Artificial intelligence is everywhere… at least in conversation.
A new study from researchers[1] at the University of California, Davis, and Michigan State University took a hard look at what people actually do online instead of what they claim to do. They combed through millions of real browser histories, covering roughly fourteen million website visits. The result? AI tools barely show up.
For most users, visits to AI sites made up less than one percent of their online life. Many people didn’t touch them at all.
That finding feels oddly quiet compared to the buzz around ChatGPT or Copilot or whatever tool makes headlines next. The researchers weren’t interested in hype; they wanted numbers. How often do people really open these systems, who does it most, and what happens before and after those moments?
What the Data Really Showed
Students used AI more than the general public, though not by much. Their AI activity made up about one out of every hundred page views. The broader population landed closer to half that rate. And while a few “heavy users” appeared… those who let AI make up more than four percent of their total browsing… they were rare.
ChatGPT dominated the category. Around 85 percent of all AI visits went to OpenAI’s chatbot. It wasn’t even close.
When researchers mapped where people went before and after those visits, the pattern stood out. Just before AI, most users were at search engines or login portals. Immediately after, they drifted to education pages or professional tools. That chain suggests people slot AI into work or study tasks rather than casual browsing. It’s not a place to hang out. It’s a pit stop.
Personality, Not Just Curiosity
Then came the psychology layer. Each participant had completed surveys measuring personality and attitudes toward technology. Patterns emerged, though not dramatic ones.
Students who leaned heavier on AI tended to score higher on what psychologists call the “Dark Triad”: Machiavellianism, narcissism, psychopathy. Those traits, simplified, describe people who are strategic, self-assured, or indifferent to social rules. Among the general public, the pattern softened, leaving only a faint link with Machiavellianism.
No cause-and-effect story here, just an observation. Still, the connection is interesting. People high in those traits often like efficiency and control. They might see AI less as a novelty and more as a leverage point, a tool that amplifies output without requiring approval or help.
The Illusion of Self-Reporting
Another piece of the puzzle: what people think they’re doing versus what they actually do.
Participants had estimated their AI use through surveys before their data was analyzed. The numbers didn’t line up. Correlation existed, yes, but weakly, proof that self-reports paint a blurry picture. Humans tend to overstate, understate, or just forget.
That matters because many studies and public polls still rely on asking people about their habits. This research shows how unreliable that can be. If we want to know how AI fits into daily life, the evidence will likely come from behavior logs, not memories.
The Quiet Reality Beneath the Noise
Even with its scope, the project had limits. Only web-based activity counted. Mobile app use, which might be higher for some, was left out. Chrome users dominated the sample, since that browser allows easy data export. Despite those gaps, the message stays the same: AI plays a small role in everyday browsing for most people.
It might not stay that way forever. As AI slides deeper into search engines, word processors, and chat platforms, usage will probably rise without anyone noticing. At some point, people won’t “go to” an AI, they’ll simply use the internet, and the AI will be there, humming quietly in the background.
For now, though, the contrast is striking. The world debates how AI will rewrite everything, yet for most people, it hasn’t rewritten much at all. They still scroll news sites, sign in to email, check grades, watch videos. The future everyone’s talking about? It’s loading slower than the headlines suggest.
Notes: This post was edited/created using GenAI tools. Image: DIW-Aigen.
Read next: 60% of Fortune 500 Companies Rely on AWS. What an Hour of Downtime Really Costs[2]
References
- ^ researchers (www.liebertpub.com)
- ^ 60% of Fortune 500 Companies Rely on AWS. What an Hour of Downtime Really Costs (www.digitalinformationworld.com)