AI: an ally for the agroecological transition? AI governance for agriculture

This is an undeniable opportunity: in agriculture and elsewhere, AI can give everyone access to the knowledge embedded in data. Farmers can use AI to better understand the complex regulations governing their profession, improve their practices and develop new techniques. Modern agricultural machinery generates vast amounts of data, but the latter is often captured by manufacturers (which are frequently foreign owned). A high risk of losing sovereignty over agricultural data exists, but farmers do not always fully understand the implications of sharing this information.

Data governance

For research purposes, farmers may entrust their data to third parties such as INRAE. ‘We comply with the General Data Protection Regulation (GDPR) and  compile information over a sufficiently large geographical area so that individual data is made anonymous within a broader dataset,’ explains Hadi Quesneville, INRAE’s Chief Data Officer. However, even within public research, a clearer definition is needed of how the data of farmers is used. ‘INRAE has established a governance framework for the ethical and secure use of research data and AI. Moreover, awareness-raising, training and acculturation among farmers is essential to fostering an informed use of data and AI,’ adds Hadi Quesneville.

INRAE is a partner in the report published in February by the Blue Shift Institute, which proposes early-stage strategic solutions to support organisations.

Increasing the agency of farmers

The issues of responsibility and opacity sometimes generated by AI is also problematic. Automatic decision-making raises the issue of liability: in the event of an algorithmic error that results in an overdiagnosis or inappropriate recommendation, legal responsibility remains unclear. The lack of transparency around algorithmic ‘black boxes’ undermines user trust and can deter farmers from engaging with new technology.

The FAO Innovation Office brought these reflections on AI in agriculture to the global stage by organising a multi-stakeholder meeting in April 2025. An INRAE delegation was among the 90 researchers, consultants, representatives of the digital industry, investors and institutional stakeholders that met to discuss the issues. Three days of productive exchanges are expected to lead to a roadmap in the near future.
 

According to Isabelle Piot-Lepetit, an INRAE researcher at the MoISAjoint research unit, the adoption of AI varies depending on farmer profiles, which can be broadly categorised as follows: innovators, who develop solutions themselves; followers, who wait for their peers to test innovation before adopting them; and the hesitant, who lacks time, training and support. The cost of AI is also  a factor. Paradoxically, the technology offers highly accessible entry points for small-scale farmers and livestock producers —via smartphones, for example— but in the future may only be an option for large farms that can bear the high costs of AI infrastructure like sensors, software and training.

Across all types of farming, support for farmers in their adoption of AI appears to be a determining factor.

Agricultural advisory services therefore find themselves at the heart of these new issues. ‘AI could transform the role of agricultural advisers without replacing them. In fact, access to large volumes of data can equip advisers with knowledge and help them  counter the increasing spread of misinformation among farmers,’ notes INRAE researcher Pierre Labarthe (Agir joint research unit), while emphasising the still largely theoretical nature of these observations and the need for empirical research to better understand the impact of AI on agriculture.

For these new technological uses, living labs —participatory and innovative research laboratories— have an important role to play by involving farmers, researchers, citizens, companies and public authorities in the co-creation of solutions tailored to the complex challenges of the agricultural sector. Initiatives such as the INRAE-led Occitanum programme to mobilise digital technology for the agroecological transition in the Occitanie region are a good illustration.

The coming years will be decisive in defining the roadmap for AI to support productive and sustainable agriculture. As Carole Caranta, Deputy Director General of Science and Innovation at INRAE, confirms: ‘AI is clearly a strategic lever for enabling agriculture to meet the major challenges of tomorrow such as the agroecological transition and adaptation to climate change. As a catalyst of scientific knowledge across all fields of agronomic research, it is also a source of actionable solutions for all farmers, who can anticipate health and climate risks, optimise the use of production and management tools, and be relieved of repetitive tasks.’