Artificial intelligence is no longer a novel concept in the realm of big food. Major players like McCormick and Unilever have been leveraging AI for years to accelerate product development and refine flavor profiles. McCormick, the company behind popular brands such as Frank’s RedHot and Old Bay, has reported a 20% to 25% reduction in development timelines by utilizing AI to identify promising flavor combinations and prioritize physical prototyping.
Similarly, Unilever has integrated AI deeply into its food research and development processes. Their systems can digitally test thousands of recipes in seconds, significantly reducing the need for physical trials and leading to viable concepts more rapidly. The development of Unilever’s Knorr Fast & Flavourful Paste, for instance, was reportedly completed in about half the typical time. Even in packaging innovation, AI has been employed to model formulation behavior, as seen with Hellmann’s Easy-Out squeeze bottle, saving months of laboratory work.
Beyond these established giants, the broader food industry is witnessing a surge of startups aiming to capitalize on AI’s potential. Companies like Zucca, Journey Foods, NielsenIQ, and AKA Foods are marketing “virtual sensory” platforms. These AI-powered systems promise to digitally screen recipes, suggest formulation adjustments, and predict consumer preferences before any physical prototypes are created. Their value proposition centers on reducing reliance on traditional taste panels, mitigating the risks of product launch failures, and compressing development cycles by identifying promising concepts earlier in the process. Industry analysts project a substantial growth in the global AI in food and beverages market, from an estimated $10 billion in 2025 to over $50 billion by 2030, fueled by escalating investments in data-driven product development, automation, and personalization.
However, the path for AI in food innovation is not without its challenges. Some early pioneers in the field have encountered limitations. For example, McCormick’s initial AI work was a collaboration with IBM, which has since shifted its focus away from AI-driven food projects. Food scientists who have tested these emerging platforms indicate that the technology is still in its nascent stages, and many of the ambitious claims may be more geared towards attracting investment than delivering a complete replacement for human expertise.
Brian Chau, a food scientist and founder of the consultancy Chau Time, observes that many AI food startups are still in the data-collection phase. They are working to aggregate sufficient real-world information to make their predictive models meaningful. “Most AI companies emerging today are, to some extent, overstating their capabilities – this is true for most startups,” Chau notes. “They need to attract investors, build datasets, and secure genuine industry partnerships before this technology can truly operate at scale.” He further explains that current platforms often resemble large language models trained on existing recipes, manufacturing data, and consumer trends, rather than systems capable of independently generating novel and viable products. “When I tested one platform, the output was essentially what you would get from any general AI system,” he recalls. “There wasn’t much added value without proprietary data from real companies.” Chau believes the long-term success of this technology hinges on startups’ ability to forge partnerships with major food manufacturers willing to share their internal formulation data, a prospect many companies are hesitant about due to intellectual property concerns. “Without major industry players feeding real data into these systems, it’s very difficult for them to become truly predictive,” Chau concludes. “It’s a numbers game.”
From a scientific perspective, the primary hurdle for AI in food development isn’t computational power, but rather the inherent complexities of biology and human perception. Dr. Julien Delarue, a professor of sensory and consumer science at the University of California, Davis, suggests that some expectations surrounding AI-driven sensory tools may stem from a misunderstanding of AI’s realistic modeling capabilities. “There is likely a bit of hype involved,” Delarue states. “This doesn’t mean AI isn’t useful; it’s just that its utility might not align with certain expectations.” While AI can undoubtedly aid in analyzing chemical data and enhancing efficiency in food development, predicting how humans will perceive complex flavors remains a significant challenge. “Attempting to predict what people will perceive from a complex mixture of compounds – the answer is no,” he asserts.
A fundamental obstacle is the inherent variability of human sensory perception. Individuals perceive the same chemical compounds differently, influenced by genetics, culture, personal experiences, and even individual history. “There is no such thing as the average consumer,” Delarue emphasizes. “Trying to predict what the ‘average’ person might perceive is likely a dead end.” Unlocking this complexity would necessitate access to significantly more granular data, specifically individual-level perception data, which is a monumental undertaking. This variability makes it difficult for any model, whether human or machine, to serve as a universal proxy for taste.
Even the companies developing these AI tools acknowledge that human judgment remains central to the process. David Sack, founder of AKA Foods, clarifies that his company’s platform is designed to organize internal R&D knowledge rather than replace food scientists or sensory experts. “Food R&D teams possess a wealth of valuable knowledge, from past formulations and sensory data to tacit know-how held by individuals,” Sack explains. “However, this information is often fragmented and difficult to leverage systematically.”
AKA’s platform aims to enable teams to test ideas digitally before committing to costly physical trials, allowing scientists to concentrate on the most promising formulation avenues. “It does not replace food scientists or sensory experts,” Sack reiterates. “Ultimately, humans define the goals, constraints, and success criteria. Sensory experts design and interpret panels. Scientists decide what to test and what to launch. AI can reduce the number of tests required, but it does not eliminate the need for real human tasting or validation. Humans must always remain involved when the end consumer is human.”
Jason Cohen, founder and CEO of Simulacra Data, a company specializing in AI analysis of sensory and consumer data, echoes this sentiment. “Consumers decide with their palate whether they like a product,” Cohen states. “We still begin with real human sensory data. AI simply helps us extrapolate insights faster and more affordably.” Cohen, who also founded Analytical Flavor Systems (acquired by NielsenIQ in 2025), believes AI’s primary value lies in identifying off-flavors, narrowing down formulation options, and prioritizing promising concepts for testing, rather than supplanting human perception.
Chau posits that large food corporations are uniquely positioned to benefit from AI-driven tools due to their existing control over extensive proprietary formulation, sensory, and manufacturing data – a resource that most smaller brands are still striving to build. Delarue foresees AI’s principal contribution to the food industry being in efficiency gains. He anticipates AI will assist researchers in faster data analysis, better complexity management, and operating within the increasing constraints related to health, sustainability, and cost. “Designing food today is considerably more challenging than before,” he remarks. “You aim not only to create food that people enjoy but also food that is healthy, sustainable, and affordable. AI empowers us to manage this complexity more effectively.”
However, when it comes to the nuanced domain of taste itself, humans remain the ultimate arbiters. “Consumers will always be the ones to decide what tastes good,” Delarue concludes. “Not machines.”
Original article, Author: Tobias. If you wish to reprint this article, please indicate the source:https://aicnbc.com/18524.html
