Meat-Eaters Still Distrust Alternative Protein. New Technology Could Pave the Way

This past January, Israel set a precedent when it became the first country in the world to approve the sale of cultivated beef (the United States and Singapore had already approved the marketing of cultivated chicken). Cultivated meat, grown from animal cells under laboratory conditions, is part of a bustling corner of the food industry — alternative protein — in which technology, science and entrepreneurship join forces to produce substitutes for meat, milk, eggs and fish.

Why does it even matter? Beyond the familiar moral dimension, and beyond the contribution industrial farming makes to problems like antibiotic resistance and pandemics, the greatest threat that the animal-based food industry poses to humanity comes by way of the climate crisis: the steak, the omelet and the cottage cheese on our plate carry an enormous environmental price. This industry is responsible for a third of all greenhouse gases — more than any other single source.

Add to that the economic potential, and it is no wonder that many countries are investing in alternative protein initiatives. Israel holds a place of honor on that list: total investment in Israeli startups in the field is among the highest in the world, second only to the United States. In Israel, alternative protein matters from a food-security angle as well. An active, productive industry would reduce the country's dependence on food imports and supply chains — something that could prove critical in times of war.

To advance this vision, the state is joining forces with the private sector and backing a range of initiatives to develop the field. Among the standout examples are two startup incubators — Fresh Start in Kiryat Shmona and The Kitchen Hub in Rehovot — which have already brought into the world several well-known companies in alternative protein and food tech.

But despite improved products that earn excellent reviews, like the Beyond Meat burger or Alpro's "This Is Not M*lk," regular consumption of alternative protein products remains mostly the province of vegans and vegetarians. Meat eaters, including those troubled by the climate crisis and by what goes on in the farms, rarely diversify their shopping lists.

There are two reasons for this, studies find. First, the taste and texture of the substitutes still fail to precisely mimic food taken straight from the animal. And second, the price. The new industry still operates on a small scale, with heavy upfront investment in setup and development, and it cannot compete with the prices of traditional food plants. This creates no small problem: to truly change the course of the crisis on our doorstep, human society needs a real and meaningful shift in its eating habits. Without broad adoption of plant-based or alternative proteins by most meat eaters, we will not manage to stop — or even slow — our gallop toward a world that cannot be lived in.

This challenge has not escaped the alternative protein companies, which are working energetically to improve their products while cutting production costs. And alongside the tools and methods of food engineering, another player has taken the field: artificial intelligence (AI). While traditional food-engineering technologies rely on a fundamental understanding of food components and how they react to different conditions, AI approaches the problem from another direction: by drawing on existing data about those same components and food types, together with methodologies and algorithms honed over decades, AI methods can identify patterns the human eye does not catch.

The American company Shiru, for instance, mines a public database of protein structures to find plant-based substitutes for animal-derived ingredients — a process that usually takes years of work, and that has now been cut to a mere few months. Climax Foods, a maker of plant-based cheeses that also star on Michelin restaurant menus, uses AI as an integral part of its research and development. Last year it managed to crack one of the field's great challenges, where many had failed: a cheese that melts and stretches much like the animal-based version.

Israel, too, is not lagging behind. ImaginDairy, which produces a protein originating in cow's milk through a process that does not require a single cow, used AI to develop an especially efficient, high-yielding production method. The protein is grown inside a specially engineered strain of fungus developed in-house. It was the AI models that guided the development process and sped the arrival at a fungal strain that functions as an efficient "production line." Eggmented Reality, for its part, enlists AI to scan the vast diversity of molecules found in nature and to pick out those capable of mimicking specific components of animal-based food. For its egg substitute, which the company produces for Tnuva and other food manufacturers, it needed a material that can be whipped, that can "bind" the other ingredients in a recipe — in other words, one that behaves like an egg. Using AI lets the company's people quickly identify promising materials from an almost infinite list.

The use of AI in this field holds enormous potential. The great challenge lies at the very foundation on which every modern AI technology rests: data. Every machine-learning model needs data as a starting point. But whereas language models — like the beloved, familiar ChatGPT — can draw on the nearly infinite amount of text on the internet, generous datasets on how plant-based components behave under different cooking conditions are nowhere to be found.

To solve this problem, academic research groups are working to produce public datasets on cultivated meat and fish cells, meant to serve the entire industry. Alongside them, nonprofits and organizations that have spotted the hidden potential are working to consolidate existing data so that it is accessible to all. From another direction, there are also commercial ventures that may, intentionally or not, contribute to the research effort — for example, the neural network Google unveiled a few months ago, capable of predicting the smell a given substance will emit based on its molecular structure. Because of the tight link between smell and taste, this technological advance carries a promise of future models that could predict taste itself. Such a model would accelerate the development of new products to an inconceivable degree: instead of a long process of trial and error, food scientists could run computer simulations and take to the kitchen only the winning recipes.

At a time when there is so much talk of the danger AI models pose to humanity, could a different application of that very technology actually improve our odds of survival? Will pairing advanced AI methods with traditional food engineering crack the perfect formula for tasty, affordable substitutes — and send meat consumers streaming toward a more ecological, healthy and ethical diet? It will not happen overnight — it will take sustained investment from the private market alongside public policy, as has already begun to happen on the ground. One can hope that, with the rapid advances in artificial intelligence, we will see the alternative protein field, too, keep improving and growing more efficient — until the day when the right choice at the supermarket shelf is also the easy one.