Artificial intelligence created fragrance
Symrise tested innovative methods to create a new fragrance in Berlin. An artificial intelligence (AI) called “Phylira”, developed jointly with IBM, acted as a digital perfumery assistant. Interactively and collaboratively, all interested Berliners helped the system to develop the capital’s fragrance. By the end of June, “#Berlin 3.0” was ready to be tested.
Visitors to the Binkini shopping mall between the Zoological Garden and Kurfürstendamm (Berlin, Germany) were able to take part in an exciting project until 29 June 2019. Together with an AI, they developed a new fragrance. The noses of the shopping enthusiasts were challenged for a total of six weeks. They were allowed to evaluate five different basic fragrances according to their own preferences. Together with Senior Perfumer Marc vom Ende, Philyra developed these so-called scent cores on the basis of specific terms that were to represent the spirit of various Berlin districts. The choice was between Grunewald, Kreuzberg, Prenzlauer Berg, Schöneberg and Wannsee. Everyone was allowed to play hobby perfumer and judge on a tablet whether its perfect fragrance smells more like the exclusive, wooded Grunewald or more like the alternative, colorful Kreuzberg. The algorithm of the computer learned continuously from the evaluations of each participant. At the end of each week, Philyra created an “intermediate perfume” based on the tablet evaluations. She received help from the trained nose and extensive expertise of the Symrise perfumer at the end. Afterwards, shopping center visitors were able to sniff this “intermediate perfume” and share the emotions it arouses in them. From this and from several open questions to the testers, the AI learned more. During the last week of the event at the end of June, the final computer-generated fragrance “#Berlin 3.0” was presented. Participants could receive a description of the fragrance by e-mail and test it at the Philyra booth. “We were excited to see where this collaborative fragrance creation experience supported by artificial intelligence would take us,” said Clarie Viola, Symrise vice president of digital fragrance strategy. “We are delighted with the result, which has enriched Berlin with a unique brand new fragrance”.
The artificial nose with a large brain
The Bikini Berlin is not Philyra’s first location. This AI system was developed by Symrise and IBM Research to revolutionize the perfume market. The computer can access huge amounts of historical data and thousands of formulas, it knows all available raw materials. From an endless variety of possible scent combinations, it can identify an unprecedented composition. And all this very effectively and quickly, with high precision. Because the slightest deviation in the quantity of an ingredient can ruin the entire fragrance – or turn it into a new box-office hit. Philyra does not act alone. Experienced perfumers use the enormous wealth of knowledge and speed, but the human nose refines the final fragrance once more.
Brazil is the leader in AI
Philyra’s first finished creations make hearts beat faster in Brazil. Just in time for Valentine’s Day, 12 June, the cosmetics company O Boticário presented two new fragrances, tailor-made for local users. As now in Berlin, the AI and an experienced perfumer – in this case David Apel – worked hand in hand. Philyra analysed consumer preferences and suggested an optimal fragrance combination. David Apel’s fine nose perfected the end result.
“We see great potential in the new technology. Because the fragrance market is hotly contested and companies are looking for something new and agile to shape the market sustainably. Today, Phylira is working on alcoholic-based fragrances. It is also suitable for other scented products such as deodorants, body care and detergents,” says Clair Viola of Symrise. “The algorithms of an AI like Philyra are constantly learning from their nature – from industry trends, regional preferences and existing combinations. With this wealth of knowledge, the professionals are exploiting the potential of large amounts of data and exploring additional possibilities for fragrance creation.”