AI Usage Mirrors Young People’s Offline Struggles

As generative artificial intelligence becomes a routine part of young people’s lives, new research shows that AI’s impact on mental health hinges more on the context in which it’s used than frequency of use.

A report from Surgo Health, in partnership with Young Futures and The Jed Foundation (JED), found that about 12 percent of youth who reported mental health struggles said they turned to generative AI to talk about their feelings. While short-term emotional relief was common, it didn’t consistently lead to positive outcomes over time. When AI was used as part of a broader support system, benefits were greater—but neutral or negative experiences were more likely when it served as a substitute.

Katie Hurley, vice president of community initiatives at the Jed Foundation, said the findings point to a growing disconnect between young people and the systems shaping their digital lives—underscoring the need for greater “algorithmic transparency” from AI platforms.

“It’s worth noting that these systems are designed to be validating and comforting,” said Hurley. “So when someone lacks parental support, peer support or any caring adult in their life, and they turn to one of these [AI chatbots] and feel understood, it’s natural they’ll keep going back for more.”

The research draws on national survey data from more than 1,300 youth ages 13 to 24, exploring how young people engage with AI in fundamentally different ways, shaped by social support, stress, adversity and access to care.

Hannah Kemp, chief solutions officer at Surgo Health, said the research suggests young people are turning to AI in ways that reflect whether their needs are met offline.

“Whatever is happening in people’s offline lives, their AI patterns often mirror that,” Kemp said. “When folks face barriers to accessing care, AI can become a substitute. And when they carry a lot of anxiety about the future, they’re more likely to become emotionally entangled with it.”

The report identifies six segments of AI engagement, ordered by increasing frequency of use.

About 10 percent of young people fall into the “Low-Use Anxious Skeptics” group—distressed, distrustful youth who keep their distance from AI amid uncertainty and fear of rapid change. Another 32 percent are “Thriving Light-Touch Pragmatists,” well-supported youth who maintain a healthy, arm’s-length relationship with AI.

Roughly 7 percent are “Worried Strivers,” highly pressured, future-anxious youth who see AI as destabilizing rather than supportive. And 9 percent are “Emotionally Entangled Superusers,” emotionally vulnerable youth who turn to AI for connection and coping when offline support falls short.

However, about 10 percent are labeled “High-Hope, High-Use Skill-Builders,” optimistic power users who treat AI as a tool for learning, creativity and future-building. Another 10 percent are “Curious Low-Concern Learners,” confident, socially grounded youth who use AI to explore, learn and solve problems.

“Our segmentation shows there are very different patterns,” Kemp said. “If we really want to understand how young people are using and engaging with AI, we have to move beyond screen time and look at the different segments of youth and what support they actually need.”

AI patterns vary: About 44 percent of older youth, ages 18 to 24, say they have little control over how AI affects them, compared with 29 percent of younger youth ages 13 to 17.

In addition, only 26 percent of older youth fall into the “Thriving Light-Touch Pragmatist” segment, compared with 41 percent of younger youth—a 15-percentage-point difference that suggests older youth are less likely to engage with AI in a balanced, low-intensity way while maintaining overall well-being.

Older youth are also more likely to never use AI: About 25 percent, compared with 14 percent of younger youth, eschew AI. They are also more likely to fall into the “Low-Use Anxious Skeptics” segment—12 percent of older youth versus 7 percent of younger youth, suggesting a larger share of older youth may carry broader, more systemic concerns about AI.

Hurley added that the COVID-19 pandemic may help explain why older youth have a less positive relationship with AI, noting that many were in middle or high school at the time, when education leaders focused heavily on academic learning loss but paid far less attention to social skills loss.

“When you’re separated from your peers, you don’t get to practice things like conflict resolution, how to date, how to make friends, how to move in and out of groups—a lot of critical social development was missed for many young people,” Hurley said. “They were taught to go online to get help … [and] for better or for worse, that became a go-to strategy.”

“Now they can use AI to find information on all kinds of things, but they’re not always asking the right questions to get what they’re looking for,” she added.

The report also finds that patterns of AI engagement differ meaningfully across demographic groups. Black youth and youth in families receiving government assistance are substantially more likely to be “High-Hope, High-Use Skill-Builders,” using AI for learning and future-building.

By contrast, Hispanic youth who use AI are more likely to fall into the “Emotionally Entangled Superusers” segment, while LGBTQ+ youth are more likely to be “Low-Use Anxious Skeptics.”

Adele Wang, associate director of research and data science at Surgo Health, said these demographic findings highlight an area of opportunity for young people who don’t have access to traditional care and support services.

“There’s a low barrier to access, which is why so many people are being driven to [AI chatbots],” Wang said. “At the same time, individuals receiving financial assistance and those who are more likely to be emotionally entangled superusers may be using it to fill gaps left by missing support and stability in their external environments.”

Implications: The report emphasizes the need for “public health leaders, educators and policymakers to move beyond blanket approaches to youth and AI.” Instead, it calls for segment-informed strategies that “strengthen offline support, protect youth agency and ensure AI complements, rather than replaces human connection.”

Kemp said there’s opportunity for generative AI to serve as a bridge to offline care rather than a replacement for it.

“Could it help someone quickly find a therapist who takes their insurance or a community clinic that offers free counseling?” Kemp said. “That might help smooth some of that friction, but that’s just not how [AI chatbots] are designed right now.”

Wang agreed, adding that disconnection not only from peers but also from parents plays a role in how young people approach generative AI.

“That miscommunication with parents is another key reason young people are turning to AI,” Wang said. “What’s especially interesting—and concerning—is that many don’t want to feel like a burden to others.”

Ultimately, Hurley said the solution isn’t limiting AI, but understanding the needs driving young people to it.

“It’s really important to elevate youth voices and to listen and learn from them,” Hurley said. “We can send them all the reports we want, but they’re using these things in ways that make sense to them, so it’s important to meet them where they are and hear what they’re trying to get out of it.”

Get more content like this directly to your inbox. Subscribe here.