Isn’t making perfume dependent on your sense of smell? While not unimportant, analyzing data is a significant part of the process that goes into creating a new scent, and artificial intelligence is quite skilled at doing this. The first artificial intelligence (AI)-the developed fragrance is now on the market in Brazil thanks to a collaboration between IBM Research & Symrise, a global manufacturer of perfumes and tastes with headquarters in Germany and clients including Covergirl, Donna Karan, Avon, Coty, and more.
Philyra
David Apel, a perfumer, collaborated with IBM’s Philyra, an artificial intelligence (AI) apprentice, to develop two new scents for the Brazilian cosmetics brand O Boticário in time for the nation’s Valentine’s Day celebration this year. They were especially seeking a scent to market to millennials and members of Generation Z, two groups they knew would find artificial intelligence-generated fragrances intriguing. AI was formally introduced to the fragrance business through this cooperation.
How is Philyra put to use?
A perfumer has access to 1,300 aroma-building components, including synthetic perfumes and compounds from flowers, herbs, spices, and fruits. Symrise provided Philyra with access to its database, which has 1.7 million formulations derived from different combinations of these ingredients. Also provided to Philyra were information on popular perfumes in various nations, age categories, and genders. The AI system, which is free from cultural prejudice, personal preference, knowledge, experience, or familiarity with a material, discovered options that hadn’t been considered before after processing the data with a deep-learning algorithm. Based on the data, Philyra produced perfume compositions that need to appeal to a specific market. The AI-generated formulae were improved by the perfumer.
It’s critical to realize that Philyra’s deep-learning algorithm enables it to learn the potential combinations of different components and does not just carry out instructions given to it by humans.
The AI’s analysis of the data and formulation of formula suggestions produces formulae that humans, for whatever reason, had never before considered. It turns out that understanding a perfume’s composition is more important than using your sense of smell when developing a new scent.
AI software can learn and provide suggestions:
- The novelty of a scent by contrasting it with other perfumes that are sold commercially.
- Substitutes and complements for raw materials to be used in formulations
- Dosing of the raw material
- Predict the behavior of people
For Valentine’s Day launch of O Boticário, Philyra worked with perfumer David Apel to create three different versions: an AI-only version, a version where the perfumer created some changes, and a third version where the AI output was only used as a suggestion but the finished version was the perfumer’s method. The “vast majority selected for the 100% Automation scent” when all three were tried.
AI scents Egeo Over Me and Egeo With You
This year, the songs Egeo Over Me and Egeo With You by Philyra and Apel for O Boticário were released on June 12 in time for Brazil’s Valentine’s Day. However, the brand encourages every person, regardless of gender, to choose whichever scent resonates with them the most. Egeo ON Me is indeed the feminine scent, while ON You is indeed the scent that is more usually male. Fruits, flowers, wood, caramel, or even condensed milk are all present in the smells.
Adding machines to the process
There are concerns about the future of human perfumers in the fragrance industry, much as there were when ai systems were initially deployed in other sectors. One person, Abel, sees a.i. as an opportunity for human-machine cooperation rather than a danger. Machines without prejudice can assist people in overcoming obstacles to creativity and offer fresh ideas. Additionally, the time needed to produce a new scent is significantly reduced when employing AI, which generally takes 6 months to four years.
The use of AI by a fragrance company to create novel combinations may have started with IBM Research and Symrise, but it won’t be the last. CARTO, a method created to leverage the Odors Valuation Is a process to enhance olfactory effectiveness in final formulae of tastes and perfumes, was presented this year by flavor and fragrance manufacturer Givaudan. It turns out that both art and science go into making scents. AI will support human knowledge in scent development but the emotional advantage of master perfumers over AI remains.