By John Lovett
Agricultural scientists are accelerating discoveries with artificial intelligence to move toward more sustainable and resilient agriculture. But there are limitations not including human imagination.
The inaugural AI in Agriculture Symposium, hosted Sept. 15 by the Arkansas Agricultural Experiment Station, brought more than 150 people together, in-person and online, to hear how AI and machine learning are propelling research in multiple agricultural disciplines, from animal science to biology and robotics. The experiment station is the research arm of the University of Arkansas System Division of Agriculture.
“AI and machine learning are indeed pushing the boundaries of agricultural and biological research,” said Aranyak Goswami, a computational biologist and assistant professor with the experiment station. “It allows us to integrate and analyze enormous datasets from poultry pathogen genomes to cattle gut microbiomes and plant multi-omics and uncover patterns that traditional methods would simply miss.”
For example, Goswami added, AI models can link genetic variation to disease outcomes or build biological “clocks” that estimate gut maturity in livestock which provide insights with real-world applications in animal health and productivity.
While AI excels at handling scale and complexity, Goswami noted that choosing the right model still depends heavily on human expertise and imagination.
“We need to decide which models make sense for a given biological problem and also ensure that computations are efficient enough not to overwhelm even high-performance computing systems,” Goswami said. “In that sense, AI does not replace human reasoning but rather amplifies it giving us new tools but still relying on scientists to frame the right questions and interpret the answers responsibly.”
Goswami is a member of the experiment station’s Center for Agricultural Data Analytics and is affiliated with both the animal science and poultry science departments for the Division of Agriculture and the Dale Bumpers College of Agricultural, Food and Life Sciences at the University of Arkansas.
Making connections
Rich Adams, assistant professor of agricultural statistics within the entomology and plant pathology department, explained how his work integrates statistical learning and comparative biology to investigate how genome structure and function change over time, especially in agricultural and ecological contexts to better understand, for example, plant and insect interactions.
His research group is developing novel methods to better predict evolutionary associations among traits, offering new insights into the mechanisms that drive biological complexity across diverse organisms. Adams said their new estimators demonstrate promise in this area for guarding against statistical outliers and other analytical challenges that complicate genomic analyses.
To complement this work, Adams’ team is also harnessing machine learning advances to study protein diversity shared among plants, fungi, insects and microbes providing a window into the ways these organisms interact and the broader consequences of those interactions.
Source : uada.edu