A recent academic investigation has shed light on a concerning correlation: individuals exhibiting elevated levels of narcissism are more susceptible to developing problematic habits when interacting with generative artificial intelligence. This susceptibility is not uniform, with cultural orientations such as vertical individualism and personal intolerance of uncertainty playing crucial mediating roles. The study, detailed in BMC Psychology, provides a deeper understanding of the psychological mechanisms underpinning excessive and potentially harmful engagement with advanced AI systems.

Problematic use of generative AI describes an engagement pattern characterized by overuse, diminished control, and detrimental impacts on an individual's daily existence. This could manifest as dedicating excessive time to AI interactions, persistently using these tools despite adverse outcomes, or feeling incapable of moderating their usage. Such over-reliance on generative AI might extend to professional duties, educational pursuits, decision-making processes, creative endeavors, social interactions, or emotional support. These patterns can disrupt academic achievements, professional responsibilities, personal relationships, sleep quality, and other vital life aspects. Some individuals may increasingly depend on AI to evade challenging tasks, ambiguities, monotony, solitude, or distressing emotions. This problematic behavior might also include compulsive checks, repetitive prompts, or continuous engagement, mirroring other forms of unhealthy digital technology usage.

An additional concern is the potential for cognitive overreliance, where users increasingly transfer their cognitive functions, such as thinking, writing, problem-solving, or judgment, to AI systems. This can lead to a reduced willingness to undertake these activities independently. It is important to distinguish problematic use from frequent use; intensive AI engagement can be beneficial and productive as long as it remains controlled and does not cause impairment. Although the concept is still evolving, problematic AI use generally refers to engagement marked by impaired control, functional detriment, psychological dependence, or continued use despite significant negative repercussions.

Kağan Kırcaburun, an assistant professor at Düzce University in Turkey, initiated this research to examine the connection between malevolent personality traits and problematic generative AI use. His particular interest lay in exploring how individualistic cultural perspectives and intolerance of uncertainty might influence this relationship. Kırcaburun highlighted that while previous research had established links between problematic technology use and adverse personality traits, only a limited number of recent studies had specifically investigated this connection within the context of generative AI.

The study involved an online survey conducted through Prolific, engaging 677 participants who confirmed active use of generative AI. The average age of the participants was approximately 40 years, with an equal gender distribution. The survey included assessments designed to measure problematic generative AI use (using the Problematic GAI Use Scale, adapted from a scale for conversational AI), intolerance of uncertainty (the Intolerance of Uncertainty Scale), individualistic orientation (the Individualism and Collectivism Scale), and various negative personality traits (the Dark Triad Dirty Dozen Scale and the Short Sadistic Impulses Scale).

Adverse personality traits are character flaws linked to self-interest, manipulation, emotional detachment, and disregard for others. These traits include Machiavellianism (strategic manipulation and exploitation), narcissism (grandiosity, entitlement, and need for admiration), psychopathy (callousness, low empathy, and impulsivity), and sadism (deriving pleasure from others' humiliation or pain).

The research outcomes revealed that participants exhibiting more pronounced problematic use of generative AI also tended to score higher on narcissism, and marginally higher on Machiavellianism and sadism. Furthermore, a greater degree of problematic generative AI use correlated with increased intolerance for uncertainty and a higher level of vertical individualism. Intolerance of uncertainty refers to a predisposition to find ambiguous or unpredictable situations distressing and to react negatively to a lack of clear outcomes. Vertical individualism, on the other hand, is a cultural viewpoint that highly values personal autonomy and competition while accepting disparities in status, achievements, and power.

Further analysis showed that for women, vertical individualism completely mediated the relationship between problematic generative AI use and narcissism. For men, this relationship was partially mediated by both vertical individualism and intolerance of uncertainty. The research team hypothesized that narcissism leads to a more pronounced vertical individualism, which, in turn, increases an individual's propensity for problematic generative AI use. “These findings collectively underscore the significant role of individualistic value systems and processes related to uncertainty in elucidating how detrimental personality characteristics contribute to problematic AI usage,” the study's author concluded.

This research significantly enhances the scientific understanding of problematic generative AI use. However, it is crucial to acknowledge that the study's design precludes drawing definitive causal conclusions from its findings. Additionally, all data relied on self-reports, introducing the potential for reporting biases to influence the results. Lastly, given the relatively low average levels of problematic AI use observed in this sample, the findings might primarily reflect variations in general engagement rather than clinically significant instances of problematic behavior.