In April, the Israeli-headquartered Razor Labs and the Gemological Institute of America, announced that they have developed a system to apply artificial intelligence (AI) to the crucial task of determining a diamond’s shape and cutting style.
According to the organizations, an algorithm enabled the determination of diamond shape and cutting style with 99 percent accuracy. GIA’s large dataset of real-world examples and expert shape determinations from experienced diamond graders provided the foundation for training the Razor Labs’ algorithm.
“As GIA introduces more automation into its grading process, this is an important remaining component that has become more challenging to classify,” said Tom Moses, GIA executive vice president and chief laboratory and research officer. “The introduction of more varieties of fancy shapes and cutting styles made this an important initiative to complete.”
Already in 2018, GIA had teamed up with IBM Research to develop an automated system for grading the clarity of a diamond. There, too, AI technology was developed from analyzing tens of millions of diamonds submitted to GIA laboratories in the past.
GIA was not the only prominent organization to have taken the step. In 2016, Sarine Technologies unveiled its first two automated color and clarity grading equipment. Today, the Israel company claims to the first organization to provide automated 4Cs grading, driven by AI, which enables objective, reliable grading with unprecedented levels of consistency.
According to Sarine, which also is Israeli headquartered, the results produced by its system are dependably more reliable than those obtained by human graders. Color grading, for example, which relies on the gemologist’s eye “is obviously prone to inconsistency,” it stated. “This is due to the subjective nature of human perception. Errors margins in manual color grading are noted not just between different gemologists, but between different work sessions of the same gemologist.”
“The way to overcome this is via standardized reliable technology-based color grading that demonstrates consistent accuracy rates that surpass those achieved by manual Color grading” Sarine states.
HARMONIZING GRADING STANDARDS
Gemologists typically are called on to identify or grade a gemstone. In the case of the former, they will be expected to classify the type and variety of the specific gemstone, if it is indeed one and not another object with a similar appearance; to determine whether it was formed in nature or synthesized in a factory or laboratory; whether it was subject to any treatments that caused it change its appearance, and if so, to identify those treatments.
As far as grading is concerned, gemologists will be requested to describe a gemstone according to predetermined criteria, such as the famous four Cs (clarity, color, caratage and cut) in the case of the diamonds, thereby scoring it for quality.
But, while there no official standard, there is a dominant organization in the industry that has has set what arguably could be described as the benchmark standard – namely the one against which almost all other are judged. It is GIA, which was established in 1931 by a retail jeweler called Robert M. Shipley.
For many years attempts to remedy the situation by creating a single, harmonized grading standard, recognized and endorsed by ISO, were unsuccessful. That ended in September 2020, when ISO published International Standard 24016, specifying the terminology, classification and the methods to be used for the grading and description of single unmounted polished diamonds over 0.25 carats in weight.
The Gemological Institute of America (GIA) has teamed up with Razor Labs to develop automated systems for grading the shape and cutting style of diamonds, after working with IBM Research on an automated system to grade clarity. As GIA introduces more automation into its grading process, this is an important remaining component that has become more challenging to classify,” said Tom Moses, the gem institute’s executive vice president and chief laboratory and research officer.
Sarine Technologies, whose system produced the scan of a polished diamonds exact proportions seen above, claims to the first organization to provide automated 4Cs grading, driven by AI.
But the work is not yet done, for an ISO quality control standard for the grading of diamonds of 0.25 carats and less does still not exist. However, work currently is underway in ISO technical Committee 174 on the still-to-be-approved International Standard 6893.
GROWING NEED FOR AUTOMATION
The harmonization of standards and the automation of the grading process are likely to have a fundamental impact on the industry in the coming years. As the online trading of diamonds not only becomes commonplace, but essentially essential, a fast, reliable and easily communicated system to accurately describe every diamond on offer is essential.
The long-used system by which diamond companies would cherry pick what stones to send to be graded, sometimes selecting the lab according to the type of merchandise concerned, is becoming archaic. To an ever-greater degree, almost every diamond being put out into the market requires some type of report.
But a system that always requires the intervention of highly human graders, regardless of the number of goods that have to be handled, is clearly not sustainable. Systems that are able to do the work automatically, such as those being developed by GIA and Sarine, becomes a favored alternative, especially if they are made available for use on-site at the major diamond trading companies.