Is A.I. Planting its Seeds in Crop Protection and Sustainable Agriculture?

Machine learning can make agricultural companies turn greener by speeding-up pesticide development, allowing for localized treatment of infested crops, and minimizing the use of resources

Everyone agrees that pesticide companies are evil: they purposely kill thousands of defenseless living being every year, they keep developing more and more aggressive poisons, and they make millions with this business model. There’s even the hashtag #monsantoevil! They must be bad, right? However, if you look at the top 10 companies responsible for environmental violation, none of the “Big 6” Biotech Corporations -Monsanto, BASF, Bayer, Dow Chemical Company, Dupont, and Syngenta- is on the list. Even weirder, you may have heard about Monsanto betting on A.I. to become greener. Maybe the Lords of the Harvest are not as bad? Maybe #monsantoNOTevil and neither are machines?

Machine learning can make agricultural and agrochemical companies turn greener

Pesticides, herbicides, and preservatives exist for a reason: to reduce plant and crop diseases. And some of them are used in organic farming, too. Despite the chemicals currently available, approx. 1016% of global harvest is lost to plant diseases every year. In addition, another 10–12% is lost each year in the post-harvest system. Crop disease losses are estimated to cost $220 billion. What’s more: every year, 48 million Americans suffer from illnesses related to foodborne pathogens. That’s why agrochemical companies continuously develop new systems against pests, pathogens, viruses, and weeds.

However, nobody likes the process of developing a new preservative, pesticide or herbicide. Not even crop, plant, and wood protection companies. Developing a new crop protection product costs between $250 million and $286 million, according to Monsanto and Syngenta. What’s more, it takes a decade or more for a crop protection product to reach the market. The same numbers apply to wood preservatives, as the head of wood protection R&D at BASF Wolman, Joerg Habicht, commented: “I think a general range of around 12 years (plus or minus 4 years) for a wood preservative is a fair estimate for 80% of all cases”. This is because researchers not only have to develop the correct formulation, but also test it in the lab and in the field. And if, at the end of the tests, the product is not as effective as expected, the researchers have to go back to synthesize a new formulation and test it.

1. Pesticide & herbicide development with deep learning

What A.I. can do is speed up the initial phase. Rather than assessing and ruling out molecules one by one as possible active ingredients or whole formulations, deep learning algorithms can predict the likelihood that a given active ingredient or formulation will have the desired protective effect on crops, plants, and wood. This is called in silico testing because the investigations take place in the machine. The AI process is similar to what pharmaceutical companies use to develop new drugs and medicines.

Don’t miss our next blog post on AI in drug discovery: subscribe to our newsletter!

As the experimental part is heavily reduced by the software, the costs for running the experiments shrink too. What’s more, the process becomes greener as fewer resources and chemicals are used.

But there’s another area where Monsanto and agricultural companies can apply A.I. and become more sustainable.

2. Detection and localized treatment of infested areas with computer vision

There’s something else nobody likes to do in agriculture: to use (and pay for) the full amount of pesticide or herbicide formulation when you can use half of it, a tenth, or even less. Awhere, Blue River Technology, and Gamaya took the leap in A.I.-based precision agriculture. Their products provide farmers with information on what’s going on in their crop fields -no matter how large they are- and empower them to produce more while using less. How? With drones and machines with brains: the drones fly over the field and collect images, which are then processed by machine learning algorithms. These latter can identify issues and potential threats, such as places where irrigation is needed, and areas infested by pathogens. In this way, the farmers can treat their fields more selectively, without treating crops and areas that do not require any action. In See & Spray, Blue River’s product, the cameras are also located on the agricultural machinery itself and the release of pesticides or herbicides happens instantly. In addition, in all solutions, weather or satellite information can be combined with the system, making the whole process more holistic. Even in this case, technology can reduce the resources and chemicals used, lowering chemical costs by 80% and reducing global herbicide use by 1 billion kilograms, as Blue River estimated. In addition, these innovations will optimize the farm’s productivity and reduce harvest losses.

Those are the biggest A.I.-applications we at Balzano have noticed in agriculture. They clearly show how A.I. can make agrochemicals eco-friendlier and the Lords of the Harvest less evil; we don’t need to be technological utopians to see that. What other innovations do you see in the field? I’d love to hear all about them in the comments!

Originally published at