Thanks to artificial intelligence and digital twins, technology is redesigning agriculture to make it more efficient and sustainable.
Open a perfect cob, without parasites and with firm kernels. It seems simple, but behind that gesture there is a bet that every farmer makes against time, drought and disease. There are many variables and include hailstorms or parasites, but if some unknowns cannot be controlled, man has tried to minimize the risks with technology.
While this innovation effort has its roots in time, the arrival of artificial intelligence (AI) is opening a radically new chapter in this age-old challenge. «When farmers buy seeds, they must be reasonably sure that the production will be abundant and of good quality», explains Jean-Luc Pellet, Geneticist Product Manager at the Bayer Crop Science Research and Development Center in Olmeneta (CR), where the corn hybrids that will populate crops throughout the Mediterranean are designed and selected.
«Maize is the most used crop in the world (in 2019 there were almost 1,100 million tonnes, followed by 734 of wheat and 495 of rice, ed.). And every year we try to produce new seeds that must have the best possible characteristics to adapt to different soils, latitudes and weather that is increasingly unpredictable due to climate change.”
Seed designer For some time in agriculture, hybrids have been created that come from parental lines of different seeds to try to combine their genetic characteristics (resistance to diseases, productivity, etc.), but if this was previously done manually, today technology comes to the rescue.
«Based on the genetic information of plants, which we obtain from the fields, we provide artificial intelligence with data on hundreds of thousands of varieties. With this data, AI makes virtually millions of combinations of hybrids to design new ones,” says Pellet. «For example, the AI, by crossing line A with line B, tells us that we will obtain line C, which estimates that it will produce 150 quintals per hectare instead of 140».
The complication lies in the fact that each plant has many characteristics, which add up to a single seed, and they must be combined together to obtain the most effective result possible.
«Not only is the yield evaluated, but also, for example, the strength», continues Pellet, «that is, the ability of the plants to stand upright in the face of extreme climatic events and this can be achieved for example with the innovative short-statured corn.
And also the health of the grain, i.e. its ability to resist fungal attacks, which are responsible for the production of aflatoxins (substances harmful to humans) in the cob. Furthermore, we try to make the crop more sustainable, creating hybrids that require less fertilizer with nitrogen input, usually used to maximize yield, because this substance has a high carbon footprint. Finally, we try to produce plants with delayed senescence, capable of producing until the end of their development.”
Digital twin All these virtual hybrids are then tested in a digital twin, a digital model with over 50 terrain characteristics, which simulate real terrain in different countries and at different latitudes, to see what results they will produce, also based on climate forecast data. “Through millions of interactions, AI helps us create a supply of seeds with a wide genetic variety to meet the different needs of different farmers,” adds Pellet.
Another significant challenge to crops is disease, weeds and pests, something that can bring any farmer to their knees. The current goal is to develop highly selective treatments, designed to act specifically on the target organism and limit the impact on beneficial species and the environment. Even in this case, however, AI comes to help. “We use it to design products with unprecedented precision and sustainability,” explains Rachel Rama, Head of Small Molecules at Bayer Crop Science.
“AI analyzes billions of molecular combinations in virtual chemical spaces and predicts which ones will bind to a parasite’s target protein. This allows you to create chemicals designed for a specific parasite, reducing the collateral impact on beneficial species and the soil, ensuring their correct degradation in compliance with sustainability standards. In practice, AI first identifies the “lock”, i.e. a protein present in the parasite, then digitally experiments with millions of virtual “keys”, i.e. molecules, to see which one adapts best. Only the best-designed molecules then move on to synthesis and testing in the real world.”
By doing so, the aim is to create new chemical products that are more targeted and have a better sustainability profile. The first of these, a herbicide scheduled for introduction in 2028, will use a new mode of action to control post-emergence weeds in extensive crops; it is the first new mode of action marketed in the last thirty years.
«This approach is currently also in use for the development of new fungicides and insecticides, with the aim of increasing their effectiveness and reducing their environmental impact at the same time», concludes Rama.
Let the farmer know… The high complexity and the growing number of extreme situations that were rarely present in the past make the decision-making phase for agronomists and farmers particularly difficult.
To support experts in this area, AI-based assistants are being developed. “These are large, specialized language models based on proprietary agronomic and product datasets,” explains Robson Monastier, Global Agronomic Digital Innovation Lead at Bayer Crop Science. «You can ask complex questions, such as how to choose the most suitable hybrid for specific soils, climates or yield objectives, or how to manage problems such as diseases, pests or weeds. In these cases, our model analyzes thousands of technical documents, product labels, research summaries and information on local trials in just a few seconds, to provide effective answers based on concrete evidence.”
In a world that changes so quickly, artificial intelligence does not replace experience; becomes a “wise advisor” who has studied millions of cases and can suggest the best choice. The final objective returns to the starting point: to win the bet that began with that single seed, to bring safer and more sustainably grown food to everyone’s tables.
