IBM Computational Creativity Food
A food truck powered by the same artificial intelligence that beat humans in rounds of Jeopardy! has been rolling out across the US over the last few weeks, serving up computer-generated recipes like Baltic apple pie and pork tenderloin with vanilla.
It’s a concept that stems from IBM’s Watson supercomputer, which famously beat humans at their own game – "Jeopardy!" – back in 2011.
Only this time, instead of answering general knowledge questions, the technology will be used to help chefs churn out unexpected and surprising dishes like Czech pork belly moussaka and Ecuadorian strawberry dessert to attendees of the South by Southwest festival in Austin, Texas next week.
Developed in partnership with the Institute of Culinary Education (ICE), the computer works like a roulette wheel: after choosing a main ingredient and setting a few parameters like regional cuisine, the computer spits out a recipe that challenges ICE chefs James Briscione and Michael Laiskonis to come up with dishes like Cuban lobster bouillabaisse or Swiss-Thai asparagus quiche.
Next week, attendees of the South by Southwest festival will be able to sample the IBM Food Truck’s first multi-course meal, which will include computer-generated dishes like Czech pork belly moussaka made with peas, celery root, parsley roots, red bell pepper, Swiss, cheddar and cottage cheese, and Russian beet salad with prunes, cucumbers, pickles and beans.
But more than just being a gimmicky food truck, the technology also speaks to a larger idea called computational creativity: machines that try to manufacture creativity.
What was once regarded as the final bastion and uniquely human attribute, creativity, is being turned into another automated and mechanized tool by engineers at IBM, who say that cognitive computing could be used to help chefs become more creative.
"We’re all familiar with search engines and how they can help us find things,” said IBM research scientist Dr. Lav Varshnay in a YouTube video. "But can a computer help you come up with things that have never existed before? That’s what we’re trying to do."
In addition to pulling up novel flavour pairings based on food chemistry, the supercomputer can also model human perception and therefore predict foods that would be flavourful and palatable for the human taste bud.
After spitting out a series of ingredients – some of which, like pork tenderloin and vanilla, may cause a few raised eyebrows – chefs are tasked with pulling the flavours together.
The technology could also be extended to other creative industries like fashion, travel and design.
Engineers at IBM have in some ways beat renowned Spanish chef Ferran Adria with their Watson supercomputer, as the Michelin-starred chef is also working on a tool to help chefs innovate their art.
Bullipedia, a culinary version of Wikipedia set to launch next year, will include a database of foods and ingredients that can be used as a creative tool for chefs to stretch their imagination.
Watch an introductory video on how the cognitive cooking tool works above.