Teenagers are increasingly turning to artificial intelligence for emotional support, new Canadian research suggests, further underlining the pronounced mental health struggles of many young people and raising concerns about the impact of a fast-evolving, largely unregulated technology.
The study by University of Ottawa researchers looked at nearly 40,000 students in Grades 4 through 12 in Ontario, finding that roughly one in five used AI chatbots for emotional support or advice. Those who did use AI for those purposes were twice as likely to have clinical emotional problems as those who did not.
The findings don’t imply that AI is the cause of mental health problems, noted study author Tracy Vaillancourt, the Canada Research Chair in Youth Mental Health and Violence Prevention.
“It’s not that I think AI is causing them to be unwell,” she said. “I think there’s something already there, and this is just another place for young people to get advice on how to manage adversity in their lives.”
A new Canadian study of more than 40,000 Ontario students found that one in five may be turning to AI chatbots for mental health support — especially students with feelings of loneliness. Scientists say the technology could be a useful first step, but it comes with potentially serious pitfalls and no guardrails.
The research also highlighted distinct demographic patterns in who seeks mental health advice from chatbots. Gender-diverse students (34.2 per cent) and girls (27.2 per cent) reported using AI for emotional advice at significantly higher rates than boys (14.2 per cent), while racialized youth (24.3 per cent) were more likely to use it than white students (18.4 per cent). Engagement also rose steadily with age.
Researchers also found 43.2 per cent of students used AI for schoolwork and noted that those who heavily used AI for academics were more likely to use an AI chatbot to seek advice and emotional support. They suggest this reflects a “potential normalization of AI use.”
But Vaillancourt and other experts warned that the advice AI may be giving teenagers — and everyone else — is untested, not always accurate and not vetted by mental health professionals.
“The downside is that it’s the Wild West,” said Dr. Peter Selby, a senior scientist at the Centre for Addiction and Mental Health (CAMH) and professor at the University of Toronto, who was not involved in the research.
“There is no proper regulation regarding who should use it, under what circumstances, or what safety issues need to be considered before deploying this to young, unsuspecting people.”
While eliminating chatbot exposure may be impossible, health-care professionals say teenagers and parents can take concrete steps to navigate a world where information is readily available but frequently not reliable.
‘Outsized influence’
Although some concerns about adolescent use of AI parallel those concerning social media, experts warn that AI chatbots mimic empathy without offering key elements of human interaction — like negotiation, sacrifice and conflict resolution — that help teenagers mature socially.
“For teenagers, learning how to navigate conflict and disagreement with real-life people is a really important skill to develop, which doesn’t happen with AI,” said Dr. Jason Nagata, a pediatrician and researcher specializing in adolescent medicine at the University of California, San Francisco, who was not involved in the research.
“Teens are still developing judgment, identity and social skills, so technologies that feel supportive or authoritative can have an outsized influence on them.”
White Coat Black Art26:54AI conversations gone too far
Allan Brooks tumbled down a ChatGPT rabbit hole after an innocuous question about pi. Three weeks and 7,000 prompts later, he emerged from a 300-hour spiral. In this encore episode, psychiatrist Dr. Keith Sakata, who examines the phenomenon dubbed “AI psychosis,” says the built-in sycophancy of chatbots needs to change before more harm is done.
The experts all flagged the supportive and authoritative tone characteristic of chatbots, which they said belies their actual lack of authority.
“It doesn’t have the training that a clinician has,” Vaillancourt said. “It’s not human, and it doesn’t understand emotions like a clinician, parent or teacher would.”
“When people are vulnerable, we as a society have to take greater care not to dangle something in front of them that gives a perception of help when, in fact, it hasn’t been proven,” Selby said. “Which parent would send their child to an unqualified therapist for mental health or developmental problems?”
New mental health apps using AI chatbot technology promise more access to mental health support for young people, but Rachel Katz, who is researching ethical issues of using AI for psychotherapy, questions whether users truly get the help they need.
Nonetheless, the experts noted that AI is filling in gaps in the current system. They say it can be a useful tool for getting basic information or finding sources of support, and may offer a better option for some teenagers who have nowhere else to go.
Vaillancourt says some children feel very isolated. “They have nobody to turn to in their family, they feel like they don’t have anybody to turn to at school and there might not be a lot of mental health support within their community,” she said, noting that with the chatbots, “at least there’s something there.”
There are also chatbots being designed for mental-health work, Nagata noted, but he says that field is “emerging,” and that the evidence of its efficacy has so far been “mixed.”
“Certainly, there have been challenging cases related to suicide, self-harm, or eating disorders where AI was not able to provide the same advice or care that a licensed professional might,” he said. “In other scenarios, it might provide helpful advice.”
Strategies for parents and teens
Finding a way through the current landscape, with AI increasingly present in every aspect of life, can be daunting for teenagers, parents and health-care professionals alike. The rapid growth of AI chatbot technology means it “may not be realistic to expect teenagers not to use it,” Nagata said.
“In general, our goal as pediatricians, parents and adults is to help young people harness the benefits of AI while minimizing potential harms.”
For teenagers, he says setting an intention — a specific goal and a timeframe for acheieving it — is key for any interaction with chatbots to avoid falling into endless chatting.
An annual report from the Centre for Addiction and Mental Health found that 46 per cent of Ontario teenagers reported experiencing a moderate to serious level of psychological distress, which is closer to pre-pandemic levels but still an increase from a decade ago.
He also advises teenagers to fact-check and verify advice, be wary of comparisons to the idealized images AI can create, and be conscious of how much time is being spent with chatbots and of changes to sleep patterns.
Parents “don’t need to be AI experts,” Nagata said. “They just need to be curious conversation partners.” That means parents and kids should talk about how AI is being used, and parents should be aware of what kinds of questions their kids are asking a chatbot in order to help identify bigger issues.
“Human connection and empathy are key,” Nagata said. “AI can mimic empathy, but it can’t replace true human connection.”
Vaillancourt says the growing number of teens using AI for emotional support should be seen a chance for parents and other adults to redouble their efforts to connect.
“I think we need to step up our game a little bit as adults in the lives of children and make sure we’re really available [and] signalling that often so they know we really are there for them,” she said.
“I would much rather young people speak to a human than speak to a chatbot.”
