AI, agriculture and the future of food


“So far nobody seems to have asked the question, ‘Are there any risks associated with rapid deployment of agricultural AI?'” said Asaf Tzachor, a researcher at the Center for the Study of Existential Risk at the University of Cambridge. in a press release.

The potential benefits are enormous. Increases in agricultural productivity could help feed the approximately 2.4 billion people around the world who are food insecure and malnourished, and revolutionize the way farmers use their land.

That could get expensive. The analysis points to potential flaws in the agricultural data that power AI-powered systems and the possibility that autonomous systems could prioritize productivity over the environment. This could result in unintended errors leading to over-fertilization, hazardous use of pesticides, inappropriate irrigation or erosion, jeopardizing crop yields, water supplies and soil. And large-scale crop failures could exacerbate food insecurity.

Cyber ​​security is another potential point of failure. The researchers said cyberattacks could disrupt entire food systems. The more dependent agricultural systems are on intelligent machines, the more disruptions could arise if they fail or are destroyed.

Then there are people – and without inclusive technology, the researchers warn, AI could simply exacerbate already existing inequalities in agriculture. If large farmers benefit, small farmers in the Global South, for example, could be left out of agricultural profits altogether.

Potential solutions mentioned by the researchers include data sharing, citizen input, and digital “sandboxes” where developers can predict potential failure points for agricultural AI.

“Technological modernization in agriculture has achieved a lot,” the researchers write. But irresponsible developers may “ignore and thereby perpetuate drivers of food insecurity, labor exploitation and environmental resource depletion.”

Responsible artificial intelligence in agriculture requires a systemic understanding of risks and externalities

Nature Machine Intelligence


Comments are closed.