This is the weirdest recipe IBM's supercomputer chef has created

Flickr / KariSöderholm works in the kitchen with an IBM computer known for destroying humans with chess and Jeopardy.

However, all the chefs are occasionally experimenting with cooking. Even though there is a successful recipe on the shelf, Chef Watson's app has some very tasty results.

The chef Watson application takes the ingredients that the user needs and mixes the recipe according to their shared taste memory.

On August 25, Reddit AMA with IBM, Bon Appetit magazine, Culinary Education Institute announced that the original recipe of Chef Watson chicken breast tacos might be delicious, but it is actually impossible.

The Reddit user's ZipBoxer wrote as follows. "I have played the Chef Watson application, but sometimes it will bring fun results." For example, [a] needs mustard powder. / 2 cups of peased green beans) cut into 3/4 pieces and placed on the grill. "

Baked 3/4 peas may taste of chicken tacos, but reasonable human chefs will not try to burn them. Researcher Patrick Wagstrom at IBM Watson has blamed this anomaly for natural language failure, super intelligent machine results, lack of commonsense reasoning, and a sharp understanding of the language.

Mr. Wagstrom says it as follows. "Since it is probably necessary to open and slice the original ingredients, we recommend naturally parallel ones to wrap the green peas."

In fact, Chef Watson's initial formulation requires "cut vegetables to 3/4-inch size". From the viewpoint of Chef Watson, peas are vegetables, so it will work

Although it may seem to a stranger, I am receiving a unique form of culinary training called machine learning. According to Washington Post, the Watson chef captured a large amount of unstructured data "recipe, book, academic research, tweet" and analyzed patterns which can not be detected by human eyes.

For this pea-loving expression ceremony, Chef Watson analyzed 10,000 Bon Appetit recipes. Then it looks for a statistical correlation between the components that often appear in the equation.

It is not the first time Watson's chef recommended the impossible steps. According to Mr. Wagstrom, in the early version of this software, "one was a refrigerated goat, two was a tequila to annihilate".

This ingredient list was not organized by the chef, but organized by IBM's supercomputer Watson. They trained thousands of recipes to understand which combinations of foods will work and to understand the synergistic effect; they will use scientific data to find people at the molecular level with fun things I trained. This gives them the idea of ​​pleasure; they tell it to make a recipe that has never been used to it, and that gives it the idea of ​​surprise. Basically, they teach it to it - and more importantly, they teach it to make their own dishes

I'm looking for recipes of chocolate, apricot, beef burrito with mush Makamar's chicken breast. This is not an acid of Heston. These ideas were born from IBM's supercomputer "Watson", and its first recipe "Cognitive Cooking and Chef Watson" will be released this spring. Watson has traditionally obtained data on foods that humans enjoy. Computers can learn and find recipes, taste characteristics and reasons behind compounds, so that it can suggest a combination of (new?) A new flavor. The resulting list of materials was converted to recipes by the American Culinary Education Institute chef and the publisher promised "a rare combination of materials that humans could not imagine." Curious, I caught an early copy. How do you cook?

This is not the first time that IBM Watson was called on to stimulate artificial intelligence. Last year, Watson created the IBM Chef Watson using the BonAppétit application. The app will discover combinations of unexpected flavors, encourage creative daily mealtime challenges, and encourage family chefs around the world to bring new ideas to their kitchen. The free Watson-led app includes knowledge gained from Watson's training to learn about 10,000 recipes in the BonAppétit database, and how to use ingredients in various dishes and cooking styles. According to IBM, IBM combined this information with an understanding of the system's food chemistry and human preference palatability and integrated valuable insights to participating users. You can use cooking app here