[论文解读] Farmer.Chat: Scaling AI-Powered Agricultural Services for Smallholder Farmers

Namita Singh, J. Wang’ombe+10 more|arXiv (Cornell University)|Sep 13, 2024

FinTech, Crowdfunding, Digital Finance被引用 7

一句话总结

Farmer.Chat 是一个生成式 AI 聊天机器人，为小农户提供多语言、多模态、情境感知的农业建议；在 four countries 部署，拥有 15k+ 用户和 300k+ 查询，显示出可扩展的 extension 服务和高用户满意度。

ABSTRACT

Small and medium-sized agricultural holders face challenges like limited access to localized, timely information, impacting productivity and sustainability. Traditional extension services, which rely on in-person agents, struggle with scalability and timely delivery, especially in remote areas. We introduce FarmerChat, a generative AI-powered chatbot designed to address these issues. Leveraging Generative AI, FarmerChat offers personalized, reliable, and contextually relevant advice, overcoming limitations of previous chatbots in deterministic dialogue flows, language support, and unstructured data processing. Deployed in four countries, FarmerChat has engaged over 15,000 farmers and answered over 300,000 queries. This paper highlights how FarmerChat's innovative use of GenAI enhances agricultural service scalability and effectiveness. Our evaluation, combining quantitative analysis and qualitative insights, highlights FarmerChat's effectiveness in improving farming practices, enhancing trust, response quality, and user engagement.

研究动机与目标

• Address the information access gap for smallholder farmers through scalable AI-powered extension.
• Develop a multilingual, multimodal, context-aware knowledge base paired with retrieval-augmented generation.
• Evaluate user trust, engagement, and adoption factors in real-world deployments.
• Demonstrate regional deployment feasibility and openness of the codebase to the research community.

提出的方法

• Design a user-centered, scalable AI platform with ease of use, multilingual multimodal interactions, and a comprehensive knowledge base.
• Build a Knowledge Base Builder to ingest, structure, and index diverse data formats (text, images, videos) into vector embeddings.
• Implement AI modules using Retrieval-Augmented Generation (RAG) and a Query Orchestration pipeline (planning, execution, tooling) for real-time, personalized responses.
• Provide local language, voice, and video support with translation and speech processing pipelines (translation via Google Translate, ASR, TTS).
• Integrate with frontend platforms (WhatsApp, Telegram) and allow future IVR or mobile app deployments.
• Establish a continuous feedback loop with conversation logs and analytics to drive content and model improvements.

实验结果

研究问题

• RQ1 RQ1: How can generative AI enhance the scalability, accessibility, and contextual relevance of agricultural extension services?
• RQ2 RQ2: What factors influence user trust, engagement, and long-term adoption of AI-driven advisory tools like Farmer.Chat?
• RQ3 RQ3: How does deployment of Farmer.Chat impact agricultural outcomes, farmer satisfaction, and community-level adoption across diverse regions?

主要发现

• Over 75% of queries are answered by the system, indicating strong baseline performance.
• Farmer.Chat engages 15,000+ farmers and handles 300,000+ queries across deployments in four countries.
• Kenya deployment alone includes 8,805 users across seven counties and 225,500+ queries, with livestock (Dairy 27.9%, Chicken 20.65%) and crops like Potato (11.72%), Avocado (10.93%), and Coffee (12.63%) driving usage.
• The platform supports multilingual and multimodal interactions to improve accessibility for low-literacy users, including voice notes and translations.
• Users report high satisfaction with response quality and relevance, with engagement centered on yield, pest control, and weather topics.
• The system integrates external services (e.g., TomorrowIO weather, Plantix disease diagnostics) to augment decision-making.