AI model aims to improve plant-based texture

The taste of plant-based alternatives often falls short. Texture in particular proves problematic for consumers, according to researchers at Wageningen University & Research. With a $50,000 grant from the Bezos Earth Fund, the team will now develop an AI model to help change that.

Texture shapes the eating experience

Researcher Jack Yang explains that texture is a key hurdle. “Consumers often find meat and cheese substitutes too dry or too rubbery,” he says. The project therefore focuses on the building blocks of food. Molecules form structural elements, and those elements determine the final texture. Artificial intelligence will be used to map which combinations work best.

New and open source

AI is already used to predict recipes, but a model that specifically targets texture — and is open source — does not yet exist. “That’s where Wageningen is fairly unique,” says Yang. “Many parties focus on what happens within a building block, but we also look at why. That gives us real traction.”

Building better plant-based food together

Once completed, the model will be shared with other players in food technology. “It really has to become a tool for the entire food tech community,” Yang emphasizes. This shared system can then grow with industry input, speeding up the development of more flavourful alternatives to meat and dairy.

The research is part of OLiMPuS, a collaboration between Food Sciences & Technology, Plant Science Group, and Wageningen Food & Biobased Research.

Wur.nl