Can computers be creative?
In 2011, IBM Watson
, a supercomputer that combines artificial intelligence and analytical software, made headlines after defeating two Jeopardy! champions
. Since then, Watson's skills have been applied effectively in medical advancements, psycholinguistics, tourism and as a culinary chef
As Watson's team point out, even the best chefs build flavour profiles with only two or three ingredients at a time, a time consuming process - whereas Chef Watson can sift through millions of ingredient combinations and consider countless options simultaneously. As a cognitive system, it learns from human expertise and extends what people can do on their own. Coupled with Bon Appétit's recipes
, Chef Watson reads data from thousands of recipes to understand how different ingredients are used in cuisines and dishes, as well as having an added understanding of food chemistry and the psychology of people’s likes and dislikes. With as little as one user-selected ingredient, Chef Watson can suggest a totally unique flavour profile, measurements, and preparation steps for a dish.
, Lead Engineer for IBM Chef Watson, will be speaking at the RE•WORK Future of Food Summit
in London on 21 June. I caught up with him to hear more about his work and what the future holds for foodtech.What was your motivation for applying IBM Watson's intelligence to food?
After Watson won at the Jeopardy! TV show in 2011, IBM worked on turning what was then a supercomputer into a panoply of services available on the cloud to let entrepreneurs develop their own cognitive apps. The IBM BlueMix platform now offers services for natural language processing, machine learning, voice recognition, image recognition, and data insights. One aspect of cognitive computing that was missing to Watson was creativity. Could Watson help its users be more creative? We thought that an application that would inspire cooks by suggesting novel combinations of ingredients would be a great way to answer that question.
What do you feel has been essential to the success of IBM Chef Watson so far?
First there's the universal appeal of food. The US has become a food-focused culture: food TV channels like the Food Network turned a public that largely didn’t care about cooking into foodies, and social web sites started a whole crowdsourcing movement, where people review restaurants, post stories and pictures of the food they make and the food they eat.
Then, as we all know here, food tech is trending, the number of food tech startups has exploded in recent years.
But I think there's also the fact that this is a very unexpected move from IBM. Who would have thought that IBM would run a food truck and publish a cookbook? This is definitely not your grandparents' company.
How can your technology and ethos be applied to other areas?
The algorithms that we created for Chef Watson are fairly generic. In a nutshell, they ingest a corpus of existing creations (recipes from Bon Appetit magazine), implement several scientific theories that help score combination of ingredients (such as the foodpairing method), and then try to combine ingredients in a way that is novel and scores high. There are many other domains that could use this approach, such as perfumery, fashion, interior decorating, product design...
What do you see in the future for IBM Chef Watson?
When it comes to cooking, I think of three broad categories of users: the makers, who create and cook recipes from scratch, the remixers, who modify existing recipe to satisfy their taste, and the consumers, who love to eat but prefer to have someone else do the cooking. These categories aren’t mutually exclusive, and the same person can be a maker, a remixer or a consumer on different occasions. Chef Watson today addresses mostly the makers, and to a smaller extent, the remixers. We recently launched a cocktail app called Watson Twist that focuses on the consumers, people who are in a bar and want to create new cocktails that a bartender will prepare for them. But I'd like to see more variations on Chef Watson that address the remixers and consumers better.
What new developments in foodtech can we expect to see in the next 5 years?
Within 5 years, we'll see applications that successfully tackle meal planning for the masses. The war has begun, and food geeks can choose between a myriad of companies offering boxed meals, giving meal recommendations, helping you cook with you have on hand, optimizing your dinner choices for your definition of "healthy", and so on. In the coming years, a handful of winning models and companies will emerge.
What are the most pressing issues in the food supply chain? How can these be solved using technological or scientific breakthroughs?
One major issue is food waste. About 1/3 of the food produced globally is wasted. Part of the problem can be alleviated with technological solutions such as Chef Watson, meal planning, or a kitchen filled with IoT gadgets.
But it's often too convenient to think that issues can simply be solved with technology. There needs to be a change in mentalities as well. Do American restaurants need to serve portions so large that nobody can finish them? Do supermarkets need to sell packages of 8 burger buns when the average US household size (and therefore the number of buns one should need for a single meal) is 2.5? In a world where I can get almost anything delivered to my door the next day (if not the same day), do I really need to stock on food "just in case"?
Another issue is quality and transparency, and I'll talk about it a bit more in the next question. Legislation is often too complaisant.
What are the top 3 things you would like to see happen in the food industry?
• Ability to verify the nature and provenance of the products in my plate using an affordable, portable device. Even just the nature of the products would be a great start. Recent studies have showed that a third of the seafood sold across the United States is incorrectly labeled, a third of tested restaurant lobster dishes actually contain cheaper seafood. And of course there was the recent horsemeat scandal in Europe. This is outrageous!
• Instant measurement of the flavor characteristics of a dish using an affordable, portable device. Something like a pocket gas chromatography mass spectrometry measuring device. I would use it to come up with better food / wine pairings, and reverse-engineer recipes in restaurants, tweak my own recipes...
• Eco-friendly agriculture that doesn't involve eating crickets or Soylent. I'd rather have meat less often than eat bland substitutes.
Florian Pinel will be speaking at the RE•WORK Future of Food Summit in London on 21 June 2016. Other speakers include Abi Glencross, PhD in Cellular Agriculture, King's College London; Nick Holzherr, Founder & CEO of Whisk.com; Richard Ballard, Co-Founder of Growing Underground; Ian Hales, Research Associate, Bristol Robotics Lab, and more.
Early Bird tickets are available until 29 April, for more information visit the event page here. Our previous foodtech event sold out, so book early to avoid disappointment!
Future of Food
Future of Food Summit