Global blind spots in air pollution public opinion: an LLM-assisted mapping of research coverage and survey availability

Abstract

Abstract Air pollution is a global health crisis, but evidence on public perceptions of pollution exposure and mitigation demands remains fragmented. We first map "what research covers" by reviewing the survey-based literature on public views of air pollution, coding 456 studies through a transparent, LLM-assisted screening and classification workflow. Then, we map "what surveys offer" by cataloging 112 air-quality survey items in cross-national survey programs, coding them manually using the same classification framework. We find persistent geographic and exposure-level biases: survey-based research focuses on high-income countries, China and India, while cross-national item coverage is more clustered in Europe. Thematically, both domains emphasize perceptions, with gaps in behavior, policy preferences, and health—particularly in high-exposure regions. These findings highlight that where mitigation is most needed, research coverage is most limited and accessible survey data most scarce. Methodologically, the paper develops a validated LLM-assisted workflow to facilitate future large-scale literature mapping and classification.