The Human Taste Test: Why AI Still Can't Replace the Chef

For major food corporations, artificial intelligence has been a quiet kitchen assistant for years. At McCormick, home to Frank's RedHot and Old Bay, algorithms have trimmed flavor development timelines by a quarter. Unilever uses similar systems to digitally test thousands of recipes in seconds, a process that helped develop a Knorr product in half the usual time. Even packaging, like Hellmann's squeeze bottle, has been optimized through digital modeling that saved months of lab work.

Despite this integration, executives are adamant: AI is not the head chef. "Human creativity and judgment lead the way," says Annemarie Elberse of Unilever. McCormick's chief science officer, Anju Rao, calls it a co-creation tool, stating their greatest asset remains their people.

This hasn't stopped a wave of startups—Zucca, Journey Foods, AKA Foods—from pitching their own "virtual sensory" platforms. They promise to predict consumer liking and screen recipes digitally, aiming to shrink development cycles and reduce failed launches. The market for such technology is projected to swell from $10 billion to over $50 billion by 2030.

Yet skepticism runs deep. Early pioneers like IBM have stepped back. Food scientist Brian Chau notes many startups are still gathering data, their models often little more than sophisticated recipe aggregators. "They need to attract investors, they need to build datasets," he says, arguing real predictive power requires proprietary data from large manufacturers, who are often reluctant to share.

The fundamental hurdle isn't processing power, but biology. Dr. Julien Delarue of UC Davis explains that human taste perception is wildly variable, shaped by genetics, culture, and experience. "There is no such thing as the average consumer," he states, making a universal taste proxy impossible. AI's near-term role, he argues, is in managing complexity—balancing health, sustainability, and cost—not inventing flavor.

Founders of these AI tools agree. David Sack of AKA Foods says his platform organizes internal knowledge but doesn't replace scientists. Jason Cohen of Simulacra Data notes AI helps extrapolate insights faster, but they always start with real human sensory data.

The consensus is clear: while AI can streamline the journey from concept to kitchen, the final verdict on taste will always belong to people.

Source: CNBC