New research finds there is no substitute for real-world observations, but that — under the right circumstances — AI-generated images can be used to improve the performance of computer models used for species identification and biodiversity monitoring.

In a Policy Forum, Noam Angrist and colleagues discuss a science-backed research agenda for understanding how to successfully implement educational programs at scale. “The frontier of education research and practice is to improve implementation,” Angrist et al. write. “There is growing evidence on what works – now we need to translate that evidence into action and systematically study the process, so that we can do it repeatedly and at increasing scale to achieve effective, equitable, and enduring learning for all.” Although most children around the globe attend school, learning outcomes remain low in many parts of the world. A growing body of research has identified programs that can improve learning, yet interventions that succeed in small, controlled studies often lose effectiveness when implemented at scale and by governments. According to the authors, a major obstacle is that education research rarely measures how faithfully programs are implemented or how widely they are adopted, making it difficult to determine whether poor outcomes result from flaws in the program itself or from shortcomings in its implementation. Here, Angrist et al. call for stronger implementation science in global education, modeled partly on efforts in health care, to close the gap between research discoveries and real-world practice. The authors propose a four-stage framework to turn educational research into effective, equitable practice at scale: efficacy trials, effectiveness trials, policy plans, and large-scale implementation. This approach emphasizes testing whether interventions work across diverse populations and real-world settings, as well as whether they are affordable and sustainable. “We hope the implementation science for global education agenda motivates a shift toward research that informs the decisions and challenges that confront implementers and government decision-makers,” the authors write. “By ensuring that more science makes it into implementation and that more implementation is made scientific, implementation science can support the ultimate goal of all children learning.”

Rivers may be delivering far more microplastic pollution to the oceans than previously realized, with developing countries accounting for nearly all of the global flow, according to a new study. The worldwide analysis also finds that heavy rainfall can unleash surges of microplastics into rivers – a climate-sensitive pollution threat that could intensify as extreme weather becomes more common. Microplastics have become a pervasive and long-lasting form of pollution, with recognized threats to ecosystems, water quality, and human health. Each year, tens of millions of metric tons of microplastics enter the environment, and much of that material ultimately reaches the oceans. Rivers are a major pathway for transporting these particles from inland regions to coastal waters. However, estimates of global riverine microplastic pollution vary widely because of inconsistent sampling methods and limited understanding of how rainfall and runoff affect their transport. Here, Hehao Qin and colleagues developed a new framework to produce more reliable, high-resolution estimates of microplastic concentrations and export from rivers. According to the authors, their approach harmonizes differences in particle size and sampling methods, while using machine learning to account for both human activity and natural processes. It also tracks how rainfall can either increase or dilute microplastic concentrations. Using this approach, Qin et al. estimate that rivers worldwide delivered roughly 263,000 tons of microplastics to the ocean in 2022, with 96% originating from the Global South – Southeast and East Asia alone contribute to more than half of this total. Plastic use, poorly managed waste, and human development were the strongest drivers of geographic differences in microplastic concentrations, while weather and seasonal conditions influenced short-term fluctuations. The findings show that extreme rainfall events can substantially increase the movement of microplastics into rivers, particularly in regions with high plastic use and poor waste management. Qin et al. warn that climate change could worsen this problem as extreme rainfall becomes more frequent and intense, potentially creating short-lived but highly concentrated pollution pulses into rivers and marine environments.