The computer that makes a meal
If IBM’s Watson could beat “Jeopardy” champion Ken Jennings, can the same artificial intelligence also be used to combat food waste?
The creators of IBM’s Chef Watson certainly think so.
The Watson division of the technology giant has developed an algorithm that pairs unexpected ingredients based on their flavor compounds — think chocolate and edamame in a burrito, or porcini mushrooms in pudding.
Right now, the technology — which underpins a 60recipe cookbook, “Cognitive Cooking With Chef Watson” — is used by chefs, bartenders and home cooks to experiment with flavor combinations. But it could have much broader applications, potentially helping to cut down on food waste, a problem that persists on an individual and global scale.
“There are other areas where we’re going to use computing in food,” said Florian Pinel, lead engineer for Chef Watson. “We can use the same algorithm we created for Chef Watson to provide personalized meal plans. This will allow us to reduce food waste and create varied meals for people who have severe dietary constraints.”
Imagine, Pinel posited, a digital tool that can scan and analyze the contents of your refrigerator and suggest recipes based on your leftovers, your location, the time of year and any food allergies you may have.
“We’re going to see important developments in that area,” Pinel said.
Food-matching software could also help identify when ingredients from halfway around the world can be replaced with a locally farmed substitute with a similar flavor profile — potentially lessening the carbon footprint of produce or grains that would otherwise have to be shipped from afar, said Pascal-Xavier Van de Goor, an account manager at Foodpairing. The Belgian company uses an algorithm to pair ingredients based on similar aromas, and sells the information to chefs, bartenders and food manufacturers looking to experiment and test new products.
Foodpairing has amassed a database of about 2,000 ingredients that can help predict whether consumers will like flavor combinations in products that have not even been introduced yet, based on how successful previous similar combinations have been.
“Foodpairing only gets more and more intelligent and stronger to predict the next successful flavors,” Van de Goor said. “It can grow to the point of personalized food recommendations based on your diet and allergens, and discover new things that might be better for you and the climate.”
“This will allow us to reduce food waste and create varied meals for people who have severe dietary constraints.” Florian Pinel, Chef Watson lead engineer