Yu Fang stood in a sunlit field at the University of California, Riverside, watching honey bees dart between flowers—except for the ones near the treated honeycombs. Those they avoided completely. It was the real-world proof the team had been waiting for: 130 chemical compounds, flagged by a machine learning model, could safely steer bees away from danger.
Honey bees are essential to global food production, pollinating everything from almonds to apples. But they’re also vulnerable, especially to pesticides used in farming. Even legal, targeted sprays can harm bees when they forage on treated plants. Colony collapse has become a growing crisis, and scientists have been searching for solutions that protect both crops and pollinators.
Enter Anandasankar Ray and his team. Instead of trying to remove pesticides entirely—a challenge for modern agriculture—they asked a different question: Can we keep bees away from treated areas without hurting them? The answer came from an unexpected place: artificial intelligence.
Ray’s team built a machine learning model trained on how bees respond to smells. With over 200 odor receptors, bees have an incredibly sensitive sense of smell—great for finding flowers, but hard to repel. The model analyzed chemical structures and known bee behaviors, then screened more than 50 million compounds. From that massive pool, it narrowed the list to about 130 promising repellents.
"Bees rely heavily on their sense of smell to forage, but that sensitivity makes it tough to find odors that push them away instead of drawing them in," Ray explained. "Our goal was to flip that script and find a way to use scent as a deterrent—safely and effectively."
In lab tests, bees avoided the top candidates. Then came the field trials. Seven compounds were tested on freely foraging bees. Every single one worked. Bees steered clear of honeycombs laced with the repellents—and showed no signs of harm.
The implications go beyond farms. In cities, hospitals, or office parks, these scents could discourage hive-building in unsafe places. For seedless crops like certain grapes or cucumbers, where pollination isn’t wanted, the repellents offer a precise tool.
Ray envisions a future where pesticides come blended with bee-safe repellents—protecting both plants and pollinators. "Protecting pollinators doesn't have to come at the expense of food security," he said. With further testing and development, these AI-discovered scents could become a standard part of sustainable farming.
For now, the team’s work stands as a hopeful sign: sometimes, the best way to protect nature is to listen to it—and then gently guide it in a safer direction.