Affective generative AI use associated with psychological distress in youth. - Psychology Today

AI Use and Emotional Problems in Children and Teens

Affective generative AI use associated with psychological distress in youth.

Posted September 25, 2026 Reviewed by Tyler Woods

The use of artificial intelligence (AI) is rising rapidly among children and adolescents. As this trend rises globally, it is important to understand the relationship that AI may have on the mental health. A new large-scale study published in JAMA Pediatrics discovered an association between affective generative AI use and mental health problems in youth.

“Seeking emotional support from generative AI is a unique marker of psychological distress in youth, suggesting that clinical and digital literacy frameworks must distinguish between functional AI assistance and the use of algorithms as an emotional refuge,” wrote Kamilla Bonnesen, Amanda Krygsman, Sarah Hobson, Daphne Korczak, and Tracy Vaillancourt who authored the research led by the University of Ottawa.

What is Affective Generative AI?

Affective AI, also known as affective computing, was a term coined in the mid-1990s by MIT professor Rosalind Picard, an American scientist, entrepreneur, and AI pioneer in affective computing. Picard defines affective computing as “computing that relates to, arises from, or deliberately influences emotions.”

Generative AI, or GenAI, are machine learning models that can create generated content such as text, code, music, audio, images, or video when prompted. Examples of generative AI include ChatGPT, Claude, DALL-E, Gemini, Midjourney, and Perplexity.

The combination of generative AI and affective computing may increase user engagement with more natural and emotionally appealing interaction. This becomes an increasing concern as AI use among youth skyrockets globally.

Young people are using GenAI frequently, and the numbers are rising. The use of GenAI among youth is common. Roughly a third of American children from the ages of four to 17 years used GenAI apps on their devices according to a University of North Carolina at Chapel Hill study published in earlier this year in February. Across 20 European countries (United Kingdom, Ireland, Germany, Austria, Belgium, Spain, Switzerland, Slovakia, Norway, Finland, Luxembourg, Italy, Malta, Latvia, Estonia, Portugal, Poland, the Czech Republic, Croatia, Serbia), 72% of children surveyed reported having used generative AI according to the EU Kids Online 2026 report.

Beyond Homework Help

More alarmingly, what is also rising is the use of AI by youth for emotional support. According to the University of Ottawa researchers for their new study, students using AI for schoolwork increases the likelihood of using it for emotional support purposes. Understanding the prevalence of youth using AI for functional tasks is important as it can increase the potential use for emotional support.

“Our results also reveal an association between functional and affective engagement: the more students used AI for school-related tasks, the more likely they were to also report affective AI use, suggesting a potential normalization of AI use where more frequent use is associated with lower barriers to entry for intimate emotional self-disclosure,” the University of Ottawa researchers reported.

How Big is the Potential Problem?

Ever since the release of ChatGPT to the general public by OpenAI in November of 2022, the use of AI, particularly large language models (LLMs), has been widespread and growing. In North America, AI use by youth is common and increasing. AI has been used in some form by 86% of American children between the ages of nine and 17 years old, according to a 2026 Common Sense Media report. In Canada, an estimated 73% of Canadian students depend on generative AI for their homework, which is a significant increase from 52% in 2023 according to a 2025 KPMG Canadian survey.

The widespread use of AI among youth extends beyond North America. In a UNICEF survey spanning 10 countries (Mexico, Brazil, Jordan, Montenegro, Armenia, Colombia, Pakistan, Macedonia, Serbia, Dominican Republic), on average, 60% of children polled reported using AI for homework help. Globally, youth are rapidly adopting AI into their daily lives, and it is not just for homework help.

In the same UNICEF survey, on average, 47% of children polled use AI for finding information, and 11% use it to seek advice about things that worry them. In Italy, 30% of 15–19-year-olds surveyed reported using AI daily, and 42% consult AI for help making important life decisions and when they feel sad or anxious, according to a 2025 CSA Research report. In Brazil, 10% of internet-using children use generative AI to chat about personal issues according to the ICT Kids Online Brazil Survey 2025.

Beyond Loneliness and Mattering in Youth

For the recent University of Ottawa cross-sectional study, the researchers analyzed Ontario Health and Peer Relations Study data from over 39,700 students in grades four through 12 during the period from May 2025 to March 2026 from four school boards in Ontario, Canada.

“We found that seeking emotional support from generative AI was prevalent and associated with more emotional problems,” the University of Ottawa team reported.

The Canadian researchers found that 21% of the students reported using generative AI for emotional support purposes overall, and affective AI use was more common among older students and among students who use AI for schoolwork.

Not surprisingly, their research also showed an association between youth affective AI use with higher scores of loneliness and lower scores of “mattering.” Mattering refers to the human need to be significant to others who are important to them and the concept was introduced to modern psychology in 1981 by Rosenberg and McCoullough with their publication of “Mattering: Inferred significance and mental health among adolescents.” The lower the score on mattering, the greater the risk of mental health issues in youth. The Canadian researchers say their findings suggest that students engage with AI chatbots to “fill social voids.”

Interestingly, after adjusting their model for loneliness, mattering, demographics, and functional AI use, the researchers discovered that there was still an association between affective AI use and high clinically significant symptom scores in youth. The Canadian researchers discovered that among youth, affective AI users had a 26% greater incidence of “clinically significant emotional problems.”

“Affective AI use was significantly associated with higher scores of emotional problems and a greater prevalence of clinically significant symptoms, independent of mattering and loneliness,” wrote the Canada-based researchers.

These findings may have clinical impact. The University of Ottawa researchers recommend that mental health practitioners regularly screen for affective AI use.

“Clinicians should normalize screening for affective AI use as this behavior serves as a robust marker for underlying mental health concerns that likely requires more intensive professional intervention than a digital interface can provide,” the researchers concluded.

Copyright © 2026 Cami Rosso All rights reserved