Designing a Farm Emergency Plan Utilizing Artificial Intelligence

Area of Science:

• Agricultural Safety and Technology
• Artificial Intelligence Applications
• Emergency Preparedness

Background:

• Farm Emergency Plans (FEPs) are critical for agricultural safety.
• Evaluating the efficacy of AI in generating comprehensive FEPs is a growing area of research.
• Current AI systems may have limitations in understanding the complexities of diverse agricultural operations.

Purpose of the Study:

• To assess the capability of three AI systems (ChatGPT, Microsoft Copilot, Google Gemini) in generating functional FEPs for a Midwestern row crop grain farm.
• To evaluate the completeness and accuracy of AI-generated FEPs and emergency responses.
• To determine the necessity of human intervention and prompt specificity in AI-assisted FEP development.

Main Methods:

• Three AI systems were prompted with four levels of specificity to generate twelve distinct FEPs.
• A rubric, based on literature review, was used to evaluate FEP components.
• AI systems were also tested on responses to three specific farm emergencies: grain entrapment, chemical spills, and ammonia exposure.

Main Results:

• ChatGPT and Microsoft Copilot provided useful starting points for FEPs with detailed prompts; Google Gemini was less effective.
• None of the AI systems could independently generate complete and reliable FEPs.
• AI responses to specific emergencies were informative but required human validation and additional input.

Conclusions:

• AI systems can serve as valuable tools for agricultural emergency preparedness but require human guidance.
• Effective FEP creation necessitates a collaborative approach between AI, user expertise, and evidence-based resources.
• The capabilities of AI systems are rapidly evolving, suggesting future iterations may offer more comprehensive solutions.