On a recent Swiss International flight to the United States, I was remembering the book “The Future Computed,” by Brad Smith and Harry Shum. They tell the story of Melisha Ghimere, 20-year-old computer science student who leverages AI in Nepal to help farmers monitor the health and well-being of their livestock and to prevent outbreaks of diseases. You can find my book recommendation here.
I was fortunate enough to be flying in business class. As I read, the flight attendants moved through the cabin asking passengers to choose their meal from a menu with three or four options per course. They noted down each passenger’s preference on a piece of paper.
It shouldn’t come as a surprise that they have a limited availability of each meal. Clearly, they would not carry a full selection for each individual: even if they had the physical space in the cabin, it would be irresponsibly wasteful. On this particular flight, the beef fillet was the most popular option. I heard passenger after passenger request it, only to be told there were no more left.
You might be wondering where I’m heading with this. Here’s what shocked me: how come a girl in Nepal has gone further in the practical application of AI than a major airline carrier?
It should be easy for the airline to build a model that allows them to predict the number of each choice of meal that will be ordered. Such a model would use data from individual customers’ past choices to predict what they are likely to order: a particular frequent flyer, for example, might always order beef when it is available, otherwise chicken, and never fish. The model would create profiles enabling it to predict what a first-time business class passenger will eat.
How do I know airlines don’t do this already? They do, after all, certainly use AI for other things, such as ticket pricing. One clue is that they don’t even appear to be collecting the data, at least on Swiss: flight attendants collect passengers’ choices on pencil and paper, rather than on an electronic device. Easier still, passengers could select their meals using their entertainment screens.
Internet forum discussions suggest this isn’t a problem unique to Swiss – and that many passengers are annoyed by it. Some airlines are now allowing you to guarantee your meal choice by ordering online in advance. That makes sense, but AI models could still help predict what extra meals should be carried for passengers who didn’t do so.
My point isn’t to complain about a first world problem. Obviously, a Nepalese farmer worried about her livestock’s health is more deserving of sympathy than a business class flyer disappointed not to get a beef fillet. My point is that it’s surprising how airlines seem to be willing to upset a profitable customer segment when it would be trivially easy to avoid it.
AI is now extremely low cost. A model to predict demand for business class meals could be adopted quickly and without any major investment. It could be done using Microsoft’s cloud platform Azure. That’s being used for everything from appliance makers forecasting demand for particular spare parts to elevator companies predicting when maintenance will be needed.
Perhaps your business is also overlooking a way to use AI to delight your customers? Companies such as Microsoft are eager to work with you on adopting innovative technologies – so what are you waiting for?