JMIR Publications

ABSTRACT

Background:

Generative artificial intelligence (AI) tools such as large language models (LLMs) are increasingly used to produce plain language summaries (PLSs) of scientific research. However, automated summaries often overlook local context and cultural relevance—limitations that are critical in environmental health research, where affected communities face disproportionate exposure and health risks.

Objective:

This study aimed to develop and refine community-informed prompts for generating plain language summaries of environmental health research using generative AI.

Methods:

A community-engaged research design was used to co-create prompts through iterative workshops with three Louisville, Kentucky–based groups disproportionately affected by industrial pollution (N=97). Participants reviewed AI-generated PLSs of peer-reviewed studies on volatile organic compounds (VOCs) and cardiovascular outcomes. Feedback on three prompt styles—varying in length, format, and reading level—was analyzed thematically to identify preferred structures, language, and contextual framing.

Results:

Participants consistently favored summaries that (1) began with definitions of technical terms, (2) separated sections for key findings, introduction, and conclusion, and (3) included environmental justice implications. The final co-created prompt produced concise, sixth- to eighth-grade–level summaries with improved readability and perceived trustworthiness. Participants described these outputs as clearer, more relatable, and more actionable than generic AI summaries.

Conclusions:

Community-informed prompt development improves the accuracy, cultural resonance, and accessibility of AI-generated summaries. Embedding community perspectives in generative AI workflows offers a replicable model for equitable research translation in environmental health and beyond.