Generative AI as an Interface Between Farmers and AI Algorithms: Opportunities and Challenges

Abstract Artificial Intelligence (AI) has transformed industries worldwide, yet its adoption in agriculture remains limited, particularly in low-resource settings. Farmers in developing countries face two chronic challenges: (1) the assumption of reliable Internet access is often an incorrect one to make in low-resource settings in developing countries; and (2) the complexity of AI algorithms often makes it challenging to be used by low-resource farmers, who often lack the required significant technical expertise needed to use these AI algorithms effectively. As a result, many low-resource farmers are excluded from the benefits of AI technologies, thereby exacerbating the global digital divide. This paper explores whether Generative AI, particularly large language models (LLMs) like ChatGPT, Gemini, and LLaMA, can bridge this ever-increasing divide, and make AI tools and algorithms more accessible to farmers in developing countries. LLMs represent a breakthrough in interactivity, allowing users to engage conversationally with AI algorithms. Historically, AI has lacked this interactivity, making it difficult for non-technical users to understand or act on algorithmic outputs. Generative AI models, however, offer a solution by serving as intuitive interfaces that translate complex AI algorithms into actionable insights. These models enable farmers to ask questions and receive answers in natural language, along with enabling them to provide feedback so that the underlying AI algorithms could be adapted to their specific needs. Furthermore, by leveraging non-Internet modalities like voice-based interfaces and SMS, these models can extend AI’s reach to remote areas lacking reliable Internet access. This paper takes the position that Generative AI can indeed serve as a bridge between low-resource farmers and heavy-duty AI algorithms. This paper discusses how LLMs bring a unique level of interactivity to AI. We then describe a vision for how these models can enable farmers to access algorithmic agricultural advice in an accessible and user-friendly manner. Subsequently, we outline the challenges that must be overcome to realize this vision, including technical, infrastructural, and ethical considerations. Finally, we conclude with a discussion on the broader implications of using Generative AI to address the needs of low-resource farming communities. By exploring this intersection of Generative AI and agriculture, this paper aims to contribute to the ongoing dialogue on making AI technologies inclusive and impactful for underserved populations, ensuring that the transformative potential of AI is enjoyed equitably by all sections of society.