Hallucination Behavior in Multimodal LLMs Across Agricultural Image Interpretation and Generation Tasks

ArXi:2605.27595v1 Announce Type: cross Large Language Models (LLMs) are being rapidly adopted in agricultural imaging applications, ranging from crop interpretation to synthetic field image generation. However, these models frequently exhibit hallucinations outputs that appear confident yet deviate from biological or environmental reality potentially leading to misinformed agronomic insights.