Here's a shocking truth: Despite all the hype around AI, most people barely use it—and those who do often share a surprising personality profile. A groundbreaking study published in Cyberpsychology, Behavior, and Social Networking (https://doi.org/10.1177/21522715251379987) reveals that artificial intelligence plays a far smaller role in our daily online habits than we might think. Analyzing over 14 million website visits, researchers discovered that less than 1% of browsing activity involved AI tools for the average person. But here's where it gets controversial: The small subset of frequent AI users were more likely to exhibit darker personality traits like narcissism and psychopathy.
Public fascination with AI has skyrocketed, especially as tools like ChatGPT infiltrate classrooms, workplaces, and creative spaces. Yet, here's the catch—most surveys rely on self-reported data, which is notoriously unreliable. People often overestimate or misremember their tech usage, particularly with emerging technologies. To cut through the noise, researchers took an innovative approach: They analyzed actual web browsing histories to measure AI engagement objectively. Their goal? To uncover real-world usage patterns, identify who's embracing AI most enthusiastically, and explore how these tools fit into broader online behaviors. Understanding these dynamics could help predict how AI adoption might unfold across different demographics.
"The disconnect between AI's cultural prominence and its actual usage is staggering," explains lead researcher Emily McKinley (https://emilykmckinley.scholar.st/), a PhD candidate at UC Davis. "Everyone's debating AI's societal impact, but until now, we've lacked concrete data on how—and how often—people interact with it in their daily lives. We wanted to move beyond speculation and see what the digital breadcrumbs reveal."
The study involved two distinct groups: 499 university students (typically early tech adopters) and 455 general population participants. All shared 90 days of Chrome browser history (the only platform enabling such detailed exports) and completed personality assessments. Using a predefined list of AI sites (like ChatGPT and Microsoft Copilot), researchers tagged AI-related visits and categorized other websites using LLM-powered classification. They then mapped usage frequency, contextual browsing patterns (what people did before/after AI sessions), and psychological correlations.
Key findings that might surprise you:
- Among students, AI accounted for just 1% of total browsing—with ChatGPT dominating 85% of those visits. While this surpassed traffic to some social media web platforms (like Instagram), it paled next to search engine usage.
- "Even tech-savvy students used AI far less than we expected," McKinley admitted to PsyPost. But here's the twist: The minority who were heavy users (defined as >4% of total visits) scored significantly higher on the "Dark Triad" traits: Machiavellianism, narcissism, and psychopathy. These individuals also held more positive AI attitudes overall. Demographic links (e.g., income, gender) were weak, but age and ethnicity showed no meaningful connection.
- In the general public? AI usage plummeted further—just 0.44% of visits. ChatGPT remained the top platform, and Machiavellianism showed the strongest (albeit modest) personality link. Prolific users in this group were harder to identify due to sparse activity, but favorable AI views still predicted slightly higher engagement.
The workflow factor: Contextual analysis revealed telling patterns. Before AI sessions, users often visited search engines or login pages; afterward, they frequently switched to educational, professional, or computer-related sites. This suggests AI is primarily leveraged as a productivity aid—not for entertainment. But here's the part most people miss: Self-reports of AI usage were only moderately aligned with actual behavior, reinforcing known biases in tech habit recall. (Ever think you use an app more than you actually do? Exactly.)
Limitations and future directions:
- Mobile app usage (a major AI access point) wasn't tracked.
- Chrome exclusivity may skew samples toward certain user profiles.
- The study didn't capture what people did on AI platforms—were they drafting essays, coding, or just experimenting? Content analysis could reveal deeper motivations.
- "We're now exploring how AI use ties to tangible outcomes," McKinley notes. "Does it affect academic integrity? Job performance? These are the pressing questions."
Provocative question for debate: Given the link between dark traits and AI adoption, could early power users be exploiting these tools for personal gain more than communal benefit? Or is this simply a case of confident, ambitious individuals leveraging cutting-edge resources? Share your take in the comments!
Study reference: McKinley, E., Markowitz, D. M., Zhu, R., & Van Der Heide, B. (2024). Evaluating Artificial Intelligence Use and Its Psychological Correlates via Months of Web-Browsing Data. Cyberpsychology, Behavior, and Social Networking. https://doi.org/10.1177/21522715251379987
