Sony FlavorGraph

2021: Launch an AI system that determines the correct combination of ingredients in dishes

In April 2021, Sony introduced a system that, using artificial intelligence algorithms, determines the correct combination of ingredients in dishes. The decision was called FlavorGraph. It was created together with scientists from Korean University.

As noted in Sony, chefs from around the world found successful product combinations thanks to intuition and experiments. Many of the classic combinations, such as cheese and tomatoes or pork and apples, were subsequently explained scientifically - these ingredients had dominant odor molecules that worked well with each other.

Example of a combination of tastes created using the Sony tool

To examine the reasons for this, a team of experts analyzed both information about the ingredients molecules and information about their use in real history. Then a database of FlavorGraph was created, which was filled with flavor profiles based on 1561 molecules. Almost 1 million recipes have also been studied to understand how the ingredients were combined in the past.

In the database, you can find obvious combinations that have been known for a long time (for example, ice cream and cookies), and something new (for example, white wine and mushroom mashed soup). The system did not find anything really extraordinary, like caviar and white chocolate, by April 2021, but FlavorGraph is called only a starting point. Over time, technology will get better, and this system will allow you to find more and more combinations of ingredients, as well as substitutions of ingredients that are considered unhealthy or rare, Sony emphasizes.

FlavorGraph is not Sony's first culinary AI project. Earlier, the Japanese corporation introduced the AI Chef Assisting Cooking robot assistant, which helps the cook throughout the cooking process - from preparing ingredients to serving a dish. Robots can be controlled remotely.^[1]

Notes