Nourishing Humanity's Digital Future: AI Ethics in Agrifood Systems

The FAO Science and Innovation Strategy is grounded in seven guiding principles, one of which is ethics-based. In this context, Mr. Amandeep Singh Gill, USG Special Envoy on Technology shared this insightful blog on ethics in Artificial Intelligence.

Agrifood systems are the circulatory network of human civilization: when they function efficiently and effectively, societies thrive; when they falter, communities suffer. Like the human body that depends on a healthy flow of nutrients to sustain life, our world depends on agrifood systems to literally nourish billions of people every day. As we stand at the threshold of an artificial intelligence (AI) revolution that promises to transform how we produce, distribute, and consume food, we must ask ourselves a fundamental question: will AI enhance the vitality of this essential system, or will it inadvertently drain resources from the very foundation it aims to optimize? 

The Promise and Paradox of AI in Agrifood Systems 

AI holds transformative potential for agrifood systems. AI applications are already demonstrating their capacity to address some of agriculture's most pressing challenges, from precision agriculture that optimizes water and fertilizer use to advanced supply chain management that reduces waste and improves efficiency.

The promise is compelling: AI can detect crop diseases before they spread, predict yields with unprecedented accuracy, provide smallholder farmers with expert guidance through their mobile devices, and enable real-time monitoring that prevents food loss.  

Yet this promise comes with a paradox. Ethics, in its philosophical essence, concerns itself with principles that delineate appropriate conduct - standards that help us discern whether specific interventions are just, beneficial, and sustainable. When we deploy AI systems in agrifood contexts, we must grapple with questions that extend beyond technical efficiency: How do we ensure that AI benefits reach smallholder farmers in low-income countries, not just large-scale operations in wealthy nations? How do we prevent algorithmic bias from perpetuating existing inequalities? When AI systems optimize for productivity, how do we ensure they don't inadvertently harm the environment, displace workers, or concentrate power in the hands of a few technology providers?

The critical question becomes: at what point does AI deployment - intended to optimize agrifood systems - begin to "steal" resources from the very systems it aims to improve? This includes not only computational resources and energy consumption, but also data ownership, farmer autonomy, and the erosion of traditional knowledge. 

Complementing Rather Than Replacing Human Expertise 

As we consider ethical AI deployment in agrifood systems, specialized small language models offer a pathway that aligns with principles of human autonomy, sustainability, and inclusion. Unlike massive general-purpose models requiring enormous computational resources and contributing to climate change through energy consumption, smaller domain-specific models can run on existing devices, provide localized guidance, and function even in low-connectivity environments.

These tools can offer real-time pest identification, localized weather predictions, agronomic advice tailored to specific soil conditions, and market information - all in local languages and reflecting local agricultural practices. Importantly, they position AI as an assistant to farmers, not a replacement for farmer knowledge. This approach respects the "human autonomy and oversight" principle, ensuring that farmers retain agency over decisions affecting their land and livelihoods.

Companies developing such solutions demonstrate that agricultural AI need not follow extractive models of centralized data collection and control. Instead, modular, accessible AI tools can empower smallholders, preserve traditional knowledge systems, and enhance rather than undermine farmer autonomy. 

Ethical Frameworks Guiding AI Deployment 

Within this broader architecture, specific ethical frameworks provide operational guidance.

The Principles for the Ethical Use of AI in the UN System, endorsed by the UN System Chief Executives Board in September 2022, establish ten foundational principles grounded in human rights. These principles - including "do no harm," fairness and non-discrimination, sustainability, human autonomy and oversight, and transparency - were explicitly designed to ensure AI systems respect human dignity throughout their lifecycle.

For agrifood applications, these principles carry particular weight. The "do no harm" principle requires that precision farming technologies don't disadvantage farmers who lack digital literacy or internet connectivity. The fairness principle demands that algorithms determining credit access, insurance rates, or market opportunities for farmers don't perpetuate historical discrimination. The sustainability principle ensures that AI systems account for environmental impacts, not merely short-term productivity gains. 

Addressing the AI Capacity Divide in Agriculture 

Perhaps no sector illustrates the "AI divide" more clearly than agriculture. Of the top 100 high-performance computing clusters capable of training large AI models, not one is hosted in a developing country^[1] - yet developing nations are home to the vast majority of the world's farmers. This capacity gap threatens to transform AI from an equalizing force into a mechanism that deepens existing inequalities.

The Secretary-General's report on Innovative Voluntary Financing Options for Artificial Intelligence Capacity-Building, issued in July 2025 pursuant to the Global Digital Compact, proposes concrete mechanisms to address this divide. The report categorizes countries into five AI maturity tiers, from "AI nascent" (Tier 0) to "AI developers" (Tier 4), recognizing that financing strategies must match developmental contexts.

For Tier 0 and Tier 1 countries - where most agricultural populations reside - the report proposes establishing a "minimum irreducible AI capacity" encompassing basic skills, compute access, relevant data, adaptable models, a national AI strategy, and international collaboration mechanisms.  

For agrifood systems, these capacity-building initiatives are transformative. They enable agricultural ministries to develop AI-informed policies, support universities in training agronomists who understand both farming and algorithms, facilitate the creation of multilingual agricultural data sets that reflect local crops and conditions, and provide compute resources for developing context-specific models - such as pest detection systems trained on local species rather than only recognizing pests from other continents. 

Building a Globally Inclusive AI Governance Architecture 

The United Nations has moved decisively to build globally inclusive AI governance architecture; in August 2025 it established two crucial mechanisms: the Independent International Scientific Panel on AI and the Global Dialogue on AI Governance. Ethics is not peripheral to these bodies: it is central to their mandates.

The Scientific Panel, comprising 40 independent experts serving three-year terms, will provide annual reports on AI's risks, opportunities, and impacts. Critically, its work includes assessing ethical dimensions of AI deployment across sectors, including agriculture.  

The Global Dialogue on AI Governance provides an inclusive platform for governments, civil society, academia, and the private sector to discuss critical AI issues. This multi-stakeholder approach recognizes that ethical AI governance cannot be dictated from above - it must emerge through inclusive deliberation that brings diverse perspectives, including farmers' organizations, indigenous communities, and agricultural cooperatives, into conversation with technologists, scientist and policymakers. 

Mr. Amandeep Singh Gill was appointed Under-Secretary-General and the United Nations Secretary-General's Envoy on Technology in June 2022, and since 1 January 2025 serves as Under-Secretary-General and Special Envoy for Digital and Emerging Technologies. In this role, he spearheads the Secretary-General's strategic initiatives in technology, facilitates coordination across the UN system, and engages with a diverse array of stakeholders and partners globally. He is also a member of the Secretary-General's High Level Advisory Body on AI and of his Scientific Advisory Board. His extensive experience in technology and diplomacy includes serving as Executive Director and co-lead of the UN Secretary-General's High-Level Panel on Digital Cooperation (2018-2019) and as India's Ambassador and Permanent Representative to the Conference on Disarmament in Geneva (2016-2018). In recognition of his influential work in the field of artificial intelligence, TIME magazine named Mr. Gill one of the 100 most influential people in AI in 2024.