IBM Research has teamed up with fragrance specialists at Symrise to create an AI system that can design completely new perfumes.
Developing a scent is a combination of science and art, and normally requires years of human expertise, but Philyra - an AI named after the Greek goddess of perfume - can create new formulations that only need a little tweaking by professionals to be ready for use.
Philyra uses machine learning to analyze thousands of raw materials and previously designed perfumes, identifying patterns and new combinations that haven't been tried before.
The smell of success
Philyra can suggest the amount of ingredients to be added to a new formula, and offer alternative raw materials that could produce the same effect. It can even predict how people will respond to the smell based on demographics including age and nationality.
It's not just a proof of concept; IBM and Symrise are beginning with fine fragrance, but Philyra could also develop new aromas for products like fabric conditioner, shower gel and furniture polish.
The aim isn't to replace humans, but to provide a tool that can augment their skills and knowledge. "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," says Symrise, which is planning to introduce Philyra to its perfumery school, where it can help train the next generation of experts.