Putting AI on the Menu

"I'll have the usual." 

You can hear that line in dozens of movies and it's a clear indication that a lead character spends a lot of time at a particular diner. Of course the long-time waitress (and it's almost always a waitress) knows exactly how the main character likes his eggs (and yes, it's often a "him").  

What does this have to do with AI? Randi Zuckerberg, president of Zuckerberg Media, recently launched a pop-up experience to help kids better understand science through food. Called Sue's Test Kitchen it had some very interesting experiments, like serving 3D printed pancakes or freezing with liquid nitrogen, but Andrew Brust found a much more interesting take on it: the use of AI to get to know you, the person eating the food.  

The idea, as he describes it, is to combine CRM data with AI to learn a customer's preferences with information like food allergies to create customized menus. So imagine going into a diner and being handed a menu that matches your preferences with the fresh ingredients in the kitchen that day. Your menu may never show you the tuna sandwich that your companion orders.  

Going one step further, AI can play a role in helping the restaurant predict demand for certain foods, place more accurate orders, and adjust for timing.  

It's an intriguing idea that takes AI into the mainstream in a unique way. Right now if you ask people about artificial intelligence in food, they will envision a world in which robots flip burgers or even make a cappuccino. 

But that thinking doesn't get to the heart of how AI can truly change lives. By knowing your dietary preferences and sharing that information with the kitchen, the kitchen staff can be ready to offer you a dairy-free, gluten-free meal, even if the wait staff has never seen you before.  

Some of this is already at work. Starbucks has a Digital Flywheel program that is starting to bring AI into its insanely popular mobile app. The more you order a grande decaf soy latte with 2 pumps of vanilla on your phone while you're walking to the subway, the more the system knows about what you like and when you like it. It's easy to see a point at which the system can predict at what time of day you order, where you are at the time of your order, and the weather on the days you switch from hot to iced. And as your behavior and preferences change over time, machine learning can be employed to make sure your order is always right. 

Machine learning plays a similar role in Attivio's cognitive search and insights platform. The system supports multiple machine learning relevancy models that learn continuously over time, so when employees perform a search, they are benefitting from their own past behavior as well as from employees like them. 

Which means soon you won't need to say "I'll have the usual" - for your coffee or your corporate information - it will just show up when you want it.