A food-safety expert explains why more recalls aren't necessarily a bad thing — and how AI plays a role

Willette M. Crawford has spent two decades working in food safety. She said AI is helping experts manage massive amounts of data. Courtesy of Willette Crawford; Getty Images AI is being used to improve foodborne outbreak detection and help manage recall decisions. Food safety consultant Willette Crawford said AI can help workers sift through vast datasets. High costs and non-uniform data remain challenges to AI adoption in food safety. For a brief period this summer, no fresh produce seemed safe. Lettuce, jalapenos, and blueberries were among a steady stream of recalls and food safety alerts that kept consumers on high alert. Willette M. Crawford, a food science consultant who has spent two decades researching predictive food safety, said AI has become a useful tool for enabling faster, more targeted approaches to outbreaks. She previously helped write regulations for the Food Safety Modernization Act, which shifted the FDA from reactive management of food contamination to a more proactive, preventive approach. "From a consumer perspective, people see all these recalls and outbreaks and think things are getting riskier or are deteriorating in some way. I don't think that's necessarily the case," Crawford, who is also the cofounder of the agriculture-tech company Katalyst Agricultural Solutions, told Business Insider. She added: "The food supply is not getting riskier — it's that our ability to see the risk is improving." Business Insider asked Crawford about exactly how AI is being used in the food-safety world and the ongoing challenges to its implementation. The interview has been edited for length and clarity. How is AI being used in food safety? Willette M. Crawford: AI is enabling us to connect information that has historically lived in databases or physical paperwork and to automate some of the workflows and processes. That allows food safety professionals to spend less time assembling and retrieving the information and more time interpreting it and doing whatever investigations are necessary, and focusing those limited resources where the risk actually exists. From a regulation standpoint, we're using whole-genome sequencing networks to improve outbreak detection. The AI tools are not detecting the pathogens, but they're helping us identify them more quickly in the system to make those connections across different systems, across different states, and so forth, so we can identify the recalls or the outbreak scope more quickly. How has AI changed your approach to the food-safety work you do? WC: What's changed is what's computationally possible: how early we can see a meaningful signal and how precisely we can decide where to focus. What AI adds is the ability to integrate far larger and more diverse datasets, find relationships nobody explicitly programmed in, and do it continuously, rather than as a research project. The bottleneck was never the math. It was always the data. How can companies use AI more effectively for food safety processes? WC: The companies doing this well all have one thing in common — and it isn't the sophistication of the AI tool or the model they're using. It's that they digitize their information first. The ones that have brought in AI tools but haven't digitized are still muddling through and trying to see the ROI of what they've done. Whereas companies that spend the time and investment to digitize — and it can be painful sometimes — those are the companies that are better positioned. In addition to your staff's everyday responsibilities, they now have to spend a lot of time working on the architecture of these systems, and that initial architecture is important to getting this right. If your records are incomplete, inconsistent, or not well-maintained as they already are, putting an AI system in place is not going to bring you the benefits that you expect. What are the big-picture barriers to adopting AI in the food-safety space? WC: Right now, it's still early adoption to some degree, mainly by some of the larger companies in the industry. For smaller companies or producers, cost is often prohibitive to moving forward. If it's cost-prohibitive for most people in the supply chain to approach, then even the best companies with the greatest systems still have to work within their partners' limitations. Time is a barrier. Models can be developed and configured, but it takes a lot of time to configure them, validate them against your system, and have someone who's responsible for the governance of those systems moving forward. What is the regulatory framework for using AI in food safety? WC: From a governance standpoint, we don't necessarily have a comprehensive, food-specific regulatory framework for how companies should use AI in food safety. We call it a food supply chain, but it's really not a chain — it's a complex network of siloed operations that intersect whenever a product is transformed, transported, or changes hands. Those entities don't necessarily have the same standards or priorities, or in some cases, the same regulatory requirements, and each entity generates large amounts of records and information. The food supply chain is global. While globalization is wonderful because it allows us to have certain things year-round and have more products, you're now navigating a bunch of different regulatory systems, record-keeping practices, and operational standards. Read the original article on Business Insider

Read it at Business Insider ↗

← Front page