Can artificial intelligence make a pizza richer than a ‘pizzaiolo’?
For some time now, artificial intelligence (AI) is everywhere and all industries are curious how AI systems can improve their sector. Gastronomy is also beginning to wonder how far the machine is capable of reaching. Proof of this is the fight organized on Thursday night by Can Pizza, in Barcelona. Known for its artisan pizzas with top-quality products —with 10 locations throughout Spain—, this pizzeria wanted to challenge the ChatGPT system, asking it to make a pizza recipe with ingredients that Catalan diners might like. At an event titled Pizza Kombat, the pizza master Lolo Lorenzo and the innovation consultant Matthias Schneider gambled everything with a new recipe for the pizzeria. The event will be repeated on February 9 at Can Pizza on Calle Serrano in Madrid.
The idea for the new fight – it is not the first one organized in this restaurant – arose during a conversation between Isaac Aliaga, co-founder of Can Pizza, and Schneider, an AI specialist. So they wondered how far artificial intelligence could successfully get into a kitchen. With this curiosity, they challenged themselves for a new pizza on the menu to emerge from a confrontation between an AI chat, which would be managed by Schneider, and a proposal from the team at pizzaioli led by Lolo Lorenzo.
To prepare the AI team’s recipe, data on ingredients, tastes, age groups, consumer trends or best-selling pizzas were taken into account. After several questions to define a recipe that Barcelona diners would like, Schneider’s chat proposed a pizza baked with mozzarella, shiitake mushrooms, sobrassada and Parmesan, with a final touch of goat cheese, pine nuts, honey and coriander pesto. and lime.
Lorenzo’s proposal wanted to go to the roots that have made Can Pizza’s pizzas famous —his Jerry Tomato is considered the best pizza of 2022 according to the Madrid Fusión contest—, and he prepared a recipe based on ragù, onion and pickled carrots , fresh celery and Parmesan cheese in thick slices. Alessio Geromel, director of Can Pizza Universitat, said that he is not “in favor of machines in the kitchen”, that this event was only to see “how it can affect” the restoration.
After trying the two pizzas, the diners who attended the event cast their votes and the winner was Can Pizza. Although the majority recognized that the recipe made by IA was rich and with a surprising combination of ingredients, which included sobrassada as a nod to the local public, the commitment to ragout and house pickles received the majority of votes. The attendees were also able to make their requests to the machine at the moment, with the result of personalized pizzas according to the suggested flavors or ingredients. In networks, the pizzeria has shared a video where the AI recipe competes with its Jerry Tomato recipe and Mikel Iturriaga, creator of El Comidista, does not hesitate to opt for Can Pizza again.
It’s still early for an AI cook
“In gastronomy we are beginning to see various applications of AI,” says Schneider, who acknowledges that at the moment they are based more on issues of the periphery than of the kitchen itself. In other words, it is still too early for a machine to propose you a good new recipe, but its abilities can be used when writing a letter in an original way or making visually attractive dishes using an image AI system. Of course, perhaps the suggestion is very nice but impossible to put into practice, he acknowledges. This consultant says that we are at the beginning of the possibilities of artificial intelligence, but “it can help inspire and conceive new gastronomic proposals.”
Within the creative industries, perhaps in the gastronomic sector, it is where AI still has a difficult time giving profitable results, especially in the preparation of recipes because there are too many elements at stake to create a good dish: smells, flavors, textures, points cooking, combinations… But he does see more possibilities in that it has a utility for the management of catering businesses. For example, the machine is very efficient at suggesting alternatives to an ingredient. It is able to propose up to five substitutes for butter such as olive oil, coconut oil, avocado puree, Greek yogurt and margarine. In addition, in each of these proposals, he points out the characteristics that are most similar to butter. “Sometimes it surprises you”, reflects Schneider, “sometimes she is naive and can even be a liar, but she can always be of help and inspiration”.