Lonely Kids and AI: Is AI the Cause or the Comfort?

About half of kids who use AI every day say they feel lonely at least some of the time. Among kids who barely use it, it’s a third. That gap, from Common Sense Media’s 2026 census, is real. What it means is the hard part.

Do chatbots make kids lonely? Or do lonely kids reach for chatbots because they’re lonely? The question about lonely kids and AI is a textbook case of correlation versus causation, and it happens to be one of the best ways to teach your own kid how evidence works. The short answer from the studies that follow people over time: probably both, and the second step is the one parents can interrupt.

Key Takeaways

• In the 2026 Common Sense census of 1,204 U.S. kids aged 9 to 17, 54% of daily AI users and 47% of weekly users felt lonely at least sometimes, versus 33% of kids who used AI less than monthly.
• A one-time survey cannot show direction. Common Sense says so itself: lonely kids may seek AI, or overreliance may displace “healthy coping skills.”
• Longer studies are starting to answer it. A 12-month study of adult Character.AI users linked sustained use to lower well-being through less in-person contact; a forthcoming UNC study of middle schoolers found companionship use went with growing loneliness.
• In MIT and OpenAI’s randomized trial, starting loneliness barely predicted how much people later used the chatbot, which makes “lonely people just use it more” an incomplete story.
• The best parental lever is the second step: protecting in-person time and coping practice, whatever started the use.

What the Data Show About Lonely Kids and AI

Correlation is when two things tend to show up together. Causation is when one produces the other. A survey can show correlation. Showing causation takes time, comparison groups or experiments.

Here is the correlation, from The Common Sense Media Census: AI Use by Tweens and Teens, fielded March 18 to 26, 2026:

• Kids who feel lonely at least sometimes: 54% of daily AI users, 47% of weekly, 44% of monthly and 33% of those using it less than monthly.
• Lonely kids used AI more for social and emotional reasons: 46% had used it to practice conversations or social skills and 45% to discuss feelings, compared with 36% and 31% of kids who rarely feel lonely.
• Daily AI users’ average happiness was 7.9 out of 10, versus 8.5 among kids who never use AI. Small, but consistent.

The report then says plainly what it cannot know: “the results of a single survey cannot determine whether more frequent AI use may lead to increased loneliness and lower life satisfaction, or whether children who feel lonelier and less happy with their lives are more likely to use AI in the first place.” It adds a third possibility: other factors may drive both.

The Jed Foundation’s October 2026 survey of more than 5,500 students in 21 schools found the same kind of association: students who used AI for companionship reported more anxiety, depression and loneliness. JED’s vice president of research, Kayla Anderson, told NPR: “It’s unclear, based on these data, whether AI use is contributing to poor mental health and well-being, or whether students who are already in distress in certain ways are more likely to turn to AI.”

Three Stories That Fit the Same Numbers

Any correlation between A and B can come from at least three causal stories.

Story 1: Comfort-seeking (loneliness → AI). Lonely kids look for something that listens. A chatbot is available at midnight, never busy and never judgmental. This is plausible, and earlier research cited by the MIT team found loneliness prompts more chatbot use.

Story 2: Displacement (AI → loneliness). Time with a chatbot replaces time with people, or the chatbot’s easy comfort replaces practice at handling hard feelings with others. Loneliness grows.

Story 3: A third factor (something → both). A move, a falling-out with friends, social anxiety, or a hard year at home could raise both loneliness and AI use at once, without either causing the other.

These are not mutually exclusive. The interesting question is which ones the evidence supports.

What the Longitudinal Studies Start to Answer

Studies that measure the same people more than once can test which comes first.

The 12-month Character.AI study. Stanford and Carnegie Mellon researchers surveyed 1,182 adult Character.AI users and followed up with 439 about a year later. Sustained engagement predicted lower well-being at follow-up, mostly explained by less in-person social interaction. That supports Story 2, specifically the displacement version. It is a preprint, studied adults, and kept only 37% of participants.

The MIT and OpenAI randomized trial. In a four-week trial of 981 adults, people who chose to use ChatGPT more each day ended with more loneliness and less socializing. The researchers also checked Story 1 directly: starting loneliness had a negligible correlation with later usage. In their sample, lonely people did not simply use it more. Usage was self-chosen, so this is still not proof that usage caused loneliness. Our breakdown of the MIT randomized trial has the details.

The UNC middle-school study (forthcoming). UNC developmental psychologist Anne Maheux told NPR in October 2026 that in a soon-to-be-published study of middle schoolers in one state, students who used chatbots for companionship grew lonelier over time, with similar results in young adults. Her reading is that the link goes both ways: “kids who do have worse mental health are seeking out AI to help them. But that AI use may be exacerbating those mental health challenges instead of helping.”

Her proposed mechanism is worth knowing. AI may ease loneliness, anxiety or stress “in the immediate moment,” she said, while reliance prevents young people “from developing the skills to tolerate and navigate normal daily stresses.” The paper is not out yet, so treat this as a researcher’s description, not a published finding. We cover it in the middle-school companionship study.

Where that leaves us

Comfort-seeking probably starts it for some kids. Displacement probably keeps it going for some. Third factors almost certainly play a role. No study yet follows a large, representative group of children for years. The research is early, and it is pointing in one direction.

How to Teach Your Kid About Correlation vs. Causation (Using This Exact Question)

Ages 5–8: Do umbrellas make it rain?

Ask: “When lots of people carry umbrellas, is it usually raining?” Yes. “So do umbrellas make it rain?” Watch them laugh. Then try one where it’s less obvious: “Kids who eat more ice cream get more sunburns. Does ice cream cause sunburn?” Help them find the hidden thing (sunny days) that causes both. That hidden thing is Story 3.

Ages 9–12: Track two things for a week

Pick two things your family can measure daily, such as minutes spent outside and a 1-to-5 mood rating at dinner. Record both for seven days on a simple chart. If they move together, ask your kid to come up with three explanations: being outside improves mood; good moods make you want to go outside; or sunny weather causes both. Then ask: “What would we need to do to find out which one is true?” (Answer: change one thing on purpose and see what happens to the other.) For a related exercise on how AI itself confuses patterns with causes, try our explainer on causal AI vs. correlation.

Ages 13+: Read the real study like a scientist

Show your teen the Common Sense numbers above. Ask them to write down the three stories (loneliness causes AI use, AI use causes loneliness, something else causes both) and one real-life example of each from their school. Then describe the three study designs: a one-time survey, a 12-month follow-up, and a randomized trial. Ask which design could rule out which story, and what each one still can’t tell you. A teen who can explain why the MIT trial could test the voice but not the usage has understood something most headline writers haven’t.

The question to ask: “If lonely kids use AI more, what are two different explanations, and what kind of evidence would tell them apart?”

Cause or Comfort: The Evidence Side by Side

Explanation	What it predicts	Evidence for it	Evidence against or missing
Comfort-seeking (lonely → AI)	Loneliness comes first, then more AI use	Lonely kids use AI more for feelings and social practice (Common Sense 2026); earlier studies cited by MIT team	In the MIT trial, starting loneliness barely predicted later use
Displacement (AI → lonely)	Heavy use comes first, then less contact and more loneliness	12-month study: use → less in-person time → lower well-being; MIT trial: more daily use, more loneliness	Both studies are adults and preprints
Third factor (something → both)	Life events raise both at once	Common Sense notes this possibility; JED found worse outcomes when AI was used while struggling	Hard to measure; rarely tested directly
Both ways (cycle)	Lonely kids seek AI, reliance deepens loneliness	UNC middle-school and young-adult studies, as described to NPR	Not yet published

What to Do at Home, Whichever Way the Arrow Runs

Treat loneliness as the first problem

If your kid is lonely, that matters with or without a chatbot. The census found lonely kids and kids who find friendship hard lean on AI more for social practice. Ask about friends directly. Look for one new low-pressure way to be around peers: a club, a team, a cousin’s weekend.

Protect the second step

Displacement is the part you can influence. Keep a few fixed in-person commitments on the calendar so chatbot time can’t quietly eat them.

Let comfort be a bridge, not a destination

If your kid uses a chatbot to calm down before a test, that’s not a crisis. Ask what they’ll do next. “And then who did you talk to?” turns the bot into a step toward people.

Build coping skills on purpose

Maheux’s hypothesis is that instant comfort can stunt the skills for tolerating normal stress. Practice them with your kid: naming the feeling, a walk, writing it down, calling a friend. Boring, effective, human.

What not to do

Don’t announce “AI is making you lonely.” The evidence doesn’t support that certainty, and a lonely kid will hear blame. Start from “it sounds like things have been hard with friends.”

What to Watch For Over the Next 3 Months

• Week 4: Ask your child to rate their week with friends from 1 to 5 every Sunday. Note chatbot use loosely alongside.
• Month 2 red flags: Loneliness ratings dropping while chatbot time rises; your child says the bot is the only one they talk to; withdrawal from activities they enjoyed. Persistent sadness or hopelessness is a reason to call your pediatrician. In a crisis in the US, call or text 988.
• Month 3 self-check: Has your kid added or kept at least one regular in-person social activity? Is their Sunday rating steady or rising? If so, the displacement story isn’t taking hold.

Frequently Asked Questions

Does AI make kids lonely?

The evidence so far doesn’t prove it, but it raises the concern. Surveys show lonely kids use AI more without showing which causes which. Studies that follow people over time suggest heavy reliance may reduce in-person contact and deepen loneliness, while lonely kids also seek AI for comfort.

Is it bad if my lonely kid talks to a chatbot?

Not automatically. Some comfort in a hard moment can help. The concern is when the chatbot replaces people rather than leading back to them. Watch whether your child’s time with friends and family holds steady while they use it.

Why can’t a survey tell us the answer?

A survey measures people once. If lonely kids and heavy AI users are the same kids, the survey can’t tell whether loneliness came first, AI use came first, or something else caused both. Answering that takes studies that follow children over time or experiments.

What is the strongest study so far?

For direction, the 12-month Character.AI study and the MIT and OpenAI four-week randomized trial are the most rigorous, but both studied adults. The UNC middle-school study described to NPR in October 2026 is the first longitudinal evidence on kids, and it is not yet published.

How do I explain correlation vs. causation to my child?

Start with a silly example: umbrellas and rain, or ice cream and sunburn. Then apply it to a real question, like whether AI makes kids lonely. Ask for two explanations and what evidence would tell them apart. That habit transfers to every headline they’ll read.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

---

Sources

1. Common Sense Media. (2026, June 8). The Common Sense Media Census: AI Use by Tweens and Teens. https://www.commonsensemedia.org/sites/default/files/research/report/2026-ai-use-by-tweens-and-teens-1.pdf
2. The Jed Foundation. (2026, October 27). “New JED Research Shows Link Between Teens’ AI Use, Suicidal Thoughts and Behaviors.” https://jedfoundation.org/new-jed-research-shows-link-between-teens-ai-use-suicidal-thoughts-and-behaviors/
3. Chatterjee, R. (2026, October 30). “Nearly 2 out of 3 teens and tweens use AI chatbots, new research shows.” NPR. https://www.npr.org/2026/10/30/nx-s1-6022278/ai-artificial-intelligence-teens-jed-common-sense-media
4. Zhang, Y., Zhao, D., Wang, Y., Anselmetti, R., Hancock, J. T., Kraut, R., & Yang, D. (2026). “Living with AI Companions: Sustained AI Companionship Predicts Lower Well-Being Through Lower Human Interaction.” arXiv preprint (not peer-reviewed). https://arxiv.org/abs/2609.07243
5. Fang, C. M., et al. (2025). “How AI and Human Behaviors Shape Psychosocial Effects of Extended Chatbot Use: A Longitudinal Randomized Controlled Study.” arXiv preprint (not peer-reviewed). https://arxiv.org/abs/2503.17473