Context and Motivation
Despite the evolution of food safety systems such as HACCP, food-related incidents remain frequent, primarily caused by improper operator behaviors (hygiene, equipment use, and procedural implementation). Correcting these behaviors is vital, and recent efforts have focused on how Artificial Intelligence (AI) can support this process.
AI-Assisted Continuous Cycle
The authors of this study propose a three-phase cycle aimed at improving food safety behavior through AI:
Monitoring – Smart cameras and sensors observe behaviors such as hand hygiene and the wearing of protective equipment.
Evaluation – Using data analysis and Large Language Models (LLMs), organizational culture is assessed and deficiencies are identified.
Action – Real-time feedback, personalized training, and AI-assisted communication tools are implemented, followed by a renewed monitoring phase.
Phase-by-Phase Breakdown
Phase 1: AI Monitoring
This allows for discreet and continuous observation, eliminating the distortion caused by the presence of human observers. The collected digital data supports integrated analyses of food safety culture.
Phase 2: AI Evaluation
Qualitative and quantitative components are integrated through the analysis of large volumes of data (videos, sensors, interviews, surveys). Predictive models, NLP, and LLMs offer advanced insights and forecasts regarding intervention efficacy.
Phase 3: AI Intervention
This involves solutions such as real-time feedback, VR/inclusive training, personalized recommendations, and AI-assisted communication (e.g., chatbots, subtle reminders). The emphasis is placed on the employee as a central agent in the food safety culture.
Challenges and Limitations
Technical: Classification errors, sensitivity to LLM prompts, and the need for human oversight.
Ethical and Legal: Concerns about employee data privacy and transparency in AI usage.
Organizational: Resistance to change and the need for investments in infrastructure and training.
Future Research Directions
The article outlines key recommendations including:
AI Model Enhancement: Emotional recognition, sensor integration, and longitudinal evaluation.
Evaluation Efficiency: Combining qualitative and quantitative analytical methods.
Practical Implementation: Personalized training, VR integration, culturally adapted communication, field studies, regulatory guideline development, and sector-wide collaborations.
Conclusion
The article argues that AI has the potential to shift the food safety paradigm from reactive to preventive, fostering a proactive safety culture. However, technology should be viewed as a complement, not a replacement for existing systems, and should be implemented with caution, human oversight, and ethical integrity.
Wang et al. (2025) contend that human behavior is essential in mitigating food risks, and that AI provides the necessary tools for continuous monitoring, in-depth evaluation, and effective intervention. The discreet monitoring of operators' behavior allows for the collection of accurate data, which, once analyzed by AI, enables cultural diagnostics and tailored strategies. AI-assisted interventions – from immediate feedback to VR training and chatbots – are designed to strengthen organizational culture and reduce human error. The study also acknowledges technical and ethical limitations, calling for an integrated and responsible approach. Future research will focus on empirical validation across diverse industrial settings and the development of regulatory guidelines to support ethical and effective AI adoption.
Source SCIENCE DIRECT.