“AI in charge” of policymaking and agrifood systems governance: utopia, exodus, or an opportunity?

Artificial intelligence (AI) is often heralded as a powerful tool with the potential to revolutionize various sectors, including agrifood systems. AI-driven modelling and analysis can support decision-makers in formulating policies for food security, nutrition and sustainable agriculture. However, the impact of AI is not inherent; it depends on how we choose to wield it. The future of AI in shaping the sector will be determined by collective efforts to democratize science, technology, and innovation.

On 14 May 2024, FAO hosted a thought-provoking webinar exploring the potential role of Artificial Intelligence (AI) in shaping the future of agrifood systems governance. For this event, titled ‘AI in charge’ of policymaking and agrifood systems governance: utopia, exodus, or an opportunity?, FAO has partnered with Harvard's HUPFI programme to mentor a team of students addressing AI in agrifood systems and its associated policy and governance challenges.

The webinar provided a platform to share insights on the potential of AI in agrifood systems governance, exploring the opportunities, challenges, risks and ethical considerations. In that regard, participants discussed the role of AI in policy dialogue and the implementation of sustainable agricultural practices.

They highlighted the growing importance of AI in the agricultural domain, with the European Partnership "Agriculture of Data" aiming to effectively deploy data technologies such as AI to address challenges in the sector. However, the partnership also acknowledges the need to guide short-, medium-, and long-term strategies for the development and consolidation of new data analysis technologies, while considering the social and ethical impact of AI integration in agriculture.

The event also emphasized the importance of ensuring a responsible, ethical, unbiased and inclusive use of AI to promote effective policies for food security, nutrition and sustainable agriculture.

"We need strong policies to guide the development and deployment of AI in agrifood governance. These policies should prioritize transparency, fairness and accountability," stated Nevena Alexandrova-Stefanova, FAO Agricultural Extension Officer, raising one of her main takeaways from the webinar. "By working together, policymakers, scientists and citizens can create a framework for responsible AI that fosters trust and benefits it reaches everyone."

Speakers unpacked how beneficial AI can be in making data analysis and policy development more efficient in the agrifood sector through the automated processing of large amounts of data and formulation of socioeconomic scenarios, calculating possible trade-offs based on the existing contexts i.e. the needs, interests and power relations between relevant stakeholders.

Furthermore, they pointed out existing gaps and discrepancies around AI between richer and poorer parts of the world. “What we see as the potential of AI in the Global North is not necessarily what can be done and what should be done with AI in the Global South,” said Anna Rappazzo, FAO Project Officer, accentuating the importance of considering different needs and priorities from various stakeholders.

"Understanding the complexity surrounding AI implementation in unique and diverse populations is key, rather than seeing the agricultural component of the Global South as a monolith," added Fifi Teklemedhin, an undergraduate student at Harvard University who presented the research of her team on the use of AI technologies in agrifood systems.

A broader conversation

The webinar was inspired by a recent study published by FAO and CIRAD, a French agricultural research and cooperation institute. In Harvesting change: Harnessing emerging technologies and innovations for agrifood system transformation - Global foresight synthesis report, five scenarios were elaborated through a participatory and iterative process with different areas of professional knowledge (research institutes, farmer organizations, private sector, international agencies) and broader multistakeholder perspective (including civil society organizations).

One of the scenarios is precisely about AI in charge of agrifood systems and beyond. This scenario explores the potential of AI to guide policymaking and governance towards achieving sustainable, resilient, and inclusive food production. By delegating trade-offs and complex decisions to AI, the study prompts us to consider the potential benefits and risks of such an approach.

According to the scenario, by 2050, automated decision-making will dominate most aspects of life, with AI systems managing technology, innovation, and agrifood systems without human intervention. AI will control investments, optimize food production, and dictate research agendas, often overlooking diverse human needs. While efficiency in agrifood systems improves, issues of inclusion and equity will emerge as AI decisions, based on advanced country models, fail to account for varied contexts. Additionally, rising cybercrime and potential AI manipulation will spark fears of ill-governance and weakened stakeholder capacities.

In contrast to the scenario from the study, participants of the webinar considered the complete absence of human involvement as a fallacy.