CHICAGO — Artificial intelligence (AI) offers great potential for food manufacturers across the value chain, including predicting consumer behavior, reducing experimentation cycles when making formula changes and making production more efficient.
At the Institute of Food Technologists (IFT) FIRST 2024, held July 15-17 in Chicago, Eric Krums, senior solutions consultant at Infor, Sandeep Anand, senior director and head of machine learning solutions at Infor, and Michael Waters, senior vice president and chief information officer at Lewis Food Products, Denison, Texas, spoke about how AI machine learning and product lifecycle management provide opportunities to optimize product development and maximize return on investment. The panel also shared the challenges and first steps food manufacturers should take to be successful with AI.
AI machine learning has great potential to help food manufacturers across the value chain, Anand said. It can also accelerate new product development and ingredient substitutions based on price and availability. It can also help them respond to consumer behavior.
“Machine learning allows us to take data and create predictive models,” he said. “With AI, we can look at not only ingredients, recipes, and customer demand, but also the relationship to manufacturing.”
Food manufacturers are unable to leverage technologies such as AI machine learning. Clean data is essential to build reliable predictive models that AI can leverage. Product Lifecycle Management (PLM) systems help manufacturers manage formulations and comply with regulations. They can also interact with suppliers and integrate with existing ERP systems within food manufacturers' facilities.
“This is the foundational piece that everyone needs to understand before they can even start thinking about AI,” Krumbs explains. “This is where you can store your data, label it correctly, make sure it complies with regulations, and once you're set up, you can start playing with fun tools like AI.”
Incorporating many of these foundational tools for data collection and management has allowed Ruiz Foods to align research and development with manufacturing. This integration is key to speeding up the process and positioning Ruiz Foods for success in its AI machine learning experiments. However, Warter noted there are some challenges that must be addressed with AI in the food system.
First, help people understand what AI is and how it can be used. AI Machine learning is just one type of AI, the other two are deep learning and generative AI. Machine learning makes predictions based on models and data, whereas generative AI can create information that is not true and must always be verified.
The second challenge is that AI is not an IT-only project.
“Driving change with AI starts with a strong understanding of the process and the data it generates, and then building effective models,” Waters explains.
This means involving operations, R&D, and quality assurance teams.
The third challenge Water cited is the lack of regulation and security surrounding generative AI.
“I understand that's a possibility, but how do we ensure that what we put in there is safe for us?” he said.
Krumbs agreed, saying, “How do you keep your secret sauce a secret?”
AI machine learning holds great potential for improving operational efficiencies across bakery and snacks, but it requires a foundation of clean and robust datasets to build effective models for machine learning to work.
“The value is being able to use AI across the ecosystem and make data-driven decisions that can be leveraged to build an interconnected ecosystem,” Anand concluded.
