What the 2026 conference conversation reveals about data culture, validation and the move from AI experimentation to trusted decisions.
Based on Episode 5 of the Predictably Good podcast.
Artificial intelligence was everywhere on the food industry conference circuit in 2026, but the important change was in the questions being asked. AI was no longer confined to specialist sessions about what the technology might do. It appeared in discussions about food safety culture, analytical science, regulation and operational decisions.
The focus has shifted to trust: Can the data be relied on? Can an output be validated? Who remains accountable when a decision has real consequences? In Episode 5 of the Predictably Good podcast, Creme Global’s Brendan Ring and James Doyle joined CEO Cronan McNamara to examine that shift and what may follow.
AI is becoming part of the infrastructure
Across industry and scientific forums, AI is becoming embedded in everyday tools. Digital systems now support specification searches, trend analysis, product development, rapid testing and early-warning discussions. The FDA’s New Era of Smarter Food Safety blueprint points in the same direction by connecting predictive analytics with stronger data, traceability, information sharing and food safety culture.
The challenge begins when AI moves beyond demonstration and becomes part of a consequential workflow. Food organisations already hold extensive evidence in laboratory results, supplier records, audits, specifications and incident reports. AI-assisted tools can help locate and structure that information, but they do not remove gaps, inconsistent definitions or missing context. Data still needs to be organised around a clear decision.
Prediction needs evidence
The ambition is preventive: identify a weak signal early enough to act before it becomes an incident or recall. That matters when contaminated food continues to impose a major global health burden. Yet a prediction is useful only when the evidence behind it can be examined.
One practical test is back-testing. A team timestamps historical data, restricts a model to information that would have been available at the time and compares its prediction with what happened next. This shows whether the system could have added value under realistic conditions. Traceability is equally important: analysts must be able to inspect a signal and return to its source. NIST’s AI Risk Management Framework reinforces the need for governance, human oversight and testing that reflects intended use.
Human judgement remains central
More capable AI makes the human role more visible, not less important. A system can extract information, group complex records and surface unusual patterns. It cannot take responsibility for a false alarm, a missed signal or an action based on incomplete evidence.
For now, the most credible role for AI in food safety is decision support. People need the authority and understanding to challenge, override or escalate an output. That is why data culture matters as much as data architecture: teams must be willing to test assumptions, document uncertainty and ask whether a result is useful for the decision at hand.
Putting the thinking into practice
Creme Global works with food organisations, regulators and scientific partners to structure complex information, harmonise internal and public datasets, create secure data-sharing environments and develop predictive analytical tools. Food safety specialists and data scientists work together to frame the question, evaluate evidence, test performance and make results interpretable to the people responsible for acting.
The next phase of AI adoption will be defined less by access to technology than by the ability to use it responsibly. Progress will depend on transparent evidence, fit-for-purpose validation and clear human oversight. A model can surface a signal; an organisation must still understand it, trust it appropriately and act.
Listen to Episode 5 of the Predictably Good podcast for the full conversation on AI, data culture and the future of food safety.
AI capabilities, public datasets and regulatory expectations continue to evolve. Organisations should validate systems for their intended use and assess them against applicable legal, regulatory and food safety requirements.
Key sources
- World Health Organization, Food safety fact sheet
- US Food and Drug Administration, New Era of Smarter Food Safety Blueprint
- National Institute of Standards and Technology, AI Risk Management Framework
- European Commission, Rapid Alert System for Food and Feed
- Creme Global, Predictably Good Podcast Episode 5 transcript, 2026
