Researchers in the US have used AI-powered camera traps to recognize and document bee species, validating a low-cost alternative to traditional pollinator research methods. Scientists at Oregon State University built the traps from commercially available components, costing $100-$200 in total each, housed in weatherproof containers with solar panels and batteries. The cameras were positioned at a fixed focal distance, aimed at colored platforms serving as non-rewarding visual lures. Over 18 days, the units recorded continuous time-lapse frames during daylight hours (approximately 06:00-19:00), capturing footage of 6 bumblebee species, closely matching previously documented diversity. Some 70,000 images were fed into two AI analysis models: over 6,000 original-resolution, full-frame images were analyzed by one model, and over 66,000 tiles from cropped frames were analyzed by the other. The AI identified and classified species, flagging and discarding images without insects, which reduced raw field data by over 90% without human effort. Experts then reviewed only a small subset of pre-filtered, cropped images to confirm species-level identification, suggesting that continuous multi-day field monitoring can be scaled feasibly.