The Risks of Using Artificial Intelligence to Grow Food

The use of artificial intelligence in agriculture in the future poses huge potential risks to farms, farmers and food security, but little is known about them, according to a new risk assessment published recently in the journal Nature Machine Intelligence , has not received enough attention.

Imagine a swath of farmland sprawling to the horizon where wheat is grown, harvested and processed into flour and made into bread for the urban population. Imagine again that all the power to till, plant, fertilize, monitor and harvest this field is delegated to artificial intelligence: controlling drip irrigation systems, self-driving tractors and combine harvesters. That’s smart enough to react to the exact needs of the weather and crops, but then a hacker can screw things up.

A risk assessment analysis recently published in the journal Nature Machine Intelligence warns that the future of artificial intelligence in agriculture could pose huge potential risks to farms, farmers and food security that are poorly understood. Not even getting enough attention.

The first author of the study, Dr Asaf Tzachor from the CSER Centre at the University of Cambridge, said the idea of ​​running a farm with artificial intelligence was not science fiction, and large companies were already developing the next generation of autonomous robots and decision support systems, which will be gradually introduced in agriculture. replace manpower. But so far, no one seems to have asked the question, “Are there any risks to rapidly deploying agricultural AI?”

While AI holds great promise in improving crop management and agricultural productivity, the underlying risks must be truly addressed. New technologies must be properly tested in real-world environments to ensure they are safe from unexpected failures, unintended consequences, and cyber-attacks.

In this Cambridge University study, the research team presents a series of risks that must be considered in the development of artificial intelligence in the agricultural industry, and ways to address such risks. According to the risk assessment, cyberattacks could “hijack” artificial intelligence to wreak havoc on commercial farms by poisoning datasets or shutting down sprayers, drones and harvesters. To prevent this, “white hat hackers” are needed to help uncover any security holes in the software development phase, reducing the chance of being attacked by real hackers.

An AI system that only delivers optimal crop yields in the short term could ignore the environmental consequences of achieving yield targets, leading to chronic overuse of fertilizers and soil erosion, the report noted. Excessive use of pesticides in pursuit of high yields can poison ecosystems; excessive use of fertilizers can pollute soil and surrounding waterways. The report recommends involving agroecologists in the design process of AI technologies to ensure these situations are avoided.

Automation and driverless driving can improve working conditions for farmers and free them from manual labor; however, without inclusive technology design, socioeconomic inequalities currently entrenched in global agriculture, including gender and ethnicity, will persist discrimination, etc. Agricultural AI systems that do not take into account the complexity of the workforce will ignore and may continue to exploit vulnerable communities and fail to fundamentally solve agricultural problems.

We have seen the deployment of various robots, advanced machinery and sensors in the agricultural industry to collect crop information and support farmers’ decision-making, such as detecting disease or irrigation conditions, and some agricultural machinery and equipment can even operate without a driver. work autonomously. Such automated systems are designed to increase agricultural efficiency, save labor costs, optimize production, and minimize losses and waste. While this can help farmers increase their incomes, it can also lead to increased reliance on AI.

However, the majority of the world’s farms, and the ones that feed large swathes of the southern hemisphere, are smallholder farmers who may be excluded from the benefits associated with AI. Market marginalization, low internet penetration and the digital generation gap may prevent these small farmers from using the most advanced agricultural technologies, widening the gap between commercial farms and subsistence farmers.

An estimated 2 billion people worldwide suffer from food insecurity, of which about 690 million are malnourished and 340 million children are deficient in micronutrients. In the face of climate change and population growth, artificial intelligence and precision agriculture technologies are expected to bring substantial benefits to food and nutrition security. However, when deploying such technologies at scale, we should closely consider the potential risks and aim to understand and mitigate these risks early in the technology design.

Originally published at https://www.tlw.com.