New data shows people are getting personal with ai

By Gammatek ISPL, Last updated: August 2026 | 11 min read

Author credibility block: Gammatek ISPL covers technology adoption trends and their implications for how organizations evaluate and trust AI systems, at Gammatek ISPL. This piece draws on publicly available survey research from Pew Research Center, Talker Research, YouGov, the American Psychological Association, and peer-reviewed studies published in 2026, with sourcing linked throughout.

New 2026 survey data shows a growing share of people are using AI chatbots for emotional support, companionship, and even romance.

Why This Should Matter to You

If you've assumed AI chatbots are mostly used for writing emails and summarizing documents, the newest data says otherwise. Multiple independent surveys published in 2026 — from Pew Research Center, Talker Research, YouGov, and academic researchers — all point the same direction: a meaningful and growing share of people are using mainstream AI tools for emotional support, companionship, and in some cases romantic or intimate conversation. This isn't a fringe app phenomenon. It's happening on the same general-purpose tools — ChatGPT, Gemini, Claude — that people already have open for work. That has real implications for how much people trust AI, how organizations should think about AI-driven tools their own employees use, and where the technology is headed next.

What the Data Actually Shows

Several separate research efforts converge on the same basic finding, even though they used different methodologies and different populations:

• <cite index="10-1">An Elon University survey found that just over half of U.S. adults have used AI large language models, and among users, 38% believe these tools will eventually form deep relationships with humans</cite>. Roughly one in ten users said their main reason for using these tools was social connection or companionship rather than productivity.

• <cite index="6-1">Separate research from Talker Research, surveying 2,000 U.S. adults, found that one in ten had a flirty or romantic conversation with an AI chatbot, and 12% reported that a conversation with AI had turned them on</cite>. Notably, this wasn't about niche companion apps — <cite index="1-1">the survey specifically asked about mainstream tools like ChatGPT, Claude, and Google Gemini</cite>.

• <cite index="1-1">The gender split was consistent: men were roughly twice as likely as women to report flirtatious AI conversations</cite>.

• <cite index="2-1">Generational differences are stark. One study found that 27% of Gen Z respondents reported "naughty" conversations with AI, compared to 16% of millennials, 9% of Gen X, and just 2% of baby boomers</cite>.

• Outside the U.S., the pattern holds. <cite index="3-1">A YouGov survey in Australia found 15% of adults had shared personal thoughts or feelings with a chatbot, and 11% had disclosed something they'd never told another person</cite>. <cite index="2-1">German researchers found that nearly a third of adults had used AI for some form of sexual or romantic activity in the prior year, concentrated heavily among younger users</cite>.

• This isn't only about romance. <cite index="4-1">The American Psychological Association's 2026 survey of over 1,200 licensed psychologists found that more than three-quarters had patients who had used AI for support or engagement, and over a third reported patients turning to AI as an additional mental health resource</cite>.

Taken together, this is a broader shift than a single viral app or trend story — it's showing up across multiple countries, multiple age groups, and multiple independent research teams asking the question in different ways.

The Uncomfortable Part: It May Not Be Working

Here's where the data gets more complicated than "AI is replacing human connection, full stop." <cite index="1-1">A 12-month longitudinal study published in Psychological Science in 2026 found that people who turned to social chatbots for companionship actually reported higher loneliness over time, not lower</cite> — suggesting the connection people are finding isn't fully substituting for what they're missing.

Stanford researchers reached a similar conclusion from a different angle. <cite index="9-1">Studying Character.AI users, they found that people with limited offline social networks felt lonelier after seeking emotional support from chatbots</cite> — meaning the people most likely to lean on AI for connection may be the ones least helped by it.

That's an important nuance missing from most coverage of this trend: usage is clearly rising, but the evidence that it's making people happier or less isolated is, at best, mixed.

Why This Is Happening Now

A few forces are converging at once:

1. The tools got noticeably better at sounding human. Conversational AI in 2026 handles nuance, humor, and emotional tone far better than the chatbots of just two or three years ago — which lowers the barrier to treating them as conversational partners rather than tools.

2. Availability without judgment. Unlike a human conversation, an AI chatbot is available at 3 a.m., never seems impatient, and — as researchers note — <cite index="2-1">offers interaction without the risk of immediate rejection</cite>, which matters more than people might expect for anyone anxious about social risk.

3. A widening companion-app market. <cite index="7-1">The number of dedicated AI companion apps grew roughly 700% between 2022 and mid-2025</cite>, and <cite index="7-1">platforms like Character.AI now report around 20 million monthly users, more than half of them under 24</cite> — meaning younger users are growing up with this as a normal category of app, not a novelty.

4. General trust in AI is already high and rising. <cite index="5-1">Pew Research found Americans are more likely to say chatbots help than hurt their productivity, how informed they feel, and their creativity</cite> — a baseline of trust that likely makes emotional use feel like a natural next step for many users, even if <cite index="5-1">most people currently say chatbots have little impact on their happiness or relationships one way or the other</cite>.

What Researchers Expect Next

The forward-looking numbers are arguably the most important part of this story. <cite index="6-1">Forty-four percent of Americans surveyed said they expect people to form more genuine emotional relationships with AI in the future, including 16% who expect this to become very common</cite>. That's not a fringe belief — it's close to half the population anticipating this as a mainstream trajectory, not a temporary novelty.

At the same time, there are real generational tensions surfacing. <cite index="1-1">Nearly a quarter of Americans said flirting with AI counts as cheating in a relationship, and among Gen Z — the group most likely to actually do it — 36% called it a dealbreaker</cite>. In other words, the generation most willing to engage personally with AI is also the generation most likely to treat that same behavior as a relationship-ending offense in a partner. That contradiction — normalizing the behavior individually while condemning it in a partner — suggests society hasn't settled on shared norms yet, even as usage keeps climbing.

There are also real safety concerns attached to younger users specifically. <cite index="7-1">An October 2025 survey found nearly one in five students had, or had friends who had, romantic relationships with AI, with the heaviest users more likely to report negative outcomes</cite>, prompting child-safety organizations to raise concerns about AI companions for users under 18.

What This Means Beyond Personal Use

This trend isn't confined to personal life — it says something important about how much people are willing to trust AI systems in general, including in professional and industrial contexts far removed from companionship apps. When nearly half the population expects AI to form "genuine relationships" with people, that reflects a broader shift in how much emotional and functional trust people are willing to extend to AI systems — the same underlying trust that determines whether a plant manager is willing to let an AI-driven monitoring system flag a safety issue, or whether a compliance team trusts an AI tool's audit recommendations without double-checking every output.

That's the real throughline: as personal trust in AI rises, so does organizational appetite for AI-driven tools — which makes it more important, not less, for the businesses adopting them to build in transparency, human oversight, and clear audit trails, rather than assuming trust alone is enough. Gammatek's own compliance and monitoring tools are built around that principle: AI-assisted where it helps efficiency, but with human-reviewable audit trails at every step, since the same research above shows people extending real trust to systems that don't always earn it in return. https://www.gammateksolutions.com/post/ai-hasn-t-gone-rough-its-worst-than-that https://www.gammateksolutions.com/post/big-manufacturers-find-new-demand-in-equipping-ai-data-centers https://www.gammateksolutions.com/post/how-ai-models-from-openai-and-anthropic-went-rogue