Roughly one in three young adults in the United States turns to generative artificial intelligence chatbots like ChatGPT for information regarding drugs and alcohol, according to a new study by researchers at Rutgers University. Published on October 19, 2026, in the journal JMIR Formative Research, the investigation evaluated survey responses from 995 U.S. residents between the ages of 18 and 29. The researchers aimed to understand how frequently young people consult synthetic text generators for sensitive health inquiries and how much they trust the resulting outputs.

The findings indicate that 32.3% of respondents—321 out of 995 participants—have used generative AI to look up substance-related details. While young users rated the automated tools highly for helpfulness and reliability, public health experts warn that relying on unregulated algorithms for substance guidance carries notable risks, as chatbots can generate inaccurate medical details.

How Young Adults Query Chatbots About Drugs and Alcohol

The research team, led by postdoctoral researcher Mireia Triguero Roura and co-authored by associate professor Pia M. Mauro at the Center for Pharmacoepidemiology and Treatment Science, recruited participants online through the platform Prolific in April 2026. Within this sample, general familiarity with artificial intelligence was nearly universal, with 97.3% (968 of 995 respondents) reporting prior use of generative AI for any purpose. This sits alongside broader survey data cited by Rutgers showing that 66% of young adults use AI to search for general information or obtain medical advice.

When narrowing the focus specifically to drugs and alcohol, the investigators found that inquiries were not merely one-off experiments. Among the 321 participants who sought substance-related details from AI, 48.2% reported doing so on at least a monthly basis.

The primary motivation for these queries centered on safety and health outcomes. According to the study data, 92.8% of the participants who used AI for substance information were looking up the risks and consequences of substance use. Respondents also reported prompting chatbots for details about specific substances and substance-use disorders.

Demographic Patterns and Healthcare Access

The survey data revealed distinct demographic patterns in who chooses to consult chatbots about drugs and alcohol. Specifically, Black respondents and male respondents were statistically more likely to report using generative AI for substance-related inquiries.

While the cross-sectional survey notes an association rather than a direct cause, the study authors point out that these patterns highlight real-world gaps in traditional clinical support. Triguero Roura noted in the university's October 23 announcement that the disparities in who turns to AI for drug-related inquiries may reflect underlying differences in healthcare access, social stigma, institutional trust, and general health information-seeking behaviors.

For young adults facing barriers to medical care or fearing judgment from family members and clinicians, an anonymous digital interface offers an immediate, private alternative. Across the group of substance-information seekers, respondents reported high levels of perceived trust, reliability, and helpfulness regarding the AI tools they used.

What This Means for Public Health and Safety

For parents, educators, and healthcare providers, the gap between young adults' high confidence in AI and the verified safety of chatbot outputs is a central concern. Generative AI models predict text patterns to formulate responses; they do not possess clinical judgment and are known to occasionally reproduce medical misinformation.

"Using these tools can lead to misinformed or otherwise risky substance-use behavior among young people, who are the age group most affected by adverse effects of substance use," Triguero Roura stated in the Rutgers release.

At the same time, secondary analyses emphasize the boundaries of what this specific survey can prove. While the Rutgers summary highlights the 32.3% prevalence rate, independent reporting by Remio and Superpower Daily notes that the study relied on self-reported data from an online panel designed to approximate the U.S. young-adult population, rather than a formal national census or probability sample. Furthermore, the researchers did not audit the actual text of the chatbot answers for clinical accuracy, nor did they track whether the advice altered the participants' real-world substance consumption.

Consequently, while the study confirms that chatbots have become a routine source of drug and alcohol information for many young adults, it does not demonstrate actual physical harm or verify whether the advice received was safe. Triguero Roura and Mauro emphasize that follow-up investigations are needed to evaluate how young people interpret AI-generated drug and alcohol outputs in practice.

Frequently Asked Questions

What types of substance questions were most common in the study?

The vast majority of substance-related AI users—92.8%—reported asking chatbots about the risks and consequences of substance use. Participants also queried the systems for details regarding specific substances and substance-use disorders.

Did the study prove that AI substance advice is dangerous or inaccurate?

No. While the researchers warned that AI chatbots are capable of reproducing misinformation that could lead to risky decisions, this specific survey only measured self-reported usage habits and user attitudes. It did not evaluate the clinical accuracy of the chatbot responses or measure whether users changed their behavior afterward.

Why might certain young adults prefer asking a chatbot over a doctor?

The researchers suggest that differences in who uses AI for drug inquiries—such as higher usage rates among Black and male respondents—could be linked to variations in healthcare access, fear of stigma, differing levels of trust in medical institutions, and personal habits around seeking health information.

Sources