IBM’s Chef Watson gives us a glimpse of the creative potential for artificial intelligence. Although Chef Watson’s recipes might suggest ingredient combinations and styles of dishes humans would never have considered, its ability to analyse data and overlay scientific info makes working in the kitchen alongside human chefs effective.
IBM’s Chef Watson just might be the solution home chefs need to overcome a pantry with only a few ingredients that “couldn’t possibly go together” or a welcome inspiration to professional chefs who want to evolve their menu through “cognitive cooking” and take advantage of seasonal ingredients. Watson, IBM’s artificial intelligence platform, had already shown its expertise on “Jeopardy” and for lung cancer treatment, and as Chef Watson it’s assisting IBM achieve its mission to “help people discover new ideas”—in this case new ideas for the kitchen.
How AI is being used to develop recipes
The first step in having Chef Watson help create a recipe for you is to pick any ingredient you wish to use in your culinary adventure—pineapple or anise or sea bass—any ingredient you have on hand will do. You also have the option to add up to three additional ingredients for Watson to put through its algorithm; you can enter as few as one ingredient to as many as four and any number in between. Next, you tell Watson the dish you wish to make and for what meal. Maybe you want a burger or a chutney; an empanada or ice cream. Finally, you can select a style or a theme—Chinese, easy, Eastern European and more! Voila! Chef Watson will take your input, compare it to its data and spit out 100 recipe options for you to choose from complete with ingredient lists and instructions on how to prepare it. Now, it’s all up to you to create your culinary masterpiece exactly as the recipe dictates or swap things out and customise it to your liking.
The technical details
As any reasonable person would, you might question how a machine that has no taste buds and couldn’t distinguish sweet from sour because it has no sense of taste could possibly be entrusted to create recipes. To start, Chef Watson’s creators fed it with nearly 10,000 recipes from Bon Appetit’s archives and it used natural language processing to analyse the recipes to learn the underlying logic, understand the ingredients that are most commonly used and how ingredients were combined.
In addition, Chef Watson also has the benefit of learning the science of complementary flavour compounds in such depth and complexity in an amount of time it would be impossible for a human chef to learn. Not only can Chef Watson pair two ingredients, but it can look at “pairings of six, seven, eight, nine ingredients without a problem.” Ultimately, this allows Chef Watson to advise cooks and to discover and create totally unique recipes by using flavour compound algorithms.
With this knowledge base established, once ingredients are typed into Chef Watson’s interface, it gets to work scanning its recipe database and calculating what foods appear together in dishes and then reviewing the chemical affinities of ingredients to create recipes. As it continues to process recipe requests, Chef Watson refines its algorithms by incorporating user feedback.
The professional version of Chef Watson takes it a step further. Researchers scraped the internet and added 30,000-plus recipes and added information about the molecular makeup of smells and flavour compounds in food as well as academic research known as “hedonic psychophysics” that investigates the smells and tastes people find pleasurable.
Some of Chef Watson’s innovative although often eccentric recipes are available in the cookbook, “Cognitive Cooking with Chef Watson: Recipes for Innovation from IBM & the Institute of Culinary Education” that features 65 original recipes generated by Watson. As you might imagine, Chef Watson’s specialty is unusual combinations.
Ideas and insights you can steal
Chef Watson helps us understand the types of “creative thinking” Watson and other artificial intelligence machines can support. Although Chef Watson might provide the spark to allow our creative endeavours in cooking to go beyond the restrictions our minds might see for impossible ingredient combinations or culinary concepts and to discover a new possibility that might actually taste good, cooking is still a collaboration with human chefs. Working together, Chef Watson and human chefs can realise more together than they can working alone. For the foreseeable future, Chef Watson can’t make its recipes a reality without human intervention and its algorithms help humans embrace the art of the possible.