AI for Inclusive and Resilient Agricultural Food Systems

Abstract

The session examined how artificial intelligence can make agricultural supply chains more sustainable, resilient, and inclusive—especially for smallholder farmers and SMEs in low‑ and middle‑income countries. Opening remarks highlighted the Netherlands’ strategic commitment to AI‑driven food‑security and the OECD’s focus on trustworthy, interoperable AI governance. The panel explored concrete AI use‑cases (precision irrigation, pest‑early‑warning, climate‑adaptive breeding, traceability, and logistics optimisation), identified persistent gaps (data scarcity, trust, digital divide), and debated policies and partnership models needed to scale responsible AI across diverse agro‑ecological contexts.

Detailed Summary

Speaker: H.E. Harry Verweij (Netherlands)
- Stressed that digitalisation and AI are accelerating food‑productivity, climate resilience, and nature conservation.
- Emphasised the Netherlands’ role as a global agro‑innovation hub despite its small size, citing companies such as ASML, NXP, Philips and Dutch universities.
- Highlighted precision‑farming successes: up to 90 % water savings via smart irrigation, AI‑optimised yields with minimal inputs, and disease‑prediction models.
- Outlined the Dutch ambition to share ICT‑agri solutions with low‑ and middle‑income countries, tailoring tools to local challenges (early‑warning for pests, water‑use optimisation, plant‑breeding assistance).
- Called for inclusive AI that empowers small farmers, protects data, and supports entry into global supply chains.
- Re‑affirmed partnership with the OECD, the FAO, and Indonesia, and praised India’s leadership in hosting the summit.

2. OECD Perspective – Trustworthy AI, Evidence, and Gaps

Speaker: H.E. Mathias Cormann (OECD)
- Described today’s agri‑food systems as operating in a highly volatile environment (droughts, floods, pests, conflicts, economic crises).
- Presented impact evidence:
  - AI‑enabled precision spraying ↓ pesticide use by up to 30 % without yield loss.
  - Computer‑vision weed‑targeting ↓ herbicide use by up to 50 %.
  - AI‑assisted breeding identified drought‑tolerant traits, raising yields up to 25 % during end‑season droughts (central Europe).
  - AI hybrid‑rice platform in Asia shortened breeding cycles and boosted resilience.
  - AI‑driven traceability and smart logistics reduced post‑harvest losses and improved market transparency.
- Highlighted the digital divide: 96 % of Australian farmers use digital tools versus 12 % in Chile.
- Identified structural barriers – high costs, limited digital skills, fragmented data‑governance, and lack of trust.
- Introduced the OECD AI Policy Toolkit (hosted on OSD.AI) covering >2,000 policies across 80 jurisdictions, aimed at guiding responsible AI adoption in agriculture.
- Stressed the need for interoperability and public‑private collaboration to avoid deepening inequalities.

3. Panel Discussion

3.1 FAO – Anticipatory Action & Inclusivity

Speaker: Dr. Dejan Jakovljevic (FAO)
- Framed AI from a food‑system ecosystem perspective, noting that agriculture interacts with water, transport, and other sectors.
- Warned that the digital divide excludes farmers from essential services; AI can widen this gap if not managed responsibly.
- Presented a positive example from India: a phone‑based advisory service (no smartphone required) providing multilingual guidance on shrimp cultivation, pest management, etc., thereby lowering entry barriers.
- Emphasised anticipation as the key AI function: creating decision‑support tools, situation rooms, and early‑warning systems to pre‑empt shocks (climate, conflict, market).
- Stated the urgency: ≈700 million people currently lack sufficient food, underscoring the need for rapid, coordinated AI‑enabled action.

3.2 Indonesia – AI Roadmap, Governance, and Use‑Cases

Speaker: Prof. Arvin Sumari (Indonesia)
- Outlined Indonesia’s geographic challenges (17,000 islands, 36 % land, 64 % water) and the digital‑infrastructure gap (regional time‑zone differences, uneven telecom coverage, talent distribution).
- Described AI use‑cases:
  1. Soil‑condition & nutrition prediction for allocating new rice fields (≈1 million ha target).
  2. Crop‑selection optimisation based on island‑specific soil profiles.
  3. Fertiliser‑water optimisation (experimented with AI‑driven blend ratios).
  4. Intelligent farming (end‑to‑end knowledge growth, from seed‑placement to harvest logistics).
  5. Weather & flood forecasting to reduce crop failure.
  6. Logistics‑route optimisation (e.g., rice price disparity: US  $1 in e a s t er ni s l an d s v s U S$ 5 after transport).
- Presented Indonesia’s AI Roadmap (seven pillars):
  1. AI Regulation (transparency, explainability).
  2. AI Ethics.
  3. Investment & Financing (public and private).
  4. AI Export/Data Sharing.
  5. AI Research & Innovation.
  6. AI Talent Development.
  7. AI Use‑Case Development.
- Emphasised a “quad‑helix” ecosystem (government, industry, academia, civil society) to ensure no stakeholder is left behind.

3.3 India (NITI) – Public‑Private Partnerships & Preventing the AI Divide

Speaker: Ms. Debjani Ghosh (NITI Frontier Tech Hub)
- Asserted that many projects “throw AI at every problem” without clear problem definition, leading to fragmented pilots.
- Identified food‑wastage as the most tractable AI target: logistics optimisation, cold‑chain monitoring, and market‑linkage platforms can dramatically cut loss.
- Advocated for sector‑specific Centers of Excellence (CoEs) that focus on concrete challenges (e.g., cold‑chain resilience, climate‑adapted crops) rather than generic AI labs.
- Called for aligned problem statements, clear pathways to commercialization, and collaborative governance to ensure scalable, impact‑driven solutions.

3.4 Netherlands (Wageningen) – Bridging Research & Low‑Tech Environments

Speaker: Dr. Arun Pratihast (Wageningen University)
- Highlighted three systemic obstacles:
  1. Data scarcity & non‑shared datasets—without local data, global models fail to deliver local relevance.
  2. Lack of trust – farmers often disregard AI advice that feels opaque or irrelevant.
  3. Scalability beyond technical performance – models must work with limited connectivity and literacy.
- Shared concrete projects:
  - World Serial Project (in partnership with ESA) mapping global croplands; limited by countries’ reluctance to share satellite data.
  - Cocoa agro‑forestry chatbot (multi‑language, computer‑vision disease detection) that engages farmers directly.
  - Emphasised the need for data‑as‑infrastructure, i.e., embedding farmers in data collection pipelines to create trustworthy, actionable models.

4. Closing Remarks

Moderator thanked panelists and reiterated key themes:
- Vast AI potential (precision farming, anticipatory tools, supply‑chain transparency).
- Inclusivity must be intentional: smallholders, women, and remote communities need equitable access and data protection.
- Problem‑driven approach: solutions must originate from farmer needs, not from technology hype.
- Trust, explainability, and responsible data governance are prerequisites for adoption.
Acknowledgements: gratitude to the Kingdom of the Netherlands for co‑hosting, to the OECD for thematic guidance, and to all participants for their diverse perspectives.

Key Takeaways

AI can dramatically improve productivity and climate resilience, e.g., up to 90 % water savings in smart irrigation and 30 % pesticide reduction via precision spraying.
Inclusivity is a core condition: without equitable digital access, AI risks widening the gap between well‑connected and remote farmers.
Anticipatory AI tools (early‑warning, situation rooms) are vital for pre‑empting shocks such as droughts, floods, and market disruptions.
Data sharing and ecosystem building are the biggest bottlenecks; national and international policies (e.g., OECD AI Policy Toolkit) aim to create interoperable, trustworthy frameworks.
Sector‑specific Centers of Excellence are recommended to focus AI development on high‑impact problems like food waste reduction and cold‑chain resilience.
Indonesia’s seven‑pillar AI roadmap (regulation, ethics, financing, export/data, research, talent, use‑cases) provides a replicable model for aligning governance with agricultural needs.
Real‑world pilots must be farmer‑centred: solutions should operate in low‑tech environments, be multilingual, and provide transparent, actionable advice.
Public‑private partnerships need clear problem definitions, commercialization pathways, and shared responsibility for data stewardship to succeed at scale.
Trust and explainability are non‑negotiable for farmer adoption; models must be transparent, and data ownership must be respected.
Global collaboration (Netherlands, OECD, FAO, India, Indonesia, and others) is essential to pool expertise, finance, and policy tools for a resilient, inclusive AI‑enabled food system.

See Also:

India AI Impact Summit 2026

Explorer

ai-for-inclusive-and-resilient-agricultural-food-systems