Weeds continue to be one of the most sustainable problems in agriculture. But the biggest problem facing modern farmers is not removing weeds. Mechanical tools and herbicides can do that. Instead, it has difficulty identifying and killing weeds without damaging the crop.
Paul Mikesell is a company that manufactures AI-powered robots for the agricultural sector and is the founder of Carbon Robotics, former director of infrastructure engineering at Uber. He has developed an AI system that attempts to solve the big weed problems over the past six years. His company's solution is the LaserWeeder G2, a machine that automatically detects weeds and zaps Zaps with laser arrays.
Mikesell told Business Insider that neural networks are important “not only to find where weeds are, but also to find the best place to kill them.” Neural networks are computational models inspired by how the biological brain learns to process information, and are key to the functionality of modern AI systems.
Throughout the agriculture industry, AI tools are beginning to make a difference for farmers. That's good news for industries struggling with enemies such as the climate crisis and trade changes. From complex robots to chatbots, farmers are testing a variety of tools to reach out of reach and achieve their goals.
Machine learning becomes the field
Mikesell's experience helped me build a self-driving car infrastructure with Uber. It helped shape the carbon robotics approach to agricultural AI, and applied the same technology to the agricultural tools he is currently developing.
Computer vision systems used in autonomous vehicles such as automobiles, tractors and other agricultural equipment often rely on neural networks known as convolutional neural networks.
CNNS is a form of neural networks that can be trained to detect patterns in images. Carbon robotics uploads weed images to its own database. There, human labeling manually identifies weeds and crops. These image label pairs are used to train CNNs to discover weeds that can detect weeds using the onboard camera of the laser weeder and the computer hardware of the machine itself. This means there is no need for an internet connection.
John Deere, the world's largest agricultural equipment company, uses CNNS for multiple applications, including looking at autonomous tractors and spraying weed detection systems. At CES 2025, the company presented its new second-generation “autonomous kit.” This can partially or fully automate common tasks such as cultivation and weed removal.
Sarah Schinckel, the company's director at Emerging Technologies, said AI is already improving its agricultural equipment. In 2024, she said John Deere's Sea & Spray System was used to spray over a million acres of farmland. The machine sprays only plants identified as weeds, so the system was able to weed this area using 8 million gallons, which are less herbicide than is normally needed.
“If you think about that savings and improving overall productivity and sustainability for them, that's just a victory for them,” Singkel said.
The technology also increases staffing flexibility for farmers. For example, semi-autonomous harvesting equipment provides human operator AI assistance that allows for faster adjustments to the equipment than a typical operator can respond. “Perhaps we can put people who are combining experts to put operators in taxis, helping them to achieve high performance,” Singkel said.
Farmers launch chatgpt
Large farmers are building tools with complex CNNs and other types of machine learning, while some farmers are using more accessible AI tools. Phillip Guthrie, a partner at agricultural consulting firm Nine Creeks Consulting, often offers presentations on new agricultural technologies, including generator AI. He already watches farmers pick up ChatGpt for planning and advice.
Guthrie recalled a conversation with a farmer who had trouble with the data analytics platform he used to monitor and track the weather on the farm. Analytics never worked correctly for the operation, so “he took raw weather data, threw it into ChatGpt and started analytical.” AI was able to handle analytical tasks that previous software couldn't handle.
Guthrie hopes that more farmers will begin using generative AI tools in equally specific and creative ways, and will likely bypass companies creating specialized Agri-Tech software tools.
Two visions of generation AI in agriculture
AI technologies such as CNNS, which are currently available in autonomous agricultural equipment, represent a major leap in technology. Systems like the Laserweeder G2 and John Deere See & Spray were impossible to imagine ten years ago.
However, it is unclear how these task-specific examples of agricultural AI will fit into the new generation AI tools.
Mikesell speculated that one solution could be in integration. Carbon robots like John Deere do not use the generator AI for their equipment and have not announced plans to do so. Still, he said generative AI could become a “plan-human interface” used to operate equipment like the company's automatic laser weeds.
“We want to clear this 2,000 acres into our generated AI system,” says Mikesell. “So, why not come with a solution and deploy these laser weeds in this pattern?”
Meanwhile, Guthrie believes generative AI can drive “democratization” of industries that large companies may overlook. The industry always needs heavy equipment, but farmers often express their dissatisfaction with the expensive yet extremely specific software available to the industry. “The last thing they need is another tool to do one thing. All they want is a tool to do everything,” he said.
Guthrie said that he is a generative AI that constantly improves, “There will be farmers who can build their own tools, carry out their own analyses, do their own automation, and focus on what they want for themselves.”
“It changes the way agriculture is for me.”
