Exposure to content written by large language models can reduce stigma around opioid use disorder

Widespread stigma, both offline and online, hinders harm reduction efforts in the context of opioid use disorder (OUD). This stigma targets clinically approved medications for OUD (MOUD), people with the condition, and the condition itself, among several others. Given the potential of artificial intelligence in promoting health equity, this work examines whether large language models (LLMs) can abate stigmatizing attitudes in virtual healthcare communities. To answer this, we conducted a series of randomized controlled experiments, where participants read LLM-generated, human-written, or no responses to help-seeking OUD-related content. The experiment was conducted under two setups: participants read the responses either once (N = 2, 141) or repeatedly for 14 days (N = 107). Participants reported the least stigmatized attitudes toward MOUD after consuming LLM-generated responses. This study offers insights into strategies that can foster inclusive discourse on OUD. Based on our findings LLMs can serve as an education-based intervention to promote positive attitudes and increase people's propensity toward treatments for OUD.