Every bite of a Big Mac is a carefully crafted experience, encompassing fat, salt, texture, and a recipe recognized by millions. Now, a team from Stanford University has pitted an artificial intelligence (AI) against this fast-food icon.
In a San Francisco restaurant, 101 participants tasted various burgers without knowing which recipes came from the AI. Two AI-generated patties matched or even exceeded the Big Mac in flavor. Developed using BurgerAI, an AI model that analyzes ingredient patterns rather than guessing recipes, the system evaluated 2,216 burger recipes to create over a million new combinations aimed at delivering better taste, nutritional value, and environmental sustainability.
How the AI Burger Challenges the Big Mac
In a study published in npj Science of Food, researchers like Vahidullah Tac, Christopher D. Gardner, and Ellen Kuhl describe the burger as a benchmark for computer-aided food development. The Big Mac serves as a tough competitor, sold in over 100 countries and notorious for its recognizable taste.
Interestingly, BurgerAI did not have access to the official Big Mac recipe. Instead, it used publicly available recipes to discern which ingredients complement each other, typical proportions, and frequently occurring combinations. Kuhl explains the difference from other AI systems: “Most AI systems are trained to predict what already exists. We wanted AI to invent what should exist next.”
Participants scored the burgers on a scale of 1 to 7, assessing overall appeal, taste, and texture. The Delicious Burger 1 scored 5.8 points for taste, while the Big Mac scored 5.4. The Delicious Burger 2 achieved a popularity score of 5.7, beating the Big Mac’s 5.3 in taste as well. Texture showed no significant difference.
Unexpected Flavor Success from AI
Many guests described the AI burgers as meaty, juicy, or smoky, addressing a common failing among replacement products, which often sacrifice mouthfeel for other benefits. BurgerAI managed to create burgers that met familiar expectations.
However, as researchers focused on environmental sustainability and nutritional quality, results shifted. The most eco-friendly burger, made from mushrooms, scored an environmental value of 0.06, while the Big Mac scored 0.93, incorporating land use, greenhouse gases, water scarcity, and nutrient runoff into waterways.
On the plate, though, the pure mushroom burger rated lower in overall appeal, taste, and texture compared to the Big Mac. It was often described as earthy, soft, or strong-flavored. A second sustainable option combined beef with mushrooms, achieving similar taste quality to the Big Mac but lacking in environmental benefits.
Why the AI Burger Can’t Master Everything
In terms of nutritional value, BurgerAI produced a notably strong option with a bean-based burger scoring 63.12 on the Healthy Eating Index, while the Big Mac only managed 33.71. The bean burger included more vegetables, whole grains, and plant-based protein while faring better in refined grains, sodium, and saturated fats. Its environmental score improved by a factor of six.
Despite this, in blind tastings, it fell short. Guests rated the bean burger at 3.8 for popularity, 4.0 for taste, and 3.7 for texture, all below the Big Mac’s ratings. Many described it as dry, grainy, soft, or bland. Hence, good nutrition does not translate into a satisfying bite.
“We expected a compromise between sustainability and consumer acceptance,” Tac revealed. The surprising fact that a burger with significantly better environmental impact could stand up against a globally successful fast-food product was a pivotal finding. Kuhl sees this as more than just a new recipe: “The burger is just the beginning.”
Stanford Sees More Than Just a Recipe
BurgerAI has clear limitations. The model understands ingredients and quantities but doesn’t encompass full cooking processes. Cooking times, roasting aromas, moisture loss, and culinary techniques need to be added by a professional chef. Moreover, the training data limits results; it primarily reflects Western burger traditions found in online recipes.
The environmental values also rely on databases and averages. Specific supply chains, regional farming methods, or individual production conditions may vary. Additionally, the taste test remains limited: only 101 people sampled a few burgers at a San Francisco restaurant. Larger tests are needed for reliable insights across broader consumer groups.
Nonetheless, this initiative holds significance for the food industry. AI can do more than merely replicate recipes; it can combine ingredients to balance taste, nutritional value, and environmental concerns quantitatively. The perfect burger may not have emerged, but several AI burgers came close enough to challenge the fast-food classic seriously.
Summary
- BurgerAI from Stanford University created a million new burger variants from 2,216 recipes, going head-to-head with the Big Mac in blind taste tests.
- Two AI burgers sometimes scored better in flavor than the fast-food classic, with no significant difference in texture.
- The study emphasizes that while AI can optimize taste, nutrition, and environmental impact, the ideal burger doesn’t emerge across all three categories simultaneously.
Interestingly: While BurgerAI reimagines fast-food classics, another alternative is already emerging—made from leftover carrots that would typically go to waste. A fungus transforms them into protein-rich patties and sausages that, in tests, outperformed soy in taste—stay tuned for more details in our upcoming article.
Image: © Jill Sakata Melosh
