Science & Technology14 October 20265 min read
Second UN AI-for-agri-food meeting in Delhi calls for farmer-centric AI frameworks
The second meeting of the UN Global Initiative on AI for Agri-Food Systems (GI-AI4FS) was held at the India Mobile Congress 2026 in New Delhi. It was jointly hosted by the Telecommunication Engineering Centre (TEC), which works under the Department of Telecommunications, and the Indian Council of Agricultural Research (ICAR). The meeting called for trustworthy, interoperable and farmer-centric AI frameworks and closer international cooperation on digital governance.
What the initiative is
GI-AI4FS is a United Nations-backed multilateral platform that aims to use artificial intelligence and the Internet of Things in agri-food systems. It is coordinated by the International Telecommunication Union (ITU) with the Food and Agriculture Organization (FAO), the International Fund for Agricultural Development (IFAD) and the World Food Programme (WFP). It aligns with Sustainable Development Goal 2 (Zero Hunger) and Goal 12 (Responsible Consumption and Production).
Its stated objective is to promote trustworthy, interoperable and farmer-centric AI solutions while strengthening cooperation on governance, standards and capacity building. The first meeting was held at FAO headquarters in June 2026. India's hosting of the second is read as strengthening its role as a convenor on AI governance and digital public infrastructure for the Global South.
How far AI has reached Indian farms
Several applications are already in use. Project Farm Vibes, a Microsoft initiative in Maharashtra, used AI to raise crop yields by 40% while cutting fertiliser use by 25%. An AI-based pilot for local monsoon onset forecasting for Kharif 2025 reached 3.88 crore farmers across 13 states, and 31 to 52% of surveyed farmers changed their land preparation and sowing decisions.
Farm Again, a Tamil Nadu startup, deployed solar-powered sensors costing ₹2.5 lakh across 4,000 acres. This doubled coconut yields and saved 4,00,000 cubic metres of water a year. The National Pest Surveillance System uses AI to cover 66 crops and over 432 pest types, with real-time advisories to more than 10,000 extension workers.
In crop insurance, YES-TECH uses AI and remote sensing for yield assessment and has been fully adopted by Madhya Pradesh. CROPIC lets farmers upload geotagged, time-stamped images for damage evaluation. Both support the Pradhan Mantri Fasal Bima Yojana (PMFBY). Price prediction draws on e-NAM and AGMARKET data to help farmers time their sales.
The public digital backbone sits under the Digital Agriculture Mission:
- AgriStack issues Farmer IDs linked to land, livestock and crop records. Over 7.63 crore have been generated against a target of 11 crore by 2026-27. It also drives the Digital Crop Survey, which covered 23.5 crore plots in Rabi 2024-25.
- The Krishi Decision Support System combines satellite, weather and soil data to produce crop maps, yield estimates and drought assessments.
- Bharat-VISTAAR, proposed in Union Budget 2026-27, is a multilingual AI tool meant to bring AgriStack portals and ICAR packages into one advisory system.
- Kisan e-Mitra, a voice-enabled chatbot in 11 regional languages, had answered over 93 lakh queries as of December 2025, handling 8,000 a day on schemes such as PM-KISAN and PMFBY.
The Soil and Land Use Survey of India (SLUSI) is running a village-level soil mapping project at 1:10,000 resolution. About 29 million hectares have been mapped against a target of 142 million hectares, across six funded states. Under Rashtriya Krishi Vikas Yojana (RKVY), 2,282 agri-startups have received support, with up to ₹5 lakh at the idea stage and ₹25 lakh at the seed stage.
In October 2025 the government, with the World Economic Forum, IndiaAI and BCG, released the "Future Farming in India: AI Playbook for Agriculture". It introduced the IMPACT AI Framework, built on three pillars: Enable (strategy, policy and digital public infrastructure), Create (collaboration to design and validate local solutions) and Deliver (integrating AI into extension services so it reaches small and marginal farmers).
What stands in the way of smallholders
About 86% of landholdings are small and marginal, under 2 hectares. AI models trained on large, mechanised, mono-cropped farms often give skewed advice when applied to India's fragmented, multi-cropped and rain-dependent farms.
The WEF's 2025 report says farmers face information clutter from scattered digital sources and are wary of dropping tried practices for algorithmic advice without physical validation.
AgriStack depends on the Digital India Land Records Modernization Programme (DILRMP). A 2026 parliamentary report notes that 97.4% of cadastral maps are digitised. Un-updated spatial records, unresolved tenancy disputes and the lack of conclusive titling still block a single source of truth.
Other concerns are data misuse, such as predatory pricing and discriminatory credit scoring, in the absence of agricultural data protection, and the high capital cost of sensors, drones and autonomous machinery, which could concentrate benefits among wealthy landowners. Data centres may use 9.3 trillion litres of water a year by 2030, and AI hardware could generate 2.5 million tonnes of e-waste annually by that year.
Remedies proposed
To address the cost barrier, a Farming-as-a-Service model is proposed through Custom Hiring Centres, Farmer Producer Organisations and Kisan Drones. Smallholders could rent AI-enabled machinery, such as spraying drones run by women self-help groups, on a pay-per-use basis.
A 'phygital' extension network is also proposed, linking Krishi Vigyan Kendras with digital systems and equipping Krishi Sakhis with AI diagnostic tablets. Other measures are faster rollout of AgriStack and the Digital Crop Survey across all districts by FY 2026, upskilling extension workers into "digital agropreneurs", and sharing open-source models and data governance standards through GI-AI4FS.
The proposals also include extending the Data Empowerment and Protection Architecture (DEPA) to agriculture, giving farmers consent-based control over who accesses their farm-level data while allowing secure sharing with verified agri-tech startups.