AI System Learning to Create New Era of Fragrances

“The nose knows,” but could artificial intelligence change that popular saying? Engineers from the Thomas J Watson Research Center at IBM are working on an AI system that might one day create the scents we have in our laundry detergent, air fresheners, or even perfume.

The IBM team collaborated with global flavor and fragrance producer Symrise on the AI that not only makes new fragrances but considers trends within the industry and in materials used to produce the scents.

The AI is called Philyra, and the researchers hope it can revolutionize industries surrounding new fragrances.

“Philyra does more than serve up inspiration – it can design entirely new fragrance formulas by exploring the entire landscape of fragrance combinations to discover the whitespaces in the global fragrance market," wrote Principal Research Scientist from IBM Research Richard Goodwin.

How Philyra works

One of the biggest keys to succeeding with fragrancing is novelty. The researchers programmed their AI to have a knack for predicting the next big fragrance by examining data from other popular smells. 

Philyra to identify fragrances close to existing fragrances. They expect that the larger the distance between fragrances, the more unique the smell will be to the public. A seasonal pumpkin spice smell, for example, could be adjusted slightly to provide a new take on a classic scent. 

“In the case of fragrances, the art and science of designing a winning perfume is something humans have explored for hundreds of years," Goodwin wrote. "Now, perfumers can have an AI apprentice by their side that can analyze thousands of formulas and historical data to identify patterns and predict novel combinations, helping to make them more productive, and accelerate the design process by guiding them towards formulas that have never been seen before.”

Symrise has already used Philyra to create two unique perfumes. Both those scents will launch next year for O Boticario -- a massive global beauty company. During that process, Philyra created a concoction of smells from a database. Those combinations were then considered by a master perfumers who tweaked them if needed. By understanding what Symrise's consumers most frequently bought, Philyra could suggest scents in line with other successful smells. 

"Symrise’s longer-term goal is to introduce this technology to their master perfumers around the globe and continue to use the solution for the design of fragrances for personal care and home care products," the companies explained. "Symrise also plans to introduce Philyra into their Perfumery School to help train the next generation of perfumers, firmly embedding AI into the heart of its organization."

History of making fragrances

Becoming a perfumer -- a scent professional of sorts -- requires a naturally excellent sense of smell and creativity. At the basic level, perfumers have to have a wide variety of knowledge about smells both good and bad. Perfumers have to distinguish between smells, even picking up on subtle notes of a fragrance through much stronger scents. 

[see-also]

For centuries, being a perfumer for groups normally meant taking up an apprenticeship with another skilled perfumer to learn the trade. It wasn't until 1970 that schools opened up to teach the art of it. The ISIPCA was the first school designed for people with extraordinary noses; the entrance examination was rigorous, and it required a background in organic chemistry. 

Skilled perfumers can get some of the most impressive and unique job locations. Companies like Robertet, Firemenich, Takasgo, and Symrise -- the company involved in this study -- hire perfumers. Even NASA has a 'chief sniffer' position. 

Don't expect this AI to take over hundreds of unique perfuming jobs. It still requires human oversight from people with the training and knowledge to know what scents work and what won't. But for now, the IBM and Symrise team will continue working on applying a very human experience to an AI system.