By Florian Pinel
My time at South by Southwest (SXSW) could be called, “Five Days in a Food Truck,” or, “What it’s Like to Cook with a Computer.”
For the better part of a week, chefs James Briscione and Michael Laiskonis and I prepared everything from Caribbean snapper fish and chips, to chocolate burritos with an Austrian twist for the throngs of techies who hustled between presentations by Neil deGrasse Tyson, Edward Snowden, and more.
IBM at SXSW makes sense. Food at SXSW does too. But food from IBM at SXSW? That’s the question I got most when handing out plates of Vietnamese apple kebab on day one, through day five’s two dishes of chili and poutine (more on this Canadian favorite, later). We were there with the Institute of Culinary Education to demonstrate how a computer could act cognitively and creatively – using unique recipe generation as the example. Watch my interview with Engadget to get a close up of how the system, operating over the cloud, comes up with never-before-thought-of recipes.
Day 1: Vietnamese Apple Kebab
When the scientists and chefs get word of the most-voted-upon dish, we put the system through its paces – adding our own twists. We entered Day One winner “kebab” into the system, but it was up to us to decide the region and a key ingredient. Did you know that strawberries and mushrooms share a flavor compound, making them complimentary? ICE Chefs James Briscione and Michael Laiskonis knew all about g-dodecalactone thanks to the system (try cross referencing that in a textbook!)
To get a recipe complete with complimentary ingredients and other constraint options (such as removing or adding meat or vegetables), our cognitive cooking system connects to SoftLayer’s cloud. All of those algorithms have to churn through quintillions of options. SoftLayer’s baremetal servers can crunch combinations at incredible speed; the chefs waited seconds for Watson to tell them about g-dodecalactone.
Day 2: Caribbean Snapper Fish and Chips
Voters ordered up fish and chips on Saturday. The creativity of the system to combine fennel and coriander met the creativity of the chefs to take the fish out of deep fryer and serve it ceviche style – raw snapper with plantain chips.
What we want the system to do in the future is create dishes for those with dietary restrictions or allergies. Or, imagine living in a place where certain kinds of foods are difficult to find. The system could match flavor profiles of the called for (but unavailable) ingredient with something that’s on hand.
Day 3: Austrian Chocolate Burrito
The burrito actually made its debut at IBM’s Pulse conference in Las Vegas. Twitter voters clamored for it again, and on day three the Austrian chocolate burrito was back on the menu. I think this dish is a good example of the system’s flexibility. It allows for the chefs to make ingredient replacements. As Chef Laiskonis put it “by looking at this dish through an Austrian lens, we could go with subtle flavors like orange, apricot, and cinnamon.”
Day 4: Belgian Bacon Pudding
Speaking of Chef Laiskonis, ICE’s Creative Director, as a pastry chef he made a personal plea to the Twitter voters for pudding. The voters heard him, and on Day Four, the team made Belgian bacon pudding. The system must have found another compound like g-dodecalactone. How else could we explain putting porcini mushrooms in a pudding?
Day 5: Peruvian Potato Poutine and Indonesian Chili con Carne
And some dishes went off menu. A Canadian contingent tweet-jacked the Food Truck voting with the dish, #poutine. So, while the French fry and gravy favorite was not on the official list of dished to vote for, Twitter spoke, and we listened. #Poutine drew so many votes over the last few days that we couldn’t deny the people of what they wanted, and added it to on-menu winner, #chili – with a twist.
For poutine, Chef Briscione incorporated the system’s suggestions of tomatoes, cauliflower, onions, and cumin (not usual additions by the Quebecois). “Messing with one of their favorite foods really scares me!” said Briscione.
The chili con carne’s additions of lemon grass and cubeb peppers give the dish a distinct Indonesian flare. “Is Texas ready for this chili?” asked Briscione.
Whether trying to put a new twist on an old standard in a restaurant, or meeting government health guidelines for school lunches, this cognitive cooking system will deliver combinations, vetted by reams of data, that will surprise and please. And we can apply these “creative” algorithms to many other domains. A cognitive system could impact industries as varied as fragrances and fashion, to retail.
About the IBM Food Truck
- For more about this research and how it can transform your business, contact Research Staff Member and Manager of Consumer Modeling Anshul Sheopuri.
- To learn more about Watson and the Food Truck project, read IBM Fellow Rob High’s perspective.
- To learn more about Watson and the Food Truck project, read IBM Research scientist Florian Pinel’s perspective.
- Get a chef’s perspective from ICE Creative Director Michael Laiskonis.
- Download images of the Watson Food Truck.
- Have more questions? Check out the Watson Food Truck fact sheet.