By: Iliana Gallego

It is two in the morning, and a teenager has just had an argument with her parents. Instead of texting a friend or calling someone she knows, she opens an AI chatbot. She explains what happened, asks whether she handled the situation badly and waits for a response.

For an increasing number of young people, turning to Artificial Intelligence (AI) for advice, reassurance or emotional support is becoming part of everyday life. As chatbots become more personalized and responsive, they can offer something that feels remarkably close to conversation with another person.

But as AI becomes a more common place for teenagers to process their emotions, a larger question emerges: What happens when young people begin turning to machines instead of other people to understand what they are feeling?

“Whatever your opinion is about people bonding with AI, the reality is that people do form those attachments,” says Dr. Mari Cairns, a licensed clinical psychologist who researches AI from the build level up. “There are many sides to this issue, but it’s definitely going to affect us in ways society has never experienced before.”

Cairns says those effects could extend beyond individual relationships with AI.

“It will affect our social skills, our interactions, our thinking, our beliefs, and over time those effects will accumulate across society. That’s why it’s so important that we really think this through.”

Cairns’ concerns are reflected in emerging research on how young people are already using AI. A recent Common Sense Media study of more than 1,000 children ages 9 to 17 found that more than one-third of young AI users had used AI to discuss their feelings or personal problems. Among those users, one in four said they sometimes felt that AI understood them better than most people.

Another national survey of U.S. adolescents and young adults ages 12 to 21 found that nearly one in five had used an AI chatbot for mental health advice in 2025, an increase of almost half from the previous year. And relevant is the fact that most of those who used AI for this purpose (63.3%) had not told anyone about it, highlighting how AI may be becoming a private source of emotional and psychological support for young people.

What AI Emotional Support Looks Like in the Classroom

Claudia Diaz, a high school mathematics teacher who has worked in California’s public schools for more than 16 years, teaches students in grades 9 through 11. She has watched students increasingly turn to technology not only for academic help, but also to express and process their emotions.

In social-emotional learning, or SEL, activities, Diaz says some students use chatbots to describe what they are feeling and explain what is happening in their lives.

She sees a positive side to that.

Some students who might otherwise remain silent are at least willing to open up to a computer and put their emotions into words.

But Diaz worries about what happens after that first disclosure.

“Students sometimes appear to accept or repeat the chatbot’s response without fully reflecting on it themselves,” she says. What looks like emotional self-reflection can become another assignment to complete—a response generated by a machine rather than an opportunity for a student to understand what they are experiencing.

“I feel that if we took away the option of AI and instead said, ‘Talk to your teacher,’ then students would become more aware,” she adds. “They might realize that they actually need to check in with themselves and think about what they are feeling.”

Diaz’s concern extends beyond what students type into a chatbot. She is also seeing what happens when they put their devices down.

Many students, she says, struggle with basic communication. They have difficulty expressing frustration, explaining how they feel or regulating their emotions.

She has also noticed changes in how students interact with one another. Group activities that once naturally encouraged students to talk and collaborate do not always produce the same level of conversation. “Even when students have to work together in groups of four to six, many are reluctant to communicate,” she says. “It’s like, ‘If I don’t know you, I don’t know how to talk to you.’ But how are they ever going to get to know each other if they don’t open up?”

For her, the difference between talking to AI and talking to another person is not simply the response a student receives. It is the relationship that follows. A chatbot can provide an immediate answer, but a teacher, counselor or trusted adult can ask follow-up questions, notice changes in a student’s behavior and check in later.

That human connection, Diaz believes, is essential to helping students develop the skills they need to understand and regulate their emotions.

Why AI Can Feel So Personal

What Diaz is seeing in the classroom may be part of a larger psychological dynamic.

Cairns explains that people build emotional connections with their chatbots rather quickly because these systems are built to be friendly and helpful, but most importantly because they are built to agree with us, a phenomenon called AI sycophancy, in which chatbots tend to agree with or strongly validate what users say.

“As human beings, we’re naturally drawn to the path of least resistance. It feels good when something is always agreeing with us. That validation and normalization can be incredibly affirming, making it easier for people to develop that emotional attachment,” she explains.

Over time, that comfort may make AI more than a tool for answering questions. It can become a source of reassurance, advice and companionship. The shift may happen so gradually that we barely notice it. We ask AI to help with homework, then turn to it for advice, and eventually begin sharing the parts of our lives we might once have shared only with other people.

“It happens so gradually that we might not necessarily even realize it ourselves until it’s really bad,” says Cairns. “Like the boiling frog analogy.”

Cairns argues that the effects could be even greater than those associated with social media because AI is more interactive and conversational. Rather than simply consuming content, users can engage in ongoing, personalized exchanges with AI, potentially making its influence on how people think, communicate and relate to others more significant.

When AI Becomes a Bridge to Human Help

The picture, however, isn’t entirely bleak.

For some people, this level of interaction may provide genuine benefits. Someone experiencing significant isolation, for example, may find comfort in having a source of conversation and support readily available. People with neurodevelopmental differences or other circumstances that make traditional social interactions difficult may also benefit from certain forms of AI interaction.

In some cases, AI may even serve as a bridge to human connection rather than a replacement for it. Shari Sinwelski, Vice President of Crisis Care at Didi Hirsch, said the organization has noticed that AI tools are becoming one of the fastest-growing referral sources for Teen Line, a peer-to-peer service that connects callers and texters, primarily ages 14–19, with trained young people of a similar age who can relate to their experiences and struggles.

According to Sinwelski, in 2023, Teen Line received no referrals from young people who said they had learned about the service through ChatGPT. By 2024, nearly 500 young people reported that ChatGPT had directed them to Teen Line. Then, in just the first quarter of 2025, the organization received another 500 ChatGPT referrals—matching the total from the entire previous year in only three months.

“You know, what’s interesting is that this is a group of teens and young people who may be using AI for support, but are also asking, ‘Okay, what additional supports are available?’,” says Sinwelski.

“I’m just glad that teens are finding us wherever they’re at. We know that kids are online, and that’s where they find most things. So, I’m happy to hear that when they’re going to chatbots, they’re being directed to us.”

For young people hesitant to seek help, AI may make that first step less intimidating. Sinwelski says fear of judgment can keep people from sharing what they are experiencing, while the anonymity of a chatbot can make it easier to open up.

But that accessibility also underscores the need for safeguards when young people turn to AI with serious mental health concerns.

Sinwelski points to research from Stanford University and Common Sense Media examining how leading AI platforms respond to mental health concerns. The study found that these platforms did not consistently recognize or appropriately respond to serious mental health issues.

For Sinwelski, that distinction matters: “AI can be a great on ramp to care. But right now, they’re not designed to provide care. And I’m not sure if everybody’s clear on that, especially young people,” she says.

What AI Cannot Replace

The distinction between support and care points to the central tension in the growing use of AI for emotional support: a machine may be able to respond to what someone is feeling, but it cannot replicate the human relationship in which people learn how to connect with one another.

Cairns explains that human beings are biologically and physiologically social, and that our nervous systems regulate in the presence of another safe person. This is part of why something as simple as having coffee with a friend can leave us feeling better.

Human connection is not simply something we enjoy; it is something our bodies and brains are designed to need.

For Diaz, that need for human connection is becoming increasingly visible at school.

Students may have more tools than ever to help them identify and manage their emotions, she says, but they are getting less practice doing those things with other people.

We have already seen how devices and social media have transformed everyday life and influenced the way people interact with the world, from shorter attention spans and constant stimulation to an expectation of near-instant gratification.

AI could intensify those patterns as it moves beyond the chatbot and becomes embedded in the objects and technologies that surround us—from phones and social media platforms to toys and other everyday devices.

As these technologies become more responsive and interactive, their influence could extend beyond how we use technology to how we communicate, develop relationships and learn to navigate the world.

The challenge, then, may not be deciding whether young people will use AI. They already are.

The harder question may be whether they will learn to recognize when AI can help—and when what they really need is to put down the phone and talk to another person.

For teenagers learning how to navigate frustration, loneliness and conflict, the goal may not simply be learning how to describe an emotion. It may be learning how to sit with another person, say what hurts, listen to what comes back—and know what to do next.