Over the past few weeks, GPT-4 has proven to be a helpful tool in the kitchen. It offers suggestions for ingredient substitutes and can even generate recipes based on the items you have on hand. This AI model, known as GPT, is both polite and knowledgeable. It can provide information on local delicacies, imitate the Cooking style of famous chefs, and offer simplified versions of recipes.
During a recent trip to Venice, GPT informed me about the local specialty fish, the branzino or European sea bass. It can even replicate the artistic style of renowned chefs like Yotam Ottolenghi. For example, when I requested a recipe for “beans on toast in the style of Yotam Ottolenghi,” GPT provided a recipe for “spiced beans on sourdough toast” which included various spices and Greek yogurt.
However, GPT’s results may not always be revolutionary. When compared to the recipes of YouTuber chef Joshua Weissman, GPT’s formulaic recipes for burgers, fried chicken, and chocolate-chip cookies fell short. Nonetheless, AI has the potential to help us overcome such limitations and improve our cooking skills.
Of course, there are some issues with AI models. For instance, the output generated by GPT may show bias towards Western cuisine. There have also been instances where AI chatbots designed for eating disorders have provided inappropriate advice. Furthermore, without skilled prompting, AI-generated recipes may turn out to be bland or even include ingredients that someone is allergic to.
Nevertheless, the convergence of cooking and computation makes sense, as recipes are essentially algorithms. They are sets of instructions designed to solve the problem of what to cook and how to cook it. Finnish engineer Sami Matilainen has even visualized Nordic classic recipes using flowcharts, emphasizing the complexity of food preparation.
The connection between cooking and programming goes further with the esoteric programming language known as Chef, which uses cooking terminology to write computer programs. This highlights the similarities between recipes and algorithms.
Instead of relying on GPT to provide variations of dishes I already knew, I decided to experiment with fusion cuisine. Combining my favorite cuisines of Iranian and Sichuan Chinese, I asked GPT for a fusion recipe. The first suggestion, “kung pao jeweled rice,” seemed uninspiring. However, when I asked again, GPT recommended “mapo bademjan,” a combination of spicy mapo tofu and hearty khoresh bademjan. The resulting dish resembled a lasagna with layers of Chinese sauce and fried aubergines from the Iranian khoresh. It had a unique flavor and texture that exceeded my expectations.
AI has the potential to capture and record culinary experiments, family recipes, and unconventional cooking techniques. It offers an opportunity to preserve and enhance the culinary traditions that may otherwise be forgotten. Written recipes have been around for thousands of years, but AI brings back the oral tradition of cooking while adding new dimensions to the experience. AI can provide more personalized and nuanced responses to food, reintroducing adjectives and sensory details that go beyond traditional recipe formats.
In the future, AI-powered cookbooks could constantly learn and refresh, providing personalized cooking experiences. This would redefine what it means to cook and create a kitchen environment that listens and learns from cooks’ tweaks, fusions, and improvisations. The scope of culinary experience would expand, allowing everyone to access and contribute to a growing collective knowledge of cooking.
The post The Role of AI in Cooking and Recipe Creation appeared first on TS2 SPACE.
