Scholarlink

<p>This study examined the impact of generative AI use on university students’ creative self-efficacy—a key motivational construct within the broader domain of creative cognition—through a moderated parallel mediation model. Drawing on the dual-process model of creativity and self-efficacy construct, we investigated divergent and convergent thinking as mediators and creativity traits as a moderator. Data were collected from 963 Chinese undergraduate and graduate students via an online survey. Parallel mediation analysis revealed partial mediation: generative AI use positively predicted creative self-efficacy directly and indirectly through both divergent and convergent thinking. Moderation analyses showed that creativity traits significantly strengthened the positive effects of generative AI use on divergent and convergent thinking at higher levels, with negative or attenuated effects at lower levels. Moderated mediation confirmed conditional indirect effects, yielding strong positive pathways for high-trait students and negative pathways for low-trait individuals—a “double-edged sword” pattern. These findings highlight the non-universal benefits of generative AI, underscoring individual differences as a critical boundary condition. Theoretical implications extend creativity models to human–AI contexts, while practical recommendations advocate trait-sensitive AI integration in higher education to promote equitable creative development.</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

Amplifier or inhibitor? A moderated mediation analysis of generative AI’s impact on creative self-efficacy through dual-process thinking

  • 期刊论文

  • 发表日期：

    2026年6月15日

查看全文
  • Puyan Jin

摘要

This study examined the impact of generative AI use on university students’ creative self-efficacy—a key motivational construct within the broader domain of creative cognition—through a moderated parallel mediation model. Drawing on the dual-process model of creativity and self-efficacy construct, we investigated divergent and convergent thinking as mediators and creativity traits as a moderator. Data were collected from 963 Chinese undergraduate and graduate students via an online survey. Parallel mediation analysis revealed partial mediation: generative AI use positively predicted creative self-efficacy directly and indirectly through both divergent and convergent thinking. Moderation analyses showed that creativity traits significantly strengthened the positive effects of generative AI use on divergent and convergent thinking at higher levels, with negative or attenuated effects at lower levels. Moderated mediation confirmed conditional indirect effects, yielding strong positive pathways for high-trait students and negative pathways for low-trait individuals—a “double-edged sword” pattern. These findings highlight the non-universal benefits of generative AI, underscoring individual differences as a critical boundary condition. Theoretical implications extend creativity models to human–AI contexts, while practical recommendations advocate trait-sensitive AI integration in higher education to promote equitable creative development.