Could AI Make Plant-Based Meats Taste Like The Real Thing?

Stanford engineers use mechanical testing and AI to improve plant-based meat textures, potentially accelerating development of realistic alternatives.

Led by Professor Ellen Kuhl, the team combined mechanical testing and machine learning to measure precisely food texture with AI, potentially accelerating the creation of more realistic plant-based products.

Published in npj Science of Food, the study demonstrated that machine learning could replicate the sensory experiences of human taste testers, marking a significant step in plant-based food development.

The researchers tested various animal and plant-based meats, including hot dogs, sausages, and turkey, alongside tofu. They found that some plant-based products already closely mimic the texture of animal meats.

“We were surprised to find that today’s plant-based products can reproduce the whole texture spectrum of animal meats,” said Kuhl, as reported by Phys Org.

The ability to replicate these textures is critical as plant-based foods are often perceived as lacking the bite or chewiness of real meat, a barrier for many consumers.

Stanford’s approach is grounded in mechanical engineering. The researchers used a 3D food texture testing method, where they applied pulling, pushing, and shearing forces to samples of meat and tofu.

These tests simulate the forces exerted when chewing. The data from these tests was then processed through a machine learning model, which created equations to describe the physical properties of the foods.

When the team compared the mechanical results with human sensory rankings, they found striking consistency. For example, plant-based hot dogs and sausages performed similarly to their animal counterparts in mechanical tests, with human testers ranking them closely in terms of stiffness and chewiness.

The implications of these findings could be far-reaching.

“Instead of using a trial-and-error approach to improve the texture of plant-based meat, we could envision using generative artificial intelligence to scientifically generate recipes for plant-based meat products with precisely desired properties,” said Skyler St. Pierre, lead author of the study, as reported by Phys Org.

By sharing their testing data online, the team hopes to encourage collaboration and accelerate innovation in the plant-based food industry, noted Phys Org.

The research team continues to expand their database of food texture data, including plans to test new products like veggie deli slices and fungi-based meats, noted Phys Org.

With these efforts, they aim to create a more standardized and data-driven approach to developing plant-based alternatives that may one day satisfy even the most committed meat lovers.