Cross-Sectional Survey Finds 32% of Young Adults Use Generative AI for Substance-Use Queries

Rutgers investigators fielded an online survey assessing frequency, demographics, and perceptions of generative AI use for drug and alcohol queries. Roughly half of AI users engaged monthly or more often. Black and male respondents showed elevated odds of AI reliance, patterns the authors link to possible differences in healthcare access, stigma, and information-seeking behavior. Respondents rated AI outputs as highly trustworthy and helpful despite known risks of hallucinated content.

The study design is observational and cross-sectional, so it documents prevalence and correlations but cannot establish causation or track behavioral outcomes such as changes in consumption. It also lacks verification of actual AI responses received by participants. These limitations leave open whether AI exposure increases risky use or simply reflects preexisting distrust of traditional sources.

Prior work on digital health information, including 2023-2025 studies in JAMA Network Open on chatbot accuracy for medication queries, shows error rates exceeding 20% for dosing and interaction advice. Combined with the Rutgers disparities, the findings suggest targeted outreach to high-use groups may be needed before AI becomes a default information channel.

The authors call for longitudinal studies that test how young adults interpret AI-generated substance content and whether exposure alters help-seeking or use patterns. Such trials would require outcome measures beyond self-reported trust.