Every second, around 75 hamburgers are sold from just one of the best-known chains in the world.. That’s equivalent to more than 2.3 billion a year on that chain alone. If we add the rest of the restaurants, fast food chains and artisanal hamburger restaurants, the number rises to tens of billions of hamburgers consumed each year. In fact, there is talk of 200,000 per minute. What is clear is that few recipes have conquered the planet with such success.
Its origin, however, remains a matter of debate. Although Hamburg-style ground beef was already known in Germany during the 19th century, it was immigrants who brought the idea to the United States. There someone had an idea as simple as it was revolutionary: placing that steak between two pieces of bread so that you could eat it with your hand.. One of the gastronomic icons of the 20th century was born. But there is a problem. What makes a perfect burger?
For some the answer is in processed cheese. Others opt for the balance between meat and fat. There are those who prefer a vegetarian version, those who seek the lowest possible environmental impact or those who simply need a meal rich in protein. Spain has been winning the award for the best on the planet for 3 years, but for color tastes and designing a burger capable of satisfying all those criteria at the same time seems impossible. Or maybe not.
A team of scientists from Stanford University has just presented BurgerAI, a artificial intelligence capable of designing completely new burgers while simultaneously optimizing flavor, texture, nutritional value and sustainability. But the real objective of the project goes far beyond the kitchen.
Those responsible for the advance, led by Ellen Kuhl, estimate that there are approximately 10⁴³ possible hamburger recipes. It’s a one followed by forty-three zeros. To get an idea of that magnitude, if by the art of birlibirloque, all the inhabitants of the planet right now, had been condemned to make one of these hamburgers since the Big Bang (13.8 billion years ago), We still would not have completed even 0.01% of the total prescriptions. And all this without even wasting time trying them. The solution? An AI.
“Most current artificial intelligence systems learn to predict what already exists – says Kuhl, professor of Mechanical Engineering at Stanford and director of Stanford Bio-X -. We wanted AI to invent what does not yet exist.” The difference seems subtle, but it completely changes the focus. Rather ask which is the most likely burger, BurgerAI tries to answer another, much more complex question: Which recipe offers the best possible balance between taste, health and sustainability?
To achieve this, Kuhl’s team trained their system using 2,216 recipes taken from Food.com. The AI learned what ingredients are usually combined, in what quantities they appear, and how they influence flavor or texture. Then he began creating completely new recipes, optimized according to each person’s profile: age, gender, physical activity and even environmental goals.
But Kuhl knew that algorithms can look brilliant on paper and fail miserably when it comes time to eat. So they decided to bring hamburgers to real life. Five recipes designed by BurgerAI were professionally prepared and served during a blind tasting in a restaurant in San Francisco, where more than one hundred diners participated. The result surprised even the scientists themselves.
Two of the burgers created by the AI scored equal to or higher than a popular fast food burger in taste, texture and overall acceptance. The so-called Mushroom Burger also reduced the environmental impact by more than an order of magnitude, while The Bean Burger approximately doubled the nutritional score of the commercial model used as a reference. These conclusions have been published in an article in Science.
“Artificial intelligence not only generated plausible recipes – adds Kuhl -. He created burgers that people really liked”. At first glance, all this may seem like a cute experiment. However, those responsible for this test insist that the burger is simply a demonstration.
“Food was a good starting point because with a single arrow we could hit two targets: the health of the planet and the health of people,” confirms Vahidullah Tac, co-author of the study.
The real interest lies in the mathematics behind it and this is the focus of the second study published by Kuhl’s team. Designing a burger involves finding the best possible balance between multiple, often conflicting factors: taste, cost, nutrition and sustainability. Curiously, This same type of problem appears in many other scientific fields. For example, a medication must be effective without being toxic.. A new material must be strong, light and cheap, and a synthetic protein must simultaneously meet numerous structural conditions.
Until now, many of these processes depended on years of trial and error. Kuhl believes that systems like BurgerAI could greatly reduce that search space, proposing entirely new designs before they are even first manufactured. Deep down, The project is not trying to prove that a machine can create a perfect hamburger, but rather that an AI can become a tool capable of designing unprecedented solutions for extraordinarily complex problems.
“The hamburger is just the beginning – concludes Kuhl -. We see food as a model system for a much larger vision: an artificial intelligence that acts as a partner in scientific discovery and engineering.”