Using AI To Give Meat Alternatives A More Tempting Texture

By Sara Miller, NoCamels January 09, 2024 4 minutes

Illustrative: Rib-eye steak produced by meat alternative company Redefine Meat (Courtesy)

Foodtech companies may have honed the taste and the look of meat alternatives, but they have yet to perfect the texture or production costs, research shows.

Presented with this pair of pressing issues, Israeli artificial intelligence expert Noa Weiss created the GreenProtein AI non-profit, which analyzes data gathered from multiple food firms to improve the production process for meat alternatives, known as extrusion.

The aim is to create something that feels more like the real thing in your mouth while reducing the cost of a notoriously expensive process.

Chunk Foods’ plant-based steak has the appearance of meat (Courtesy)

Current polling of people who would like to switch away from meat shows that these are two of the most bothersome aspects of a vegetarian or even vegan diet, Weiss tells NoCamels.

“With taste, we’re actually getting a lot closer,” she says. “But texture is still a bottleneck and it is something that consumers want… They say that the texture today isn’t good enough.”

The extrusion process involves heating a range of protein sources continuously until their structure breaks down and the molecules reform into something akin to the sinuous qualities of meat, that is also tweaked for taste.

The GreenProtein AI project uses machine learning to examine all of the data surrounding the extrusion process in order to fine tune the texture of the meat substitute.

Weiss breaks it down into two separate sets of data: the process itself, including the ingredients used, the rate at which they are heated to break down the protein structures and the cooling dyes used at the end, and the resulting texture of the final product.

Noa Weiss: Society needs to significantly shift to food not based on factory farming (Courtesy)

All of the data is fed into the ML model, which is taught how the different combinations lead to different outcomes. And with that, says Weiss, the algorithm can determine the optimal extrusion process to create a meat substitute that is as close as possible to the real thing in texture as well as taste and appearance.

“What machine learning algorithms are able to do is learn patterns that humans aren’t always able to recognize,” Weiss says.

“Especially when you feed it a lot of data, it can really spot the nuances and then predict things that you maybe haven’t tried before.”

And it is this breadth of data that sets the GreenProtein project apart from other similar attempts to use artificial intelligence to improve what is colloquially known as “mouth feel” in meat alternatives.

GreenProtein AI is collaborating on the project with international sustainability non-profit Food System Innovations. The data comes from a range of food companies that use different techniques to produce the proteins used in meat substitutes but who are all nonetheless reliant on extrusion.

Sign up for our free weekly newsletter

The latter group, Weiss says, are the targeted beneficiaries of the project as they struggle to create a substitute that is truly convincing for consumers.

“We don’t really know how to make the texture that we want with proteins and extrusion,” she says.

“But it is still the best technology that we have for texture, and the only one that has the manufacturing capacity that we’re going to need if the demand is going to be [growing] as predicted in the coming years.”

Israel already has a strong market for meat substitute products, with around 13 percent of the population embracing a vegetarian lifestyle and around five percent identifying as vegan.

A meat and dairy-free diet is so popular among young Israelis that the Israel Defense Forces recently found that more than five percent of its soldiers described themselves as vegan, more than four times the global average.

Furthermore, the country’s alternative protein industry has attracted more than $1 billion in investment over the past two years or so – second only to the USA, according to the Good Food Institute Israel non-profit organization.

It is Weiss’ own commitment to a vegan diet and global sustainability that led her to put the project together.

“I looked for problems that are not just general problems that could be solved with AI, but specifically problems that are common to a lot of different players in the field,” she explains.

“I wanted to see if there was a common problem, something that if I could help solve it, would really boost the field. And that’s when I landed on extrusion.”

Illustrative: Plant-based minute steak (More Foods)

It is therefore, she says, a blend of professional expertise and personal philosophy.

“I’m a big believer in moving away from animal-based food systems, both for the ethical aspect and the sustainability aspect,” she says.

“Science says that if we want to be able to save our planet, we as a society need to have a significant shift in the next few years to food that isn’t based on factory farming.”