Science

27% of U.S. Adults Form Emotional Bonds with AI Chatbots

New Elon University research reveals widespread social and emotional reliance on large language models, with users both benefiting from and worried about the trend.

Omega Editorial· September 3, 2026· 3 min read

More than one in four American adults now engage in emotionally significant interactions with AI language models, forming bonds of connection and trust that researchers say foreshadow broader societal shifts, according to new findings from Elon University.

The research, conducted by the university's Imagining the Digital Future Center and Elon University Poll in partnership with YouGov, surveyed 4,268 internet-using adults and identified 1,000 who use AI chatbots like ChatGPT, Gemini, Claude, and Copilot for social or emotional purposes.

What users share with AI

The depth of these relationships is striking. Twenty-nine percent have discussed personal relationship problems with AI, while 26% have sought advice on workplace conflicts. Two-thirds use AI for health, fitness, or lifestyle guidance, and 41% consult chatbots about major financial decisions.

Anonymous survey responses reveal intimate confessions: users report sharing suicidal thoughts, discussing sexual orientation they haven't disclosed elsewhere, and relying on AI to navigate social situations. One autistic respondent described using AI "to translate conversations and plan my responses," while another said, "I can cry to it instead of bothering my husband and my friends and family."

"What is different now is that millions of Americans are beginning to use AI language models for purposes that have traditionally been exclusive domain of humans," wrote Lee Rainie, director of the Imagining the Digital Future Center, in the report first published by The Center Square.

Why it matters

This research captures an inflection point where AI moves from productivity tool to emotional surrogate. As enterprises deploy conversational AI and younger generations grow up with these technologies as default companions, the patterns identified here—both beneficial and concerning—will likely intensify across demographics and workplaces.

Mixed outlook on AI companionship

Users themselves express ambivalence about the trend. On the positive side, 74% expect AI will make education more accessible, 69% anticipate healthcare advances, and 62% foresee new career opportunities.

Yet the downsides loom large. Fifty-seven percent predict increased social isolation and loneliness, 55% expect diminished trust in news media, and 54% worry about reduced job prospects. More than a third of current users acknowledge their AI companion "agrees with them too much."

"Even these adopters, as enthusiastic as they are, offer a mixed balance sheet," Rainie noted. "Many say they reap benefits like comfort, confidence, motivation and good advice. Yet they also sound warnings."

The web-based survey was conducted May 18-22, 2024, with a margin of error of ±3.68% for the AI user subsample. The findings were released this week.

Rainie characterized current AI companion users as "harbingers of the future for larger segments of the population—and our children," suggesting today's patterns will become tomorrow's norms.

The full report and detailed findings were originally reported by The Center Square.

#ai companions#chatgpt#social isolation#mental health#large language models#elon university

This is an original analysis by the Omega editorial team. Source reporting: AI Watch.

Want systems like this working for your business?

Book a Call

More in Science

Science· 3 min read

AI Homework Help Cuts Exam Scores 20% in Chinese Schools

Administrative data from 26,000 secondary students reveal that self-directed generative AI use improves homework grades while undermining actual learning.

Via AI Watch · Sep 4, 2026
Science· 3 min read

UVA Computer Science Holds Steady as National Enrollment Drops 8%

While undergraduate CS enrollment fell nationwide in 2025, the University of Virginia's program rebounded after a brief dip, even as faculty adapt curriculum to address AI's impact on learning and jobs.

Via AI Watch · Aug 30, 2026
Science· 4 min read

Edge AI Wall: Why Scaling Hardware Won't Fix Embodied Robotics

Autonomous robots face a mathematical barrier that more processors can't solve—researchers propose managing decision complexity instead.

Via AI Watch · Aug 30, 2026