Scholarlink

<p>Generative artificial intelligence (GenAI) poses new challenges to academic integrity by enabling students to produce plausible academic work with limited detectable overlap. Using a two-wave prospective panel design with a six-month interval, we examined whether Dark Tetrad traits predicted GenAI academic misconduct at Time 2 after controlling for Time 1 misconduct. University students in Taiwan completed measures of Machiavellianism, narcissism, psychopathy, and sadism at Time 1 and a measure of GenAI academic misconduct at Time 1 and Time 2 (Time 1: <i>N</i> = 714; Time 2: <i>n</i> = 654). Missing Time 2 data were imputed using missForest, yielding an analytic sample of 714 participants. Hierarchical multiple regression analyses showed that psychopathy was positively associated with subsequent misconduct, whereas the other dark traits did not exhibit unique main effects. A modest interaction between Machiavellianism and psychopathy attenuated the association between psychopathy and misconduct at higher levels of Machiavellianism. These findings suggest that researchers should consider theoretically grounded risk correlates—without treating them as deterministic—when seeking to understand and prevent academic dishonesty in the GenAI era.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

The dark side of generative AI: a two-wave prospective panel study of the Dark Tetrad in academic misconduct

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年8月5日

查看全文
  • Rongjian Sun,
  • Cheng-Yen Wang

摘要

Generative artificial intelligence (GenAI) poses new challenges to academic integrity by enabling students to produce plausible academic work with limited detectable overlap. Using a two-wave prospective panel design with a six-month interval, we examined whether Dark Tetrad traits predicted GenAI academic misconduct at Time 2 after controlling for Time 1 misconduct. University students in Taiwan completed measures of Machiavellianism, narcissism, psychopathy, and sadism at Time 1 and a measure of GenAI academic misconduct at Time 1 and Time 2 (Time 1: N = 714; Time 2: n = 654). Missing Time 2 data were imputed using missForest, yielding an analytic sample of 714 participants. Hierarchical multiple regression analyses showed that psychopathy was positively associated with subsequent misconduct, whereas the other dark traits did not exhibit unique main effects. A modest interaction between Machiavellianism and psychopathy attenuated the association between psychopathy and misconduct at higher levels of Machiavellianism. These findings suggest that researchers should consider theoretically grounded risk correlates—without treating them as deterministic—when seeking to understand and prevent academic dishonesty in the GenAI era.