Dark personality traits linked to generative AI use among art students

A new study published in BMC Psychology sheds light on the psychological and behavioral factors that may be influencing how university art students in China use generative artificial intelligence tools. The research found that students who scored higher on personality traits like narcissism, Machiavellianism, psychopathy, and materialism were more likely to engage in academic misconduct, experience academic anxiety, procrastinate, and ultimately rely more heavily on tools like ChatGPT and Midjourney. These behaviors were also associated with increased frustration and negative thinking.

The study was grounded in social cognitive theory, a psychological framework that emphasizes how personal characteristics, behaviors, and environmental factors interact. The researchers focused on a group of university art students in Sichuan province, a population that faces a unique set of challenges. These include high levels of competition, expectations to produce both technically strong and original creative work, and the increasing influence of generative artificial intelligence in their fields.

The researchers began with an interest in whether certain negative personality traits—commonly referred to as “dark traits”—could help explain patterns of academic misconduct and psychological stress. These traits include narcissism (a heightened sense of self-importance), Machiavellianism (manipulativeness and strategic exploitation of others), psychopathy (a lack of empathy and impulsivity), and materialism (a strong focus on acquiring wealth or status symbols).

Prior studies have linked these traits to dishonest behavior, but the research team wanted to explore these dynamics within the specific context of art education, where creativity is often difficult to evaluate and originality is highly prized.

To conduct the study, researchers surveyed 504 students from six major art-focused universities in Sichuan. The sample was diverse in terms of artistic discipline, including students from visual arts, music, dance, and drama programs. Participants were recruited using a stratified sampling method to ensure representative coverage across schools and artistic specialties. Data collection occurred through both in-person and online surveys. Before the main survey, a pilot test with 30 students was conducted to refine the wording and structure of the questionnaire.

Free daily newsletter

Students completed standardized self-report measures assessing their personality traits, experiences of academic anxiety, frequency of procrastination, levels of frustration and negative thinking, and generative AI usage habits. The researchers used translated and validated versions of existing psychological scales to ensure the accuracy and cultural relevance of the survey. They then applied a statistical technique called structural equation modeling to examine how the variables were related to one another.

The results showed clear patterns. Students who scored higher on dark personality traits were significantly more likely to engage in academic misconduct. This misconduct included behaviors such as plagiarism and misrepresenting AI-generated work as their own. These students also reported higher levels of anxiety about their academic performance and a greater tendency to put off assignments. These behaviors, in turn, were linked to increased feelings of frustration, persistent negative thinking, and a stronger reliance on generative AI tools to complete academic tasks.

The researchers found that of the four personality traits measured, narcissism, Machiavellianism, and psychopathy had the strongest associations with misconduct-related behaviors. For example, students high in narcissism may cheat to maintain their self-image or achieve recognition. Those high in Machiavellianism may view academic dishonesty as a strategic way to gain an advantage. Psychopathy was associated with impulsive behavior and a lack of remorse, which may explain its link to dishonest practices.

Materialism also played a role. Students who strongly valued material success were more likely to cut corners to achieve high grades or awards, suggesting that external rewards can be a strong motivator for dishonest behavior.

Academic anxiety and procrastination emerged as important mediating factors in the model. Students who were anxious about their performance were more prone to negative thinking and reported more frustration with their academic experience. Procrastination added to these problems by creating time pressure and reinforcing avoidance behaviors. These psychological pressures appeared to increase the likelihood that students would turn to generative AI tools for assistance.

The researchers highlighted that reliance on AI tools was not limited to students seeking help for legitimate reasons. Rather, it often reflected a broader pattern of behavior driven by personality traits, stress, and a lack of self-regulation. Students who were already engaging in misconduct or experiencing academic distress were more likely to depend on AI technologies as a coping mechanism.

One strength of the study is its focus on art students, a population often overlooked in discussions of academic misconduct. These students face unique challenges, particularly when new technologies like generative AI blur the boundaries between original creation and automated production. The findings may help inform institutional policies in other creative disciplines facing similar issues.

However, the study also has some limitations. It relied entirely on self-report measures, which can be subject to bias. Students may have underreported dishonest behaviors or overestimated their use of AI tools. The cross-sectional design of the research also means that the observed associations cannot be interpreted as direct evidence of causation. Longitudinal studies following students over time would help clarify how these relationships evolve and whether early personality traits predict later behaviors.

While the study does not establish direct cause-and-effect relationships, it does suggest a network of associations that educators and administrators may want to consider. The use of generative AI in academic settings is growing rapidly, and the researchers argue that it is important to understand not only how students are using these tools but also why.

The study, “Dark personality traits are associated with academic misconduct, frustration, negative thinking, and generative AI use habits: the case of Sichuan art universities,” was authored by Jingyi Song and Shuyan Liu.