The key facts

  • 32% of 995 surveyed young adults in the US turned to AI chatbots for information about drugs and alcohol
  • 12% use AI for such queries weekly, 4% daily or almost daily
  • 93% of those who asked AI about substances also used Google search, meaning chatbots complement rather than replace search engines

What the survey showed

A study published in JMIR Formative Research found that 32% of 995 US residents aged 18 to 29 had used generative tools such as ChatGPT, Google Gemini and Meta AI to look up information about drugs and alcohol. The online survey was conducted in April 2026. Among those who turned to AI on such topics, nearly half do so at least once a month: 12% weekly, and 4% daily or almost daily.

Why young people ask AI

The main reason for turning to AI was to learn about the risks and consequences of substance use: nearly 93% of those who used AI on this topic cited it. About half sought information about specific substances, including questions about dosage, while 35% were interested in topics related to substance use disorders, such as addiction. Notably, trust in AI specifically on substance-related questions turned out to be even higher than overall: the average reliability rating was 3.66 out of five, compared with 3.51 for general use.

Who turns to AI more often, and what comes next

The study revealed demographic differences: Black respondents had adjusted odds of using AI for substance-related queries 2.52 times higher than white respondents, and men roughly twice as high as women. The study's lead author, Rutgers University postdoctoral researcher Mireya Trigueros Roura, warned in a press releasethat such tools 'may lead to inaccurate conclusions and other risky patterns of substance use among young people,' and linked the disparities to potentially unequal access to healthcare, stigma and trust. At the same time, the data suggest chatbots are not displacing familiar sources: 93% of those who asked AI about substances also used Google search for the same overall purpose. The authors note that the accuracy of chatbots' answers to such questions has not yet been evaluated — this, along with the impact of the information obtained on behavior, they name as tasks for further research.