4 minute read
AI Transition in Agriculture
Paula Madetzke
When I was looking into AI driven transitions for energy, I came across another field that was interesting and would like to explore further: agriculture. As the world population grows while the climate changes and makes farming more unpredictable, the need for smarter agriculture could not be greater. Agriculture production tripled between 1960 and 2015 (Intel), but along with this steep increase in production, came deforestation, intense pesticide usage, and other environmental problems that can not be sustained for the next wave of production. Luckily, AI can be used to solve many of the emerging challenges of the future of farming by reducing the number of human farmers, increasing yield efficiency, and minimizing the negative environmental impacts of agriculture.
Agricultural work can be back breaking and difficult to make a living. As nations industrialize and more opportunities outside of farm labor become available to more people, there will be fewer people who want to harvest and maintain crops. AI can both eliminate the need for as much human labor on farms, and make the remaining work less physically taxing on those who remain. One area of transition is with smart harvesters of non-cereal crops. In the case of strawberries for example, it would take a human examining a plant, knowing whether the berry is ripe, and picking it without harming the rest of the plant. With RGB cameras and trained AI, the harvester is able to make the determination of how ripe the berry is by color and where the robotic arm should reach to pick it. Existing forms of these technologies are less accurate than humans at the moment, but can go through fields much faster and reduce the need for human labor. However with continued research and development, these harvesters are likely to become more viable replacements for human labor.
Perhaps the most important area of advancement of AI in agriculture is with yield efficiency. As the climate changes, there will be less arable land and more destructive weather, all while the global population grows and requires more food. One major example of increasing efficiency is with predictive analytics. From year to year, predictions of weather patterns from AI can help farmers make timing decisions and prevent yield loss from inclement weather patterns. In the longer term, entire growing zones will change. Predictive climate models in addition to soil information could help farmers adapt what they grow and how to a new set of weather patterns. Another way to improve efficiency is to use drones AI that is able to detect problems to monitor the crops. On large farms, this would not only reduce the need for as many humans to monitor the fields for problems, but an aerial drone could scan the fields much more quickly than humans on foot. This would allow farmers to address problems when they are presumably still small, thus preventing a significant loss of yield. The AI aided surveillance would not only help with efficiency, but also the environmental impacts of farms.
Fertilizers and pesticides have allowed modern farms to produce far higher yields than they would otherwise be able to. However, the environmental impact from the runoff of these chemicals when they leave the farm can be incredibly detrimental to both human and natural wellbeing. Not only that, but farming continues to be one of the largest consumers of fresh water. AI assisted diagnostics can help reduce the need for blanket uses of chemicals and water so that less runoff is released into the environment, and less clean water is taken from it. AI trained to monitor crops can detect pests, disease, and weeds. This allows farmers to be able to “spot-treat” problem areas of their fields rather than needing to blanket the entire field with chemicals that could harm humans or the larger environment. AI powered soil monitoring could help farmers know exactly what and how much fertilizer is needed to produce maximum yields while reducing runoff. Soil, plant and predictive weather analytics could also help farmers determine how much additional water is needed for their crops to grow and when, so that they are able to have a minimum impact on reservoirs without sacrificing yield.
The incoming challenges to farming in an era of climate change, a growing population, and avoiding further damage to the environment from runoff are formidable, but AI can be a powerful tool for meeting these challenges. For the purposes of this class, there are a fair number of open source datasets for agriculture with a wide range of specific applications. I am excited to delve deeper into this important and exciting subject.