Locked inside, away from the heat, humidity and buzz of summer cicadas, sits the future of food security. Terabytes of 1’s and 0’s on hard drives are being mined by artificial intelligence as it searches years of data for patterns and insights useful in seed selection.
“We’re using data science to accelerate plant breeding,” says Morrison Jacobs, global commercial data science lead for Crop Sciences at Bayer. “We can leverage years of genetic performance, germplasm performance data, together with the genetic data to accelerate the pipeline and predict the performance of potential products.”
An Accelerated Timeline
Jacobs says today, variety selection that used to take five to six years can be done via artificial intelligence (AI) in four to six months.
“We can use machine learning or artificial intelligence to prescribe new possible genetic combinations and anticipate how a new plant variety will perform in thousands of climates and soil conditions, because we’ve observed them and logged the data,” he says.
“Historically, what we’ve done is used data to look backward at what has happened, but that doesn’t tell you what is going to happen,” says Mike Tweedy of Pattern Ag.
This new AI revolution uses things like soil, climate and genetic data to anticipate outcomes and turn that information into actionable insights.
“Using AI and some of these predictive models, we can bring insights to the grower or an adviser and say, ‘look at this specific practice on this specific field’,” explains Andy Folta, U.S. product marketing manager for Climate FieldView at Bayer. “Farmers want an output for the data they’ve been collecting for years or decades.”
“One of the models that came up when we were going through BASF’s Xarvio platform told us to spray for white mold two weeks ahead of time,” says Cumberland, Iowa, farmer Nick Christensen. “By the time you saw it in the field, you would have needed to spray two weeks prior.”
“It’s turning that information into a recommendation without having to go through five years of trial work on farm,” explains Scott Kay, vice president of BASF.
Jacobs says these field-level recommendations work in concert with ongoing research aimed at delivering better results to farmers. Machine learning is now sorting through 2.6 million different compounds matching efficacy to unique proteins in different organisms, be them pests or weeds. It’s an approach Bayer is calling CropKey.
“Think about it as a lock and you’re trying to find the key,” Jacobs describes. “Today our scientists can design entirely new and precise keys using virtual screening, which increases the number of molecules we can analyze exponentially.”
It’s how they discovered Icafolin, a new crop protection mode of action, and the first new postemergent herbicide in 30 years. It’s set to be released in Brazil in 2028 before expanding to other countries.
“We now have 30 potential new molecules targeted under investigation, and we’re validating 10 more modes of action,” Jacobs says. “AI is powering our discovery pipeline.”
These new data tools will go to work for global farmers using solutions for the future from the future.
Your Next Read — First Look: Kelly Hills Unmanned Unveils Massive Made In The USA Spray Drone
