A survey of generative artificial intelligence users found that people with pronounced narcissism are more prone to problematic use of generative artificial intelligence. In women, this association was fully mediated by vertical individualism, an orientation that emphasizes personal independence and autonomy alongside competition, achievement, status, and acceptance of inequality between individuals. In men, the association was partially mediated by vertical individualism and intolerance of uncertainty. The paper was published in BMC Psychology.
Problematic use of generative AI refers to a pattern of excessive, poorly controlled, or harmful engagement with generative artificial intelligence tools that causes significant difficulties in a person’s daily life. It may involve spending excessive amounts of time interacting with AI systems, repeatedly using them despite negative consequences, or feeling unable to reduce their use.
People with problematic generative AI use may become overly dependent on generative AI for work, studying, decision-making, creativity, social interaction, or emotional support. Such use can interfere with academic performance, occupational responsibilities, relationships, sleep, or other important activities. Some individuals may increasingly rely on AI to avoid difficult tasks, uncertainty, boredom, loneliness, or unpleasant emotions. Problematic use may also involve compulsive checking, repeated prompting, or persistent engagement that resembles other forms of problematic digital technology use.
Another concern is cognitive overreliance, in which users increasingly delegate thinking, writing, problem-solving, or judgment to AI systems and become less willing to perform these activities independently. Problematic use does not simply mean frequent use, because intensive AI use can be adaptive and productive when it remains controlled and does not cause impairment. The concept is still emerging, but the term is best used to describe AI use characterized by impaired control, functional impairment, psychological dependence, or continued engagement despite meaningful negative consequences.
Study author Kağan Kırcaburun, an assistant professor at Düzce University in Turkey, investigated the relationship between dark personality traits and problematic generative AI use. He was particularly interested in exploring the possible mediating roles of cultural orientations of individualism and intolerance of uncertainty in this relationship. He noted that previous studies already linked problematic technology use with dark personality traits, but that only one very recent study had specifically examined the relationship between those traits and problematic generative AI use.
He conducted an online survey via Prolific. Participants were 677 individuals who reported active use of generative artificial intelligence. Their average age was about 40 years, and 50% of them were men.
The survey contained assessments of problematic generative AI use (the Problematic GAI Use Scale, created by adapting items of a scale designed to measure the problematic use of conversational artificial intelligence), intolerance of uncertainty (the Intolerance of Uncertainty Scale), individualistic orientation (the Individualism and Collectivism Scale), and dark personality traits (the Dark Triad Dirty Dozen Scale and the Short Sadistic Impulses Scale).
Dark personality traits are socially aversive characteristics associated with self-interest, manipulation, emotional coldness, and disregard for others. There are four of them: Machiavellianism, narcissism, psychopathy, and sadism. Machiavellianism involves strategic manipulation, cynicism, and a willingness to exploit people to achieve personal goals. Narcissism is characterized by grandiosity, entitlement, a strong need for admiration, and sensitivity to threats to one’s self-image. Psychopathy involves callousness, low empathy, impulsivity, and limited remorse. Sadism makes a person prone to deriving pleasure or satisfaction from humiliating, hurting, or controlling others.
Results indicated that participants with more pronounced problematic use of generative AI tended to show somewhat higher narcissism, slightly higher Machiavellianism, and sadism. More pronounced problematic use of generative AI was also associated with higher intolerance for uncertainty and higher vertical individualism.
Intolerance of uncertainty is the tendency to find uncertain or unpredictable situations distressing and to react negatively to not knowing what may happen. Vertical individualism is a cultural orientation that values personal independence and competition while accepting inequality in status, achievement, and power.
Further analysis indicated that, among women, vertical individualism fully mediated the relationship between problematic generative AI use and narcissism. For men, vertical individualism and intolerance of uncertainty partially mediated the relationship. Study authors proposed that narcissism leads to more pronounced vertical individualism and that this pronounced vertical individualism makes an individual prone to problematic use of generative AI.
“Overall, the findings highlight the central role of individualistic value orientations and uncertainty-related processes in explaining how dark personality traits contribute to PGAIU [problematic use of generative AI],” the study author concluded.
The study contributes to the scientific understanding of problematic use of generative AI. However, it should be noted that the design of the study does not allow any causal inferences to be derived from the results. Additionally, all data were based on self-reports, leaving the possibility that reporting bias might have affected the results. Finally, because the average levels of problematic AI use in this sample were relatively low, the findings may reflect variations in general engagement rather than clinically significant problematic use.
The paper, “Dark Tetrad traits and problematic generative artificial intelligence use: the serial mediating role of individualism and intolerance of uncertainty,” was authored by Kağan Kırcaburun.