Have you ever used artificial intelligence to make pizza? A group of MIT students tried it, and they are quite pleased with how it came out. The pizza included the AI-made “wale[sic] walnut ranch dressing” as a topping.
In the sea of reports hinting at the dangers of artificial intelligence, the Massachusetts Institute of Technology group found a way to show that robots and humans could have a bright, peaceful future together. On Monday, MIT student Pinar Yanardag and her colleagues launched a new project called “How to Generate (Almost Anything.) Each week they will release a new product, such as art, perfume or food, that was created through the work of MIT students and their AI-powered robot.
“By augmenting human capabilities and pushing the boundaries of creativity, can AI inspire us to create things that wouldn’t have existed otherwise? A dress designed with a crazy hat, a pizza made with shrimp & jam or a scent that has never been smelled before?” the team wrote in a Medium post about the project.
In the first chapter of their project, the MIT students wanted to teach their AI to make pizza. To do this, the robot’s neural network had to process hundreds of artisan pizza recipes and come up with new topping combinations that would go nicely together.
“In general, AI models are very good at connecting different pieces of information together – that’s why there is usually a surprise factor in anything that an AI generates,” Yanardag said. “In our pizza experiment, we saw something similar where AI combined ingredients like shrimp and Italian sausage with jam, which it picked up from a dessert pizza.”
Since the students’ algorithm is still in the process of learning, some recipes looked more odd than others, and some missed important ingredients like sauce or cheese. Some pizza recipes even included ingredients that AI made up itself, like “snipped caramel cheese” and “wale walnut ranch dressing,” according to Medium post about the project.
The students also worked with Tony Naser, owner and chef at Crush Pizza in Boston. He was the human artisan who was tweaking the recipes and baking the pizzas. The project resulted in some silly pizza recipes like blueberry, spinach, and feta; bacon, avocado, and peach; and even shrimp, jam, and Italian sausage. The recipes can be seen on Medium.com. Interestingly, the shrimp, jam and Italian sausage pizza turned out to bey the team’s favorite.
“We didn’t really see it coming but it was definitely the BEST pizza of all,” they said in the Medium post. “After the tasting, Tony also decided to put some arugula on top which really elevated the taste as well (human-AI collaboration, yay!).”
It was so good, in fact, that Naser is considering putting it on the menu at his restaurant.
The project aims to show that humans can work with AI-powered robots in the future and that it’s not as threatening and dangerous as it seems to be based on every new leap AI takes. The pizza project hints at a future in which artificial intelligence can contribute to a better life for us.
“Everyday, we started to see more news about how machines are becoming better at human jobs, which makes people worried that AI will ultimately lead to mass unemployment by replacing them,” Pinar Yanardag, the creator of the project, told CNBC Make It. “In this project, we playfully wanted to show to the public that we don’t have to fear from the AI, but rather, we can work together with AI to achieve the most creative and productive outcomes,” said Yanardag.
This is not the first time MIT students have gotten creative with science and technology. A month ago, students from MIT cracked a long-lasting theory that it’s impossible to break a piece of uncooked spaghetti in only two pieces. The challenge couldn’t be solved even by the great mind and physicist Richard Feynman, but the MIT students built a machine that put that theory to rest.
Do you think the MIT students did a good job at teaching AI to make pizza? Let us know which is your favorite recipe.