Google employees programmed AI to invent new cake recipes


Running out of recipe ideas? Google’s artificial intelligence may soon come up with recipe ideas that don’t exist anywhere else.
Google engineers wanted to look into the science of pastry to try to understand the essential factors allowing to obtain crispy, spongy, or even soft cakes.

To carry out their experiment, they started by collecting hundreds of existing recipes of different cakes of all types. After converting the proportions of each ingredient and narrowing the list down to just a few essentials such as flour, sugar, butter, eggs or even yeast, Google engineers used a tool to build a model of machine learning to analyze all this data and detect what type of cake each recipe corresponded to.

By using this model to generate recipes, they were able to determine with precision for each of them which type of cakes corresponded to each recipe (bread, cookies, cakes), but also made it possible to identify the recipes considered as hybrids, possibly look like both cookie and cake for example

To check if these hybrid recipes were feasible, they went to the stove and therefore made a “cakie” (a sweetness having the appearance of a cake with the consistency of a cookie), as well as a “breakie” (cakes that look like cookies but have the consistency of a muffin).

Beyond the possibility of developing recipes that do not exist, the model of machine learning created by Google allowed its engineers to understand what gave the characteristics (crisp, soft, etc.) to each type of cake.

Source : Google


Source: Flux toutes les actualités – 01net by www.01net.com.

*The article has been translated based on the content of Flux toutes les actualités – 01net by www.01net.com. If there is any problem regarding the content, copyright, please leave a report below the article. We will try to process as quickly as possible to protect the rights of the author. Thank you very much!

*We just want readers to access information more quickly and easily with other multilingual content, instead of information only available in a certain language.

*We always respect the copyright of the content of the author and always include the original link of the source article.If the author disagrees, just leave the report below the article, the article will be edited or deleted at the request of the author. Thanks very much! Best regards!