AI is helping Corteva Agriscience dramatically accelerate the development of new crop protection products by modeling proteins and molecules with “unprecedented speed and accuracy,” an exec told a congressional subcommittee this morning.
Speaking at a House Environment Subcommittee hearing on innovations in agrichemicals, Brian Lutz, PhD, VP agricultural solutions at Corteva, said AI had “revolutionized discovery by allowing us to trade randomness and chance for prediction, specificity and design.”
Predicting the structure of proteins in seconds
Think of a crop protection molecule as a key intended to fit a specific lock or target site in pests, typically a protein, said Lutz.
The challenge is that we do “not always know all the locks, or proteins, or how they work, so we cannot know what to target.”
Historically, he said, it has taken “many months and often tens of thousands of dollars to understand the structure of a single protein. And there can be tens of thousands of proteins in each individual pest.”
Using an AI model able to predict protein structures, however, Corteva can now “predict the structure of proteins in seconds and for a few pennies”
As for the crop protection keys, he said, “Not that long ago, we had to rely on screening many thousands of potential molecules by applying them to pests, in hopes of finding the few that might be effective… Today, we are using AI to search this vast chemical universe for specific molecules that can interact with specific proteins within weed, insect, or pathogenic organisms to keep them from harming crops.
“Our goal is specificity: a match that will do its job and only its job, leaving the rest of the plant and surrounding environment and biodiversity intact. We recently used AI to model how 10,000 different molecules might be used in crop protection, all within a matter of weeks. This model was able to identify dozens of new potential crop protection molecules that our chemists could not have found otherwise. We are currently testing these.”
Biologicals… more quickly
New AI models are also allowing Corteva to accelerate discovery of new biologicals by “predicting the incredible diversity of biomolecules and metabolites that are produced by microbes and other organisms,” he said.
It can then be used to optimize fermentation processes to produce the biologicals: “We have used AI to both engineer the bacterial strains that drive fermentation reactions and to optimize the reaction conditions.”
Actionable insights for farmers
Finally, he said, AI is helping Corteva better understand when farmers should treat each of their fields based on data related to specific environmental conditions, predicted pest pressures and farmers’ unique management practices. This in turn enables it to provide actionable insights that can help farmers optimize yields.
“We are piloting a fungicide timing model here in America in which we combine field-specific information from our customers’ farms with our internal data. This combined intelligence then helps farmers know exactly when to spray to combat key corn diseases.”
In short, he said, “AI is transforming everything we do at Corteva. It is, without doubt, one of the most profound technologies to ever be invented.”
Tackling red crown rot…
Speaking at the same hearing, Dr. Boris Camiletti, assistant professor at the Department of Crop Sciences at the University of Illinois Urbana-Champaign, explained how AI could be deployed to address red crown rot, a fungal disease spreading rapidly across soybean crops in several US states.
“My team uses satellite imagery and machine learning to identify red crown rot hot spots in the field,” he told lawmakers. “We train the models with high resolution spectral data from visible to near infrared bands, and use ground truth observations to teach the algorithm what diseased plants look like.”
The models can detect disease areas and track disease progression over time, he explained. “We are also building tools that generate prescription maps, so instead of applying fungicides across the entire fields, farmers can target only the affected areas.”
The approach can be adapted for diseases in corn, wheat, almonds and pistachios, he said. “It’s a platform for precision agriculture that reduces chemical use while improving control. AI-driven tools give us the ability to detect the disease early, respond quickly, and use chemicals more responsibly. This is the future of crop protection.”
From drug discovery to new natural crop protection products
The third witness at the hearing, Dr. Daniel Swale from the Dept of Entomology and Nematology at the University of Florida, claimed that the US risks being left behind in the crop protection stakes, but that AI could help it catch up.
“Japan exceeds the US in the numbers of first-in-class pesticides by over two-fold, and China is near equal to the US in the total numbers of insecticides produced. And even more concerning, I personally believe the gap between other countries and the US is expanding rapidly.”
Meanwhile, the inability of the current crop protection toolkit to tackle some of the industry’s biggest challenges is a growing concern, he claimed. “The Florida citrus industry was valued at $9-10 billion a mere five years ago and is now experiencing near collapse due to an insect named Asian citrus psyllid that transmits a bacterium to the citrus tree to prevent fruit ripening.
“Yet control of this insect pest has become a tremendous challenge due to the limited number of effective insecticides available. We have developed first in class and highly effective insecticides that kill this pest. Unfortunately, the likelihood these novel chemicals will reach the market is low due to increasing regulatory restrictions as well as an expensive and prolonged developmental pipeline for synthetic pesticides.
“To remedy this, we have turned to employing AI technologies to discover chemicals in the natural world because the registration requirements for natural products are significantly lower.”
While it’s well understood that highly effective crop protection molecules can come from nature, he said, the historical barrier has been the “absence of technology capable of comprehensively understanding chemical structures and interactions at scale.”
But this is changing thanks to advances in AI, he said. “Companies such as Enveda out of Boulder Colorado have developed the capability to predict chemical structures in complex natural mixtures with unprecedented speed and accuracy.”
While Enveda has historically focused on AI-powered drug discovery, he said, “Together, we are applying this innovative AI technology to discover and develop novel agrochemicals.”
Training data: Garbage in, garbage out
That said, the outputs of AI-driven discovery are only as good as the inputs, noted Dr. Swale, “And currently, the inputs needed for agrochemical discovery remain poorly understood. To date, agrochemical discovery and early phase development has been restricted to the private sector, which do not share data sets due to intellectual property restrictions.”
This, he claimed, has led to gaps in public knowledge needed for appropriate AI inputs, and has limited the utility of AI for discovery of new agrochemicals. “So how do we address this challenge? I believe AI enables public entities such as universities to play a key role for the very first time in the discovery of agrochemicals. This is due to their ability to test fundamental questions oftentimes not addressed by the private sector.”
Involving the public sector in agrochemical discovery “will allow fundamental data sets and AI inputs to be public,” he claimed. “A commitment in fundamental science by the federal government is critical for the US to remain at the leading edge of innovation and advancement.”
The role of government in data collection and access
However, recent moves by the Trump administration to cut public funding into such research could thwart such progress, warned Congressman Gabe Amo (D-Rhode Island). “Innovation often begins with research at public universities funded by federal dollars. But these investments are being systematically eroded by the Trump administration.
“The president has paused billions of dollars in federal grants to research institutions and universities that’s going to cause delays to critical work, destabilizing programs and jeopardizing the very pipeline of talent and discovery that fuels our innovation economy.”
He added: “Scientists at public universities throughout the country are working right now to develop AI models and data science that will make agriculture more efficient. The massive commercial and industrial opportunities available in agriculture today are only possible because of these fiscally responsible investments, but the Trump administration is not that interested in this research.
“They’re trying to slash the National Science Foundation’s budget by $4.9 billion. That’s a 55% cut to our country’s steward of basic research. It is the very agency that supports the underpinnings of agriculture technology, including AI.
“We all suffer without federal investment in science to address challenges in agriculture and climate. Innovation may be the topic today, but the foundation is science, and right now, that foundation is crumbling beneath our feet.”
Public data under threat
Several speakers at the hearing also expressed concern over the Trump administration’s moves to defund data collection efforts across several government departments.
Dr. Lutz at Corteva noted: “We depend on the many data sets that the government has supported over the last several decades to be able to drive innovations for farmers. The fungicide timing models that we have been developing rely on climate data that we get from the public sector.”
Congressman Amo added: “Dozens of important data sets, reports and services have been thrown out the window over the past several weeks due to what I believe to be reckless actions, actions that will be problematic and cause irreparable harm.
“Without accurate and transparent forecasting and climate modeling, farmers cannot react and plan ahead. No algorithm is better than the data that it runs on, and if we let politicians dismantle the very systems that provide the data farmers use to determine when to plant, water, apply pesticides, and harvest, we’re setting ourselves up for failure.”
Further reading:
From jumping genes to designer genomes: Anthology’s blueprint for smarter bioproduction strains
Nestlé bets big on AI and biotech with new deep tech center in Switzerland
The post Corteva: AI can transform crop protection to replace ‘randomness and chance’ with ‘prediction, specificity and design’ appeared first on AgFunderNews.
