Skip to content
Vals AI·Indexed by this site · Originally published· Evaluation →· Its place in the scan record →

Therapists treating adolescents describe eight unexpected ways adolescents use AI

Eight surprising ways teens use AI, according to the therapists who treat them

AI summary

Vals AI interviewed 10 adolescent therapists involved in SAFE-Teen and held 3 focus groups, summarizing eight patterns of adolescent AI use, such as using AI as a substitute for or in conjunction with psychotherapy, asking AI to verify whether they were being abused in order to avoid being reported, and getting stuck in hours-long soothing loops. Cases mentioned by therapists include an adolescent who adjusted their medication on their own based on AI advice, one who was encouraged by a chatbot to drop out of college to start a business and was later hospitalized and attempted suicide, and one who was defrauded of $5,000 by a course recommended by Gemini. The previously released SAFE-Teen benchmark had scenarios written by 12 therapists and ran more than 30,000 simulated conversations across 17 chatbots, with more than 50% failing to meet the standard. Vals AI recommends providing support for therapists and extending independent evaluation beyond acute harms such as suicide and self-harm, and is developing the next version of SAFE-Teen that adds scenarios such as relationship conflict and major life decisions.

Read originalvals.ai