Fuente: PubMed "pollination" Curr Opin Insect Sci. 2026 Jun 18:101564. doi: 10.1016/j.cois.2026.101564. Online ahead of print.ABSTRACTUnderstanding the causes, consequences, and solutions to global pollinator decline will require more extensive and intensive monitoring programs. However, species-level identification remains a major challenge because of the difficulty of scaling manual identification workflows. Participatory science (PS) programs generate millions of pollinator sightings with associated images, but spatial and taxonomic biases along with expert capacity limit their full potential for conservation science. In this paper, I explore how artificial intelligence (AI), particularly computer vision-based detection and classification models, can be integrated with PS to enable scalable, reliable pollinator monitoring, with a focus on bees. Recent advances demonstrate that AI image classifiers can achieve high accuracy across hundreds to thousands of taxa and can be deployed on web, mobile, and edge device platforms. AI has the potential to substantially reduce expert workloads while maintaining reliability when carefully integrated into expert verification pipelines. Such pipelines can include confidence-based filtering based on quality or priority and improved models using contextual Bayesian priors, model calibration, and ensemble approaches. However, uneven training data, observational biases, and limited image availability for rare species constrain model completeness. Addressing these gaps will require targeted image collection, integration of museum and field datasets, and standardized protocols linking AI development with monitoring objectives. Rather than replacing taxonomic expertise, AI should function as a force multiplier that accelerates feedback between data collection, model improvement, and conservation.PMID:42315082 | DOI:10.1016/j.cois.2026.101564