Artificial Intelligence: the Next Frontier in Agrifood Systems Transformation
![Artificial Intelligence: the Next Frontier in Agrifood Systems Transformation](https://www.fao.org/images/loblibraries/agriculture-images/1709114612339.jpeg?sfvrsn=c2474210_1)
©freepick
"Whenever someone mentions “artificial intelligence,” or AI to me in the context of agriculture, I get excited and can’t help thinking we are at the brink of a massive paradigm shift that is reshaping the landscape of global food security. I can see the potential it offers to bring solutions closer to the one who most need it, leapfrog on some of the most persistent development and complex challenges we face. To be totally honest, when I hear AI, the first thing that comes to my mind is “Avian Influenza”, a simple professional quirk from two decades chasing viruses as an epidemiologist across the world, but this will be the subject of another story.
And while I am excited about AI’s prospects, I can’t help but hear the alarm bells ringing in the back of my mind. We are human after all, and as humans we tend to make tools that have the potential of great things but also for disastrous ones[1]. AI invokes a future where the promise of technological innovation is as significant as the real challenges it presents. We need a balanced evaluation of its role in the transformation of our agrifood systems. We need to navigate this transformation with a bit of wisdom.
And it seems that everyone is scrambling to get their hands on it and for good reasons. As the world grapples with the multifaceted dimensions of food insecurity—physical availability, affordability of healthy diets, utilization, and stability—AI emerges as a potentially critical ally. To feed the planet’s growing population, global agriculture will need to produce more food in the next 50 years than in the previous 10,000.[2] Amidst the backdrop of climate change, geopolitical unrest, and economic uncertainties, its potential to enhance the resilience and sustainability of agrifood systems is something we need. In the great stampede for AI solutions in agriculture, nobody wants to be left behind.
Artificial intelligence, although causing a resounding buzz today, is not a new phenomenon. It embodies a scientific discipline that has evolved since the 1950s, originally aimed at reproducing cognitive abilities for mathematical demonstrations and imitating human reasoning. A turning point occurred in 2012 marked by a confluence of factors: the spectacular rise in computing power in the 2000s, the rise of Big Data, the use of statistics and probabilities to detect similarities within this massive amount of data, and the advent of machines that 'learn on their own,' and whose learning capacity is based on the principle of self-learning neural networks. The arrival of generative AI in 2022 added a new dimension, transforming this revolution into a phenomenon at least as crucial as the emergence of the Internet. A significant shift occurred with the notable arrival of large language models such as Chat GPT, which has amplified the scope of artificial intelligence, prompting questions about the pace and lag of our adaptation to this emerging technology.
Fascinating as much as it scares us, willing or not, artificial intelligence has entered our lives, making scenarios that were once science fiction tangible. It is also in this accelerated phase that new fields of application have emerged, offering promises of considerable progress in areas such as health, education, or agriculture.
The issue today isn’t whether we should use AI to make our agrifood systems more capable. It is how to do so in a way that does not exacerbate inequalities but fosters sustainability. The AI landscape is not equal. High-income countries typically have more robust data and thus more effective AI models but also skills to use the technology where it may make greater positive impact. FAO's mission is to help countries in the Global South merge those gaps and channel the use of AI to make our agrifood systems more sustainable, resilient and inclusive. If we do a good job, we can feed the world. If we don’t, we create one where we’re worse off.
On the upside
Anyone can see the upside of AI-driven solutions and the world is moving quickly towards AI-driven farming. Innovations range from self-driving tractors and autonomous sprayers to sophisticated indoor farming operations that leverage computer vision and AI to optimize crop growth conditions and detect plant pests and diseases. As outlined in FAO’s Harvesting Change foresight report, advancements that include AI-driven digital twins (a virtual model that mirrors a physical asset, dynamically updating with its real-world counterpart), an unheard-of acceleration in personalized nutrition, even the distant prospect of Artificial General Intelligence (a theoretical form of AI that could perform any intellectual task that humans or animals can, potentially surpassing human intelligence in most activities) are no longer science fiction but ideas that live along the horizon of the possible.
Farmers (84 percent of 540 million farms globally are smallholders)[3], traditionally reliant on their knowledge of the land and the cycles of nature, are now deploying AI to navigate the complexities of modern agriculture. Central to AI's value in agriculture is its capacity for data analysis and decision-making. By integrating diverse data types from IoT sensors, drones, and computer vision technologies, AI systems can monitor every aspect of the agricultural process in real-time. This capability allows for precision farming—adjusting water use, fertilizer application, and pest control measures dynamically in response to environmental conditions and crop needs. The result is a marked increase in efficiency, with tangible benefits including higher yields, reduced waste, and produce more with less.
Here are a few concrete examples where AI is contributing to a larger transformation that goes beyond digital innovation, establishing itself as a cornerstone of contemporary biotechnological advancement.
AI-driven technologies are revolutionizing marine research by enabling the analysis of massive datasets gathered from satellites, buoys, and underwater autonomous vehicles. These advanced systems employ algorithms capable of detecting the unique sounds and signals emitted by marine organisms. This capability not only aids in the discovery of new marine species but also provides critical insights into the shifting dynamics of marine ecosystems. Given that only 240,000 out of an estimated 2.2 million marine species have been identified so far, the role of AI in revealing the secrets of the ocean's expansive depths is proving to be indispensable.[4]
AI in aquaculture feeding brings multifaceted benefits, enabling farmers to cut feed costs and reduce production expenses through optimized feeding regimes. These AI systems not only enhance fish health and mitigate disease risks, minimizing the reliance on antibiotics but also diminish environmental impacts by curbing waste output. As global seafood demand escalates alongside population growth, AI-driven feeding technologies emerge as key to sustaining future seafood supplies. Other initiatives have shown that AI systems can improve order and inventory accuracy, leading to better stock management, increased sales, and fresher produce for consumers. Other initiatives have shown that AI systems can improve inventory accuracy, leading to better stock management, increased sales, and fresher produce for consumers.
The potential of AI also extends beyond crop and fish production. It can play a crucial role in addressing food loss and waste. Between 691 and 783 million people faced hunger in 2022, with a mid-range of 735 million (FAO, 2023). While hunger and food security continue, an estimated 13 percent of the world’s food is lost in the supply chain from post-harvest prior to retail (FAO, 2022); a further 17 percent of food is wasted in households, food services and in retail.[5] By improving the efficiency of the food supply chain from farm to table, AI can help minimize food loss and waste and its associated environmental impact. This not only conserves resources but also contributes to the mitigation of climate change by reducing greenhouse gas emissions from decomposing waste.
In animal husbandry, the "Internet of Animals" (IoA), humorously named in homage to the "Internet of Things" (IoT), leverages AI to enhance livestock management, offering valuable insights into breeding, reproduction, and the butchering process. It enables farmers to monitor their animals' health and behavior in real time. Utilizing wearable AI devices, farmers can gather immediate data on their livestock, aiding in smart decision-making. These devices alert them to health issues, vaccination schedules, and breeding readiness, showcasing the diverse applications of AI in animal husbandry.[6]
Bridging traditional knowledge, sustainable practices and AI: here too, AI can have a positive effect. Agroecology involves developing an ecologically balanced system that incorporates multiple species with varied growing conditions—a complex endeavor. AI algorithms, georeferenced programming and data on crops, plant species, or forestry can support agroecological modeling efforts globally. Furthermore, a significant challenge in agroecology is showing the effects of using non-chemical inputs over time. Leveraging big data and AI for targeted strategies allows for the demonstration of these impacts.
Navigating the unknown and minimizing challenges
With AI promise comes challenges,
This is a technology that is evolving rapidly and there are significant issues that require deep consideration. Raphaël Gaillard, in his recently published book 'L'homme Augmenté' (The Augmented Man), my nightstand book at the moment, proposes an interesting perspective on AI. Rather than seeing it as a competitive threat to humanity, he suggests hybridization between artificial intelligence and our own intelligence. The challenge lies not in rivalry but in the success of this symbiosis, which is not without significant challenges, and some of them are daunting according to the author. However, while we are not yet at this stage, it is crucial not to underestimate the ethical issues accompanying such an evolution. Biases in training data can be reflected in AI models, and human supervision remains essential to prevent hallucination phenomena.
The path to fully unlocking AI's benefits in agrifood systems involves therefore several key steps.
Key among these is the need for comprehensive, accurate, and accessible data[7]. It necessitates the collection and analysis of comprehensive datasets, tracking millions of data points that influence agriculture development. Such extensive monitoring can unearth insights that lead to breakthroughs in yield improvement, resource management, and waste reduction. Overcoming this challenge of data limitations requires that we all collaborate. AI systems collect extensive data on farming practices and environmental conditions, potentially exposing farmers to risks like identity theft or the disclosure of confidential information. This risk may deter farmers from embracing AI technologies, thus hindering the scalability of their benefits. Ensuring robust data protection measures is therefore crucial to build trust and encourage the adoption of AI in agriculture.[8] The future of agrifood systems transformation and artificial intelligence will need to involve a "data commons" approach to enable AI-driven advancements in food security.
Another critical area of focus is AI governance. Ensuring that AI systems deployed in food security are managed responsibly and transparently.
The United Nations has continually addressed the digital equity gap, starting with the Secretary-General's Roadmap for Digital Cooperation in 2020. This initiative led to the formation of the Digital Public Good Alliance (DGPA) in the same year. The DGPA aims to harness digital public goods to create a more equitable world. In 2022, the Food and Agriculture Organization (FAO) joined the DGPA as a member, bringing a focus on digital agriculture and its potential to benefit all. By 2024, the DGPA, along with its members, is actively transforming various sectors and national Digital Public Infrastructures (DPI). These efforts aim to establish inclusive, society-wide digital services that cater to diverse needs.
In December 2023, the United Nations AI advisory body published an interim report that calls for a closer alignment between international norms and how AI is developed and rolled out. The central piece of the report is a proposal to strengthen international governance of AI by carrying out seven critical functions such as horizon scanning for risks and supporting international collaboration on data, and computing capacity and talent to reach the Sustainable Development Goals (SDGs). It also includes recommendations to enhance accountability and ensure an equitable voice for all countries.
The report also outlines principles for inclusive and public interest-driven AI governance, emphasizing the need for global participation and addressing digital divides. It stresses AI's governance should support diversity, equity, inclusion, and public policy goals, highlighting the importance of data governance and promoting public data commons. The principles call for universal, adaptive multi-stakeholder collaboration, anchored in UN values and international commitments, to ensure AI's benefits are equitably shared and support societal challenges, including those related to agrifood systems.
Addressing ethical considerations and building social license for the use of AI also stand out as primary concerns. AI works for the global good when it generates equitable outcomes. We need to guard against scenarios that risk of exacerbating inequalities if the benefits of AI technologies are not widely accessible. We need mechanisms that distribute the advantages of AI technologies more evenly, particularly benefiting regions in the Global South, which are disproportionately affected by food insecurity. Access to AI technologies reveals disparities. The cost of these tools may put them out of reach for small-scale farmers, exacerbating inequalities between them and larger producers. This disparity prompts a discussion on how to make AI technologies accessible to all farmers, potentially through models[9] that offer these tools at minimal or no cost.
The dependency on technology also poses a risk. Farmers relying on AI tools for planning face vulnerabilities if these systems experience disruptions or technical failures. This reliance underscores the importance of developing resilient and redundant systems to safeguard against unexpected downtimes. It is critical also to recognize that digital skills are still plagued by a gender gap. In 2019, UNESCO produced a critical policy paper “I’d blush if I could”” whose arguments remain highly relevant to the field of Artificial Intelligence (AI) and the gender gap in digital skills. This gap not only undermines efforts to achieve gender equality but could also limit the diversity of perspectives and innovation in AI development. We must never forget the necessity of concerted efforts to foster gender-responsive learning environments and to promote STEM (Science, Technology, Engineering and Math) education for women and girls. By addressing these disparities and supporting women and girls in acquiring advanced digital skills, the potential for more inclusive, equitable, and innovative AI solutions can be realized. This approach aligns with the broader objectives of ensuring that AI development benefits from diverse insights and contributes to the creation of technology that is fair, unbiased, and representative of the global population it serves.
Lastly, the environmental impact of scaling AI technologies needs careful evaluation. While AI could improve resource use efficiency, assessing the environmental implications of large-scale AI deployment and engineering an AI that is consistent with nature-based agricultural practices is vital to ensure that the push for digital agriculture does not compromise ecological health.
Bringing it together
The question of artificial intelligence has invaded public discourse and societal debates in recent months, and we find ourselves compelled to reflect on our propensity to jump on this high-speed train and surf on this promise of progress without downplaying its potential consequences in reinforcing cognitive biases, stereotypes, hallucinating (i.e. generating aberrant results) and ultimately widening the digital gap with the less privileged.
As agriculture navigates this digital transformation, a collaborative approach is necessary. Policymakers, technologists, and the farming community, all of us, need to engage in dialogue to address these challenges. Collaborative efforts among governments, private entities, and international organizations like FAO are crucial for accessible digital transformation in agriculture.[10] Policies supporting technology integration and continuous research are vital for addressing both local and global agrifood system challenges.
In this endeavor, FAO can provide a platform for Member countries to discuss these issues in a timely manner, offering a neutral platform to share insights and options to address the above-mentioned challenges. It can also build on an ecosystem of young innovators and start-ups to let sustainable and inclusive solutions to emerge, facilitating the lives of millions of farmers who need better access to advisory services. As part of our Digital Agriculture approach, it can accompany countries in their development of digital agriculture strategies and policies that integrate this new reality of AI policy and governance. FAO will also continue to scan the horizon of emerging technologies and innovations to increase our preparedness for what is coming next and offer policy options to maximize benefits and minimize challenges to reaching the most desirable future.
I am sure that by adopting an inclusive approach to AI development and deployment, it is possible to harness the potential of digital agriculture while upholding principles of equity, sustainability, and ethical responsibility.
The journey toward integrating AI into food systems transformation agriculture is more than a technological endeavor; it is a collective pursuit of resilient and equitable agrifood systems for the future".
** Article by Vincent Martin, FAO's Office of Innovation Director, originally published here.
Read more articles from Vincent Martin here.
[1] FAO Director-General QU Dongyu's quote: AI "can have a tremendous positive impact, making agriculture more productive and sustainable", but care is required to prevent it from introducing unwelcome new economic, social and ethical challenges and risks. Dr Qu made those remarks in early 2020 while making FAO, along with IBM and Microsoft, one of the first signatories of the Rome Call for AI Ethics, an initiative by the Pontifical Academy for Life to promote a sense of responsibility among organizations, governments, institutions and the private sector to create a future in which digital innovation and technological progress "serve human genius and creativity and not their gradual replacement."
[2] See: https://x.company/projects/mineral/. Among other things, the Google X project focuses on the need for crop diversification in order to expand our ability to meet global demand in a way that does not leae the planet more vulnerable pests, disease and a changing climate.
[3] See: Lowder, S. K., Skoet, J., & Raney, T. (2016). The number, size, and distribution of farms, smallholder farms, and family farms worldwide. World Development, 87, 16-29.
[4] The "Artificial Intelligence for A Digital Blue Planet" forum, sponsored by FAO in June, 2021 gathered international experts in data science and marine research. Over three days, participants exchanged insights and innovative approaches on utilizing AI to improve aquatic ecosystem management. See https://www.youtube.com/playlist?list=PLzp5NgJ2-dK72rgBePlkQoD1gMTjNEJHs
[5] Technical Platform on the Measurement and Reduction of Food Loss and Waste (https://www.fao.org/platform-food-loss-waste/flw-events/international-day-food-loss-and-waste/en)
[6] Kat Bidstrup from Think Digital in Australia discussed the potential of Extended Reality (XR) technology in addressing agricultural challenges globally, spotlighting the efforts of Big Digital Studios in applying XR to agriculture. She illustrated the benefits to veterinary students, who use XR to simulate interactions with virtual animals before actual fieldwork, enhancing safety and building confidence in managing large animals through virtual practice. See: https://www.fao.org/e-agriculture/news/webinar-exploring-immersive-technology-agrifood-extended-reality-training-and-education%E2%80%AF%C2%A0-0
[7] See https://digitalpublicgoods.net/standard/ about the importance of Open Data and Open AI Models for Digital Cooperation and achieving the SDGs.
[8] About the need for Member States to adopt clear data protection policies. See: https://www.fao.org/fileadmin/templates/legal/docs/AC2022-06.pdf
[9] There is a responsibility for International Organizations, UN agencies, Academia, Private Sector to make sure that the digital cooperation does not increase the digital divide, and that digital dividends are equally distributed. See: https://www.broadbandcommission.org/working-groups/ai-capacity-building/
[10] This is the aspiration of the International Platform for Digital Food and Agriculture https://www.fao.org/3/ni482e/ni482e.pdf