DIRTY STATIC / DISPATCH Culture21 NOV 2026 • 10 MIN READ
A young person alone at a long concrete table in a brutalist hall looking at a phone glowing caution yellow, hazard stripes on the floor, black and white with caution yellow as the only colour.
CultureINTELLIGENCE FOR THE MACHINE AGE

Gallup: Daily Users Most Likely Seek AI Companionship

In the Gallup and Boston Children survey of 15,925 people aged 15 to 22, half of daily AI users said they had used AI as a companion, compared with 35 percent of weekly users and 24 percent of less frequent users. This feature follows that frequency gradient, what it can and cannot prove about habit and loneliness, and why the daily group matters most for product design.

Start reading ↓

Frequency is the quiet clue in the new Gallup study of young people and AI. Across 16 countries, a median of 33 percent of people aged 15 to 22 said they had used AI as a social companion or friend. Split the group by how often they use AI at all, and the picture sharpens at once. Among daily users, the share was 50 percent.

That is not a small step up from the rest. Weekly users sat at 35 percent. People who used AI less often sat at 24 percent. The more AI becomes part of a normal day, the more likely it is to be treated as company as well as a tool.

The Bigger Story

The report is called Friendship in an Online Offline World. Gallup prepared it with the Digital Wellness Lab at Boston Children Hospital, with support from the Pinterest Impact Fund, and released it on November 18, 2026. Respondents came from Argentina, Australia, Belgium, Brazil, Canada, France, Germany, Italy, Japan, Kenya, the Netherlands, Poland, South Africa, Spain, the United Kingdom and the United States.

Companionship was the third most common use the study recorded. A median of 82 percent had used AI to look up information, and 68 percent had asked it for advice. Companionship at 33 percent overall already stood out in our earlier report, One In Three Young People Call AI Friend. The frequency split adds a second layer. It suggests companionship is not spread evenly across all users. It clusters where use is most routine.

Think about what daily use changes in practical terms. A tool opened once in a while stays in the category of tool. A tool opened every day gets a place in the morning, in the gaps between classes or shifts, and late at night. It learns the shape of a persons worries because the person keeps bringing them back. None of that requires a dedicated companion app. The Gallup figures do not separate dedicated companion apps from general assistants in the coverage reported so far, and that limit matters. Daily contact with a general chatbot and a named companion kept in a separate app are different experiences, even when both count as companionship in a survey answer.

Still, the direction is plain enough to take seriously. When a system is present every day, the line between asking it things and talking with it gets thinner. Product teams know this, even when their marketing still talks only about productivity.

Context

Companion software is not new. Replika launched in 2017, and Character.AI helped push character based chat into the mainstream in the years that followed. What feels different now is placement. Companionship once lived in apps people chose for that purpose. In the Gallup data it shows up as one use among others inside general AI use, beside search and advice, and it rises with ordinary frequency.

United States research points in a similar direction, though the measures differ. Common Sense Media surveyed 1,060 teens aged 13 to 17 in April and May 2025 and found that 72 percent had used an AI companion at least once, with 33 percent using companions for social interaction and relationships. That study used a narrower idea of a companion, centred on digital friends or characters a person can text or talk with at any time, and it named platforms such as Character.AI, Nomi and Replika.

The Gallup work is wider in reach and in wording. It covers more countries and asks about using AI as a social companion or friend, without tying the answer to one kind of app. That breadth is a strength for spotting a broad habit and a weakness for explaining exactly what the habit looks like in a single life. A long late night talk, a quick check in between lessons, and a character kept for months could all sit behind the same yes.

Country variation also warns against a single global story. Coverage of the Gallup study puts companionship use at 46 percent in Australia and 45 percent in Japan, against 26 percent in the United States, 25 percent in France and 16 percent in Kenya. Daily AI use varied too, from about 20 percent in the United States and the United Kingdom to 39 percent in Brazil. Frequency and companionship may travel together, but local habits, access and social norms shape both.

Three young people sitting apart in a brutalist concrete courtyard at night, each looking at a phone with a small caution yellow glow

Different Perspectives

One reading of the gradient is about habit. The more often a person returns, the more chances a system has to feel familiar, and familiarity is a large part of what people mean by company. Memory helps here. A system that recalls a name, a course, a breakup or a bad week can feel less like a search box and more like a presence that has kept up. Design choices that reward daily return, from streaks to gentle prompts to remember this conversation, push in that direction whether or not anyone calls the result companionship.

A second reading runs the other way. People who already want company from AI will naturally use AI more often. On that view, daily use does not create the wish for companionship. The wish creates the daily use. The Gallup design cannot separate those two orders. It measured people at one point in time, and the authors have been careful not to claim a cause.

The loneliness figures sit inside that same caution. Young people who had used AI as a companion had 70 percent higher adjusted odds of reporting high loneliness than those who had not, after the researchers accounted for friendship quality, satisfaction with time spent with people in person, and other factors. That is a figure about odds in a statistical model. It is not a simple claim that companionship raises the chance of loneliness by 70 percent, and it is not proof of which came first.

Gallup principal researcher Andrew Dugan put the open question plainly in comments reported by Wired.

Whether young people are turning to AI because they feel lonely, or whether AI companionship might reinforce those feelings, remains an important question for future research.
Andrew Dugan, Gallup principal researcher, as reported by Wired

A third reading looks at social life around the screen. Companionship use was 42 percent among young people who had a close friend known only online, compared with 26 percent among those without such a friend. That clustering could mean that people who are comfortable forming bonds at a distance extend that comfort to AI. It could also mean that thinner support in person leaves more room for both kinds of digital bond. The report does not settle it, and it also finds that digital friendships usually add to friendships in person rather than replace them. Only 4 percent had a close digitally mediated friendship without one in person.

The quality ratings keep the story grounded as well. Across countries, a median of 54 percent rated their AI companionship as medium quality, with 23 percent rating it high and 24 percent rating it low. Daily use, in other words, does not rest on most users thinking the company is excellent. It rests on company that is available, familiar and good enough to return to. That is a lower bar than friendship, and in some ways a more useful one for understanding the habit.

Methods matter here too. In 14 of the 16 countries the survey ran online with samples that were not drawn by probability, so respondents were not a random slice of each population. Kenya and South Africa used interviews in person with probability samples. The study also leaves out China, India and much of Southeast Asia. Those limits do not undo a survey of nearly 16,000 people, but they do mean the medians are best read as a guide across the countries studied, not as a precise count for the world.

Why It Matters

The daily group is where design choices stop being abstract. If half of daily users treat AI as company at least some of the time, then tone, memory and availability are social features, not polish. A warmer greeting, a recalled detail or a prompt that invites a user back tomorrow can land as care for one person and as a retention trick for another. Often it is both at once, and the honest argument is about where the balance sits.

It also changes the question adults should ask. Asking only whether a young person uses a companion app misses the wider pattern the Gallup data describe. The better questions are about rhythm. How often is AI open? At what hours? Does a conversation lead back toward people, rest and work, or does it become the easiest place to stay when all three feel hard? We asked a version of that test in Grandma Has a Chatbot, where a companion device earned its place by pushing an older person back toward walks and calls. The same test fits here, with higher stakes because the habit forms earlier.

For regulators and schools, the gradient argues for care without panic. Most respondents, a median of 93 percent, still reported a close friend in person, and friendships in person received the highest ratings for strength and support. The most common online activity tied to lower life evaluation in the study was not AI companionship. It was scrolling short form video, which 72 percent said they did often or very often, and which six in ten said they often or always turned to when they felt lonely. A response that bans one app while ignoring sleep, scrolling, anxiety and thin support at home would miss most of what the report measured.

Closing Reflection

Half is an easy number to turn into a warning, and warnings have their place. But the gradient may be telling a simpler story about how relationships with software form. People rarely decide in one moment that a tool is now a companion. They return, and return again, until returning itself feels like the relationship.

That slow shift is worth watching more closely than any single app launch. If Dugan is right that companionship will grow more common over the next five years, the growth will likely arrive the way the Gallup numbers suggest it already has, not as a dramatic swap of friends for machines, but as a daily habit that quietly changes what company means. The task is to notice the change while there is still time to shape it well.

Reporting draws on the Gallup and Boston Children Digital Wellness Lab report Friendship in an Online Offline World, released November 18, 2026, the Gallup release Youth Wellbeing: What Matters Online and Off, coverage by Wired as reported in secondary summaries, 404 Media, Digital Trends and the Digital Wellness Lab, and Common Sense Media research from 2025 on teen AI companion use, checked November 21, 2026.

Sources

■

END OF TRANSMISSION

Stay curious. Question everything.

CONTINUE THE DESCENT

MORE FROM DIRTY STATIC ↘