TOPLINE:

Generative artificial intelligence (AI) tools were used for mental health advice by 13% of US youth in a new cross-sectional study, with even higher rates for those aged 18-21 years. Findings also suggest that more than 90% of all users found the AI advice helpful.

METHODOLOGY:

• The nationally representative cross-sectional study included 2025 data for more than 1000 English-speaking individuals (age, 12-21 years) with internet access, from RAND’s American Life Panel and Ipsos’ KnowledgePanel (37% aged 18-21 years; 50% female; 51% White, 25% Hispanic, and 13% Black).
• The survey examined participants’ use of generative AI for emotional support when feeling sad, angry, or nervous; the frequency of seeking advice; and perceived helpfulness. All result percentages were weighted.
• Factors associated with AI use, including age, sex, race, and ethnicity were assessed.

TAKEAWAY:

• Nearly 13% of all participants, representing a weighted sample of approximately 5.4 million individuals, reported using generative AI for mental health support, with higher rates (22%) among young adults aged 18-21 years.
• The likelihood of engaging with this mental health advice was significantly higher among those aged 18-21 years than among younger adolescents (adjusted odds ratio [aOR], 4.0; P < .001).
• Black users reported lower perceived helpfulness of AI-generated mental health advice than non-Hispanic White users (aOR, 0.15; P = .01).
• Overall, 65.5% of users reported monthly or more frequent engagement with AI for mental health support, and 93% reported the advice as being somewhat or very helpful.

IN PRACTICE:

“High use rates likely reflect the low cost, immediacy, and perceived privacy of AI-based advice, particularly for youths unlikely to receive traditional counseling,” the investigators wrote. 

“However, engagement with generative AI raises concerns, especially for users with intensive clinical needs, given difficulties in establishing and using standardized benchmarks for evaluating AI-generated mental health advice and limited transparency about the datasets training these models,” they added.

SOURCE:

The study was led by Ryan K. McBain, PhD, MPH, RAND, Arlington, Virginia. It was published online on November 7 in JAMA Network Open.

LIMITATIONS:

The study was limited by its modest sample size, potential nonresponse and survey response biases, the lack of information about specific generative AI platforms used and the types of advice sought, and the absence of measures of diagnosed mental illness. The tightly-focused cohort limited the generalizability of the results. 

DISCLOSURES:

The study was funded by the National Institute of Mental Health. Four of the 10 investigators reported having financial ties with various sources, which are listed in full in the original article. The other researchers reported having no relevant financial relationships.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.