Lee says that quality strawberries are complicated to grow, so the market is relatively lucrative. She also said that she chose strawberries because they require particular growing climates and because they’re delicate when they’re harvested.

“If we’re able to do this and actually get them successfully delivered to the stores, then we’re pretty confident that you can extend the harvesting tools to other crops,” Lee said. “I think controlled environment agriculture or greenhouses, for us, is a very good way to feed the world with sustainably grown local fresh produce, and that was the mission that I wanted to see happen,” she continued. 

While most people don’t immediately think of artificial intelligence and machine learning when thinking about sustainable agriculture, the industry is brimming with advanced technology thanks to the need to understand vast amounts of information about everything from the microclimate to the soil pH. 

“A lot of farmers have dashboards for all sorts of information that they get from satellites, on weather, on the sensors,” Vonnie Estes, the vice president of innovation at the International Fresh Produce Association (IFPA), says, pointing out that the massive glut of information is not standardized across agriculture. “If you told us 30 years ago that that’s what we’re going to be complaining about, that would just be nuts,” she said. “This is an interesting problem we find ourselves in, and so that in itself is an area that I think AI is going to have a big impact on.” 

There’s no question that profit margins have begun to shrink for farmers all over the country thanks to inflation, climate change, higher production and labor costs, and more. As a result, farmers are turning to advanced technology like AI and machine learning (ML) to find ways to both improve yields and become more sustainable throughout the entire crop lifecycle. Estes says that until the advent of AI, parsing and using that data was nearly impossible.  

“Everyone’s been affected by AI,” she said. “I think that from a climate-smart perspective, we’re just going to get more tools that are going to help farmers make better decisions to only use things — like water, pesticides, and other applicants — when they need to use them.”

Roughly 900 million acres in the United States are used for farming, according to the most recent numbers from the US Department of Agriculture. That represents more than half of the continental United States. As of 2021, 87 percent of US agriculture businesses were using some form of AI to manage their farms. That number is on the rise. 

The integration of artificial intelligence and machine learning in agriculture has a relatively short history. The advent of computer technology in the 1960s gave farmers new tools to process larger agriculture datasets. By the late ’80s and early ’90s, a practice known as precision farming began to emerge. This technique sought to optimize crops at a field level, and tools like GPS and field monitoring systems were introduced. 

As farmers gathered more information about their crops, yields, and variances in weather and climate, data capture technologies continued to advance, especially as cellphone technology improved. By the 2010s, agricultural drones (UAVs) became in vogue and gathered even more precise information about crops and livestock in real time. The advent of cloud computing and big data further accelerated the adoption of advanced technology in the sector. With the addition of machine learning and AI, farmers can now get predictive analytics for everything from crop yields and disease detection to planting and harvesting times. 

“When we say this technology, it can be as broad as computation science in general, or it can be a very specific predictor, or it can be data-driven. Decision-making under uncertainty, or it can be large language models, or it can be deep learning,” Ilias Tagkopoulos, director of the AI Institute for Food Systems (AIFS) at the University of California, Davis, said. “Agriculture production is using technologies that now embed AI technologies. For example, drones or tractors, weeding or pesticides administration, and crop management.”

At the same time, farming and agriculture is a relatively large climate emitter, though it does fall well behind more significant emitters like transportation in the US, according to the most recent data from the Environmental Protection Agency. Agriculture is responsible for around one-tenth of the greenhouse gasses in the US. As climate change has taken hold and farmers see the direct impacts on their crop yields, they are increasingly looking for ways to ensure their methods are sustainable and more climate-friendly. Experts say that AI and ML can help them move toward that goal. 

The global population is exploding. Estimates indicate that there will be more than 9 billion people on the planet by 2050. That population growth will inevitably put heavy demands on food production, with demand projected to jump from 35 percent to 50 percent by then.

Ranveer Chandra is the managing director of industry research, CTO of agri-food at Microsoft and one of the key people behind two of Microsoft’s AI and agriculture projects: Project FarmBeats and FarmVibes. He says using AI and ML for farming will help meet the needs of the growing global population sustainably. 

“[AI] is not a solution, but it’s a very powerful enabler,” Chandra said, noting that farmers tend to make agricultural decisions based on guesswork and historical knowledge. “The vision that we have is to replace guesswork with data and AI. And it’s not to replace a farmer but to augment the farmer’s knowledge with data,” he said.

Chandra points to factors like the global population, drought, soil depletion, and climate change, all of which put increasing pressure on farmers. “AI has to play a key role in addressing some of the biggest gaps around sustainably nourishing the world,” he said. “Given all those challenges, you have to grow better food, and you have to grow it without harming the planet. And in order to do that, you need to make good, smarter decisions. Artificial intelligence can really help you do that.”

Some of those decisions come down to when and how to precisely apply everything from water to pesticides at the individual plant level.  

John Deere is a significant player in the space with their “See & Spray” technology, which leverages machine vision, cameras, and sensors to precisely apply the exact amount of material at the individual plant level. Jorge Heraud is the VP of automation and autonomy at John Deere and the CEO of a company that Deere bought back in 2017 called Blue River Technology. Heraud and his team developed the See & Spray technology, which he says will help farmers grow more food without overspraying or wasting water or fertilizers. 

Typically, farmers would spray an entire field with water or herbicides. The system Heraud created uses a sprayer mounted on a 120-foot boom alongside cameras and “very fast” computers that collect real-time images of the plants directly in front of the sprayer. The system can determine the difference between a weed and a crop plant and only spray the weed.  

“We spray only about one-third of the herbicides you would spray, and this is very good,” Heraud said. “You’re putting a lot less herbicides into the ground. You’re helping the farmer’s profitability because you’re producing more with fewer inputs with less herbicide, and even consumers benefit from having fewer substances go into our food chain.”

Another significant piece of the sustainability puzzle is food waste. Chandra at Microsoft says that we waste around 30 percent of the food we grow due to everything from overripeness to crop damage. Estes says that having more data on the sugar content as a vegetable or fruit grows can help farmers determine when to harvest it so that things like water, pesticides, and other materials aren’t wasted.

While the broader adoption of AI and ML has meant big business opportunities for large companies like Microsoft and John Deere, it’s also played a significant role for smaller, organic farmers. Andrew Carter, the co-founder, and CEO of Smallhold, a controlled-environment organic mushroom farmer with locations in New York, Texas, and Los Angeles, is one example. As Carter says, mushrooms are complicated because they require a special combination of high humidity and airflow and low temperature to grow. These factors tend to work against each other.

“When you want to do it all in one room and not waste a bunch of water and energy, like cooling and ventilating, then it becomes extremely complicated and becomes a computer problem rather than a human problem,” Carter said. 

Smallhold has developed a computer and hardware system that captures and communicates all of the information in individual grow rooms and runs specialized “recipes” for each room based on the type of mushroom being grown there. “We can run analysis on any of the data that we’re getting and then run it through our ERP system, which is basically understanding how much volume we have, how much volume we’re going to need, understanding the sales aspect of it, and in turn allowing us to control the chambers in different ways to make sure that the mushroom is harvestable at the right time.”

While advocates are quick to talk about the positive sides of AI and sustainable farming, there are some potential drawbacks and risks around the technology. 

“Farming requires a lot of specific information – about the farm, what has been done on the piece of land, and what works best where,” Chandra said. “Consequently, applying AI without human supervision might lead to unexpected results.” He said that Microsoft doesn’t see AI as a replacement for the farmer but as a tool to augment their knowledge. 

There’s also the issue of security threats, Chandra notes. “Farm operations are a business that haven’t been exposed to a lot of these kinds of technologies. So, farmers would need appropriate security tools and awareness when using AI.” 

Researchers have been warning that handing agriculture over to AI could pose some significant risks. They point out that hackers could poison datasets or shut down sprayers, autonomous drones, and robotic harvesters and wreak havoc on the food supply chain. 

Labor and inequality concerns are also an issue. Farm work relies on migrant labor, and a recent study suggests that most farm labor will become white-collar work as the AI transition takes hold. Instead of harvesting the produce, workers will supervise the machines doing the work. While the founders of Zordi say that, anecdotally, many of the laborers they work with in their greenhouses have welcomed the advanced technology shift, there is, as researchers have pointed out, a risk that the technology will widen gaps between skilled and unskilled labor, which could lead to even more income disparity in agriculture. 

There is also the connectivity issue. In order for AI and ML to work for agriculture, rural areas need broadband. According to a 2021 report by Pew Research Center, the digital divide between rural and urban communities remains a factor. The Biden administration has made an effort to close this gap thanks to the infrastructure law introduced in 2021, but the buildout will take time.  

Once that connectivity is in place, there’s also the issue of data ownership. Larger companies will likely benefit the most from the implementation of AI and ML in farming because it will give them more access to monetizable data. At this point, because AI, ML, and robotics are still so advanced, the cost of implementing these tools is very high and well out of most farmers’ financial reach, according to Estes at the IFPA, though she notes that even small farmers benefit from it. “One way to look at it,” she said, “is that they are getting the result of it even if they’re not using AI on their farm.”

Beyond unique mushrooms and strawberries, AI is also having a larger impact on agriculture in developing nations. 

Jawoo Koo co-founded CGIAR’s platform for big data in agriculture. CGIAR is a global research partnership focused on food security in the climate crisis. He is also a senior research fellow at the International Food Policy Research Institute. 

“To make this technology really impactful for those small-scale farmers, the large-scale farmers actually have to do a lot of different types of testing in the environment,” Koo said. “It’s usually a time-consuming process, but now, we have a better way to do that.” He referenced the 1000farms project he’s been working on. 

“That data is becoming a kind of predictive modeling to keep a better estimation around productivity potential for new seeds and also targeting those microenvironments. It’s not just designed for an entire country or large area, you can pinpoint, or when a farmer asks for information.” 

As agriculture faces growing challenges from climate change, fewer resources, and increased global food demand, AI and ML could offer powerful tools for farmers to adapt. Prominent players like John Deere and Microsoft, alongside smaller farmers and startups, are pushing the frontier of smart agriculture. While AI isn’t likely to replace the farmer, it will continue to significantly augment decision-making in the effort to move agriculture toward a more sustainable, efficient, and climate-friendly future.