require highly calibrated instruments, strict adherence to grading standards, and enormous amount of diamond grading information as reference for human grading and training the AI grading system.”

GCAL has always acquired or developed the latest in technological advances in diamond identification and light performance through acquisition and further scientific advancements of those systems since our early beginnings as a significant lab in the American market in 2001. First, and most significantly, we eliminated existing elements of bribery and inflated grading being practiced in most diamond labs worldwide, by instituting a “Zero Tolerance 4C’s Consumer Guarantee for Diamonds” from day one. That guarantee pays the consumer the difference of cost in the retail market where they purchased the diamond if we graded the diamond a higher grade or weight that it actually is, or was at the time of grading. Donald A. Palmieri, President and Founder, Gem Certification &Technologies such as AI and Machine Learning (ML) are not new. In fact, they have permeated almost every other interaction humans have with technology. Even within the diamond industry, these technologies continue to gain traction, even as they demand copious amounts of data. “The future of diamond grading at GIA is exciting. IBM’s AI technology, combined with GIA’s expertise, extensive data and gemmological-research capabilities enables us to deliver advancements in consistency, accuracy and speed unlike any other organisation. The proprietary system, now in limited use in GIA’s New York and Carlsbad laboratories, will dramatically expand the reach of the Institute’s independent diamond grading reports. Initially concentrating on the most popular diamond sizes, GIA will scale the AI system to bring accurate and efficient diamond grading to more diamond sizes, shapes and qualities,” adds Natarajan.

According to David Block, CEO, Sarine Technology, “One of the major advantages of automated AI-based grading is that the level of consistency is significantly better than manual grading. Unlike graders, technology is ‘immune’ to the many factors that create most of the grading inconsistencies. Human graders are impacted by many external factors, including fatigue, environmental conditions, mental conditions (mood), lighting conditions, etc. Therefore, as AI-based grading consistency is much higher than manual grading, there are expected to be very few disputes onceAI grading is done.”

In view of high computation capabilities and access to high fidelity data, which are the key to the success of AI/ML algorithms, the efficiency of the diamond grading process is expected to improve greatly. “I believe that technology-based grading will increase consumer trust, going forward. Let’s take an example of carat weight and cut, which are both performed on technology. Today everyone blindly trusts these two parameters, which are measured and graded, using technology. Now,with AI grading, colour and clarity can also be graded using technology, which will only increase the consistency and accuracy of such grading, thereby increasing consumer confidence. I would even dare to say that today’s young consumer trusts technology even more than any manual process,” says Block.

When it is designed thoughtfully and executed successfully, technology has the power to transform the diamond pipeline and revolutionise its transparency. Of course, there are many players in the international diamond market who are making great strides in integrating their technologies into the pipeline for its general good. From the perspective of the team here at Opsydia, we have created a commercially viable, trustworthy and secure system that can eliminate ongoing issues in the diamond sector, specifically the fraudulent use of girdle inscriptions to pass laboratory-grown or treated stones off as similarly graded natural diamond. Andrew Rimmer, CEO, OpsydiaBut even so, it is early days for AI/ ML in diamond grading. According to Printer, “I think for the initial stage of training and refining, the AI/ML models would require human supervision. As the saying goes, ‘Garbage In, Garbage Out’, these models require high quality data sets to train and produce accurate predictions. Creating such high-fidelity feature sets would require human expertise. As the algorithms get more accurate and reliable down the line, I believe that this technology can eventually replace humans in certain aspects of grading. Over the last three decades, we have seen that like fingerprints, no two diamonds are the same. Each one is unique. Though AI/ML can be used extensively in the grading process, skilled and experienced gemmologists will still have to be consulted in certain cases.”

Doubts persist, however, about the extent to which this technology can be used, and whether it is even financially viable to use it widely. Adds Printer, “As the initial operating costs of these technologies go down, some of them might trickle down at the user’s end, especially the technology which is dependent on image data. As the imaging and computing capabilities of the hand-held devices are advancing ever so quickly, it may be possible to just upload the image of the gemstones on the cloud and our software may produce the grades for that particular image. On the other hand, for example, processes which require spectrograph data or 3D modelling of the stone may not be feasible for the retailers due to hardware costs.”

Overall, grading labs, and tech companies associated with them, are of the view that there is no turning back on AI/ML when it comes to diamond grading. In Block’s opinion, the impact of AI-based grading will be dramatic. “It will shorten the grading cycle time from a number of weeks (sometimes even longer) to days, enabling customers to shorten their overall manufacturing and sales cycle, thereby being able to offer their goods to the market much faster. Having grading capabilities onsite at the customer’s end will also save costs of shipping and insurance.Customers will have the flexibility of setting their own priorities, rather than being dependent on the quantities of goods at the lab.,” he says.

GIA’s expertise in instrumentation ensures that no matter where or when a diamond is evaluated by GIA, it can be compared to all other GIA-graded diamonds.” Sub-surface laser tech for linking diamonds with grading reports The future of grading is about closing any loopholes and addressing the practices that are undermining the integrity and trustworthiness of the grading process. What is needed is technology that works equally well in human hands as it does when integrated into more complex processes, including those powered by AI.With this in mind, diamond security specialist Opsydia has come out with sub-surface laser technology to create a permanent link between a diamond and its original grading report or blockchain record. Talking about this breakthrough technology, Andrew Rimmer, CEO, Opsydia, says, “Transparency relies on having access to credible and helpful information, and then being able to physically link a specific diamond to this data source. By harnessing subsurface laser technology, the Opsydia system can place alphanumeric sequences, brand logos and/or coded shapes beneath the surface of diamonds, without affecting their surface, polish or clarity grade. This sub-surface identifying information can then be used to co-relate a specific diamond to its grading report. The benefit of this over and above the current standard use of girdle inscriptions is that our sub-surface ‘identifiers’ can’t be polished away or removed without recutting a stone. Even if a falsified girdle inscription is applied, the identifier beneath the surface is permanent and immutable.The adoption of new technologies within labs will provide a renewed level of confidence across the industry, which will, in turn, filter down to consumers as they look for reassurance and trust.” Opsydia has two core classifications for its sub-surface identifiers: Nano ID and Loupe ID.Nano ID features do not affect the clarity grade of diamonds, even for flawless stones, and are not detectable with a standard grading microscope. Says Rimmer, “Nano ID features can be placed up to 0.18Some sections of the industry express fears about the human element associated with AI/ML, and its impact on the grading of diamonds. A question asked often is: Who gets to monitor the algorithms, and who will decide what data will be fed into the AI/ML software? Fears are also voiced about wrong, or even mischievous, inputting, which will defeat the very purpose of adopting it in the first place.

Expressing his opinion, Block says, “That is actually a very important question, as the quality and validity of the data that is used in the learning algorithms is critical to the ability to grade correctly. The learning data is collected from a very wide range of sources, and is verified prior to using it, so the chances of such mistakes or manipulations are very low. Sarine as a public company also implements best practices in terms of data security, monitoring and data access in order to ensure the grading algorithms and data are independent and not influenced by any external source. Lastly, unlike manual grading, where there is no audit trail to the grading process, technology has a complete audit trail, enabling us to track the grading process of each and every diamond that is graded.”

Adds Natarajan, “AI-powered automated grading needs an incredible amount of data. GIA has decades of experience and expertise built on the evaluation of tens of millions of diamonds by thousands of skilled, experienced GIA staff. In addition, GIA’s expertise in instrumentation ensures that no matter where or when a diamond is evaluated by GIA, it can be compared to all other GIA-graded diamonds.”

Sub-surface laser tech for linking diamonds with grading reports
The future of grading is about closing any loopholes and addressing the practices that are undermining the integrity and trustworthiness of the grading process. What is needed is technology that works equally well in human hands as it does when integrated into more complex processes, including those powered by AI.With this in mind, diamond security specialist Opsydia has come out with sub-surface laser technology to create a permanent link between a diamond and its original grading report or blockchain record.

Talking about this breakthrough technology, Andrew Rimmer, CEO, Opsydia, says, “Transparency relies on having access to credible and helpful information, and then being able to physically link a specific diamond to this data source. By harnessing subsurface laser technology, the Opsydia system can place alphanumeric sequences, brand logos and/or coded shapes beneath the surface of diamonds, without affecting their surface, polish or clarity grade. This sub-surface identifying information can then be used to co-relate a specific diamond to its grading report. The benefit of this over and above the current standard use of girdle inscriptions is that our sub-surface ‘identifiers’ can’t be polished away or removed without recutting a stone. Even if a falsified girdle inscription is applied, the identifier beneath the surface is permanent and immutable.The adoption of new technologies within labs will provide a renewed level of confidence across the industry, which will, in turn, filter down to consumers as they look for reassurance and trust.”

Opsydia has two core classifications for its sub-surface identifiers: Nano ID and Loupe ID.Nano ID features do not affect the clarity grade of diamonds, even for flawless stones, and are not detectable with a standard grading microscope. Says Rimmer, “Nano ID features can be placed up to 0.18 millimetres (180 microns) beneath the surface of a stone and must be read at >100x magnification, or with an Opsydia ID viewer – a device capable of reading identifiers that is currently in the development phase. The assertion that Nano ID features have no impact on the clarity grade of internally flawless diamonds has been independently verified and confirmed by two leading grading houses, including the Swiss Gemmological Institute (SSEF).

“In contrast, a Loupe ID is larger and denser and can be used in strategic places, such as the centre of the table, to allow valuers, grading houses and consumers to quickly distinguish between laboratory-grown or branded items from natural diamonds. Neither Loupe ID nor Nano ID identifiers affect the surface polish of a diamond.”

Future of tech in diamond grading
GIA grades millions of diamonds per year. Currently, its automated clarity grading is in limited use at its New York and Carlsbad laboratories, primarily focused on the smaller sizes of diamonds submitted. Eventually, automated grading will be in all of its global locations. Says Natarajan,“The diamond-grading skills of highly trained and experienced gemmologists and graders remain invaluable though. Grading large and high-quality diamonds is very complex and will require the insight of GIA’s expert staff. While AI-powered automated grading systems will allow GIA to deliver advancements in consistency, accuracy and speed, the AI will also need to learn from skilled gemmologists, researchers and graders to achieve those advancements.”

As of today, technology-based grading is a small percentage of the total diamonds graded; moreover, the technology is still new. Says Block, “We are starting to see a wider acceptance of grading based on AI. The technology is starting to mature and has only been available commercially for a few years at our grading labs. In the near future, Sarine will be introducing eGrading™ technology, which will enable the grading process to be done at source, rather than in a lab, thereby saving significant time and cost, as the diamonds will be able to be graded on-site, without having to send them to a lab. As the advantages of such technology are significant, we expect that within a number of years, most diamonds will be graded using AIbased technology.” The increasing use of technology for diamond grading has caused some quarters to have apprehensions that the human factor might become redundant. Grading labs, however, feel that the human grader cannot be replaced totally. Says Palmieri, “Human contact will always be necessary and will never completely replace the professional diamond grader.”

Agrees Block, “Human involvement will not become redundant, but the type of skill sets will change, transitioning the workforce from using a loupe to working on advanced AI technologies.” “Humans will always be the master of automation and can never be replaced. Human intervention is crucial for any automated service in any industry. Depending on the functionality and calibration of the equipment one could determine the accuracy and consistency in the grading process. However, human graders will always be needed to intervene and check the results in order to maintain the standards,” opines Joncheere.

Even though most of the technology being used for diamond grading is still in the nascent stage, the increasing use of existing and emerging technologies is inevitable. Used correctly, and with adequate human supervision, it can only benefit the entire industry and all its stakeholders.