When AI Becomes a Student’s Confidant

Artificial intelligence has become increasingly intertwined with young people’s personal lives, including during their college years, when many are developing their identities and building meaningful relationships. But as AI becomes a source of emotional support and companionship, new research from The Rithm Project suggests that young people don’t necessarily have to trust the technology to turn to it for advice—they just need to perceive it as less risky than confiding in another person.

Earlier this year, the organization released research based on a survey of nearly 2,400 young people ages 13 to 24—including more than 500 young adults ages 18 to 24—examining how their AI use intersects with their social lives, emotional well-being and relationships.

Today, The Rithm Project released new research that builds on those findings through interviews with 29 young people from the original study—including 19 who are either currently enrolled in college or have completed a bachelor’s or associate degree. The interviews focused on two groups identified in the initial survey: those who turn to AI for personal and emotional support and those who interact with AI characters and personas.

Alison Lee, head of research and development at The Rithm Project, said one of the most notable findings was how young people weighed the perceived risks of confiding in AI against those of opening up to another person.

“What they were saying was ‘When I turn to AI, I know it’s hallucinating, I know that it’s glazing me, I know that it’s engaging in all these problematic behaviors,’” Lee said. “‘But those feel way more manageable as a risk than when I turn to a human.’

“Turning to a human is scary, it’s vulnerable, and you’re afraid of their judgment and emotional reactivity,” she added. “But big emotional events are a hallmark of the human experience … so it makes a lot of sense for young people to search for a place where those big emotions have a safe place to land and where they can make meaning of it.”

How AI shapes relationships: The research explored interviewees’ experiences with relational AI use, including how they began using the technology, why they turned to it for emotional support, whether their real-life relationships had changed and what they shared with or concealed from friends and family. Four common themes emerged.

1. Relational disruptions drove AI use. Almost all interviewees described turning to AI for emotional support following a specific interpersonal challenge or disruption, rather than generalized loneliness. Those experiences ranged from conflicts with friends or romantic partners to other moments when participants felt they needed someone to confide in.
2. AI felt less risky than people. Participants often perceived AI as a safer place to share personal concerns because its shortcomings, including bias or excessive validation, seemed easier to manage than the possibility of judgment, gossip or emotional reactions from another person. Those same concerns also contributed to participants concealing their emotional AI use from others.

3. AI reinforced existing relationship habits. The interviews suggest that AI served as either a bridge or a wedge between young people and their human relationships. Some participants used AI to consider alternative perspectives or prepare for difficult conversations, while others described receiving validation that reinforced their existing views of interpersonal conflicts.
4. Recognizing risks with AI encouraged safeguards. Participants who understood and expressed concern about problematic AI behaviors were more likely to describe efforts to protect themselves, such as setting boundaries around their AI use or prompting the technology to challenge their assumptions rather than simply validating them. However, recognizing these risks alone did not necessarily lead participants to change how they used AI.

Lee said the interviews highlighted how AI can serve as a valuable outlet for young people to process their emotions and gain clarity about personal challenges. The concern, she said, arises when AI becomes their only source of emotional support.

“What we don’t want is for AI to be the only crutch for them to turn to in the moment of that gateway … because it already primes itself to be the perfect place to catch you when you fall,” Lee said.

With that in mind, Lee said, two factors are particularly important in determining whether AI strengthens or undermines young people’s relationships: the kinds of questions they ask the technology and whether they have real people to turn to for support.

“If they’re actively prompting AI to understand someone else’s point of view or for self-improvement or to take an action to repair a relationship, AI often acted as a bridge,” she said. “But if you’re regularly only seeking affirmation, you’re only ever seeking to be heard and validated—and that’s what you’re prompting AI for—it is much more likely that the AI is going to start to act in ways that become a wedge between you and your real-life relationships.”

Strengthening human connections: Lee said the findings raise questions about how colleges and universities can help students navigate their relationships with AI while strengthening the human connections that contribute to their well-being and sense of belonging on campus.

She said institutions should consider where students naturally turn when they experience emotional challenges and how to expand their networks of support. While administrators have limited time and capacity to provide support, colleges can help students build relationships with a wider range of people they feel comfortable reaching out to regularly.

“A lot of young people are mostly connecting to friends or asking for help or venting through platforms like Instagram or Discord or whatever digital platforms they’re on,” she said. “That’s why Claude and ChatGPT have been so prolific. It just feels like you’re texting or [direct messaging] somebody, and for this generation, they’ve grown up texting and DMing with their closest confidants.

“So how do you make that as an avenue for help-seeking just as easily for college services and supports?” she added.

Ultimately, Lee said, reframing conversations about young people’s relational use of AI can help them consider the potential rewards—and not just the risks—of turning to humans instead of AI.

“The risks of turning to humans feels really scary and consequential and permanent and physiologically activating,” she said. “But if you were to include the rewards of turning to AI versus rewards of turning to humans, the rewards of turning to humans would outweigh what it means to turn to AI.”

For higher education leaders, she said helping students recognize the unique benefits of human connection is just as important as educating them about the potential risks of relying on AI for emotional support.

“Being seen and witnessed by a group of friends that really love you and hold you down and keep you grounded and gas you up is the beautiful, powerful reward of human relationships,” she said. “Something that I hope more young people do is say, ‘Let’s not just go all in on AI or ban AI. Actually, what we really need to understand is what are the risks and rewards of each of our choices so that we can be clear about what we actually really care about.’”

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