A Japanese app is now serving a local sushi chain by grading the quality of the tuna it serves to customers.
Tuna Scope, developed by advertising giant Dentsu and other companies, uses artificial intelligence to perform “tuna examination,” a vital artisanal skill that may soon disappear as practitioners dwindle.
How it works: Tuna Scope is based on an AI model that determines tuna quality based only on a cross-sectional tail scan.
- Developers took about 4,000 tail images — approximately equivalent to the number a human examiner might see on a 10-year path to proficiency — and input the data into the AI program.
- Using machine learning, the app reportedly mastered the skill in just a month, achieving an accuracy level of up to 85% when compared to grading results of human counterparts.
- With just one picture, the app grades the tuna on a five-point scale based on visual characteristics, including color, sheen, firmness and layering of fat.
- A particular tuna that emerged on top of the app’s ranking was later served on a sushi restaurant in Tokyo and received a 90% customer satisfaction rate.
- The app is currently used by restaurant chain Kura Sushi Inc., whose buyers cannot freely travel overseas to purchase fish due to COVID-19 restrictions, according to the Asahi Shimbun.
Why it matters: Developers decided to pass the skill of “tuna examination” to AI as experienced human practitioners dwindle, likely as a result of Japan’s aging population.
- Human tuna examiners, who draw expertise from years of training, have fallen to less than half of the industry’s golden age, according to the app’s developers.
- “[Tuna examination] is a skill that allows master tuna merchants to determine properties such as the flavor and texture of a tuna just by eye, without ever tasting it themselves,” its website read. “As the number of practitioners of this craft dwindle, these trade secrets are in danger of disappearing, so we decided to pass the torch to AI. [We] want to ensure that people around the world could enjoy the same standard of delicious tuna, even into the far future.”
- Keiko Yamamoto, a chef and sushi instructor based in London, confirms that tuna can be graded based on appearance alone. “I’ve had to cut fresh tuna every two weeks, so I know what’s good, what’s not good,” she told The Verge. “It [highest-quality tuna] looks bouncy, or soft, maybe, to your eye. Good quality tuna is silkier and shiny.”
- However, not everyone is onboard with the idea of automating the skill. Richard Cann, a sales manager at T&S Enterprises, which supplies many of London’s top Japanese restaurants, told The Verge that “We always have and always will do it by eye.”
- Cann said that they buy the tuna whole and divide it themselves. He also pointed out that their business has relationships with chefs who trust their workers, saying, “We’re a people business in everything we do.”
Tuna Scope is only available in Japan at the present. However, its developers plan to expand “in the near future.”
“By expanding beyond Japan, and continuing to acquire data from fisheries around the world, TUNA SCOPE aims to create a world standard for tuna quality in the near future,” they said. “The technology developed for TUNA SCOPE has untapped potential for applications across a wide variety of industries. By teaching AI the secrets and nuances required to perform the tasks humans have perfected over the years with the naked eye, we may gain a new ally against a host of problems that humanity faces today. That future might be closer than we think.”
Feature Images via Tuna Scope