Scholarlink

<p>Moving beyond conventional AI efficacy measures, we constructed the Generative Artificial Intelligence Self-Efficacy Scale (GAI-SES) to specifically capture both task-oriented and social dimensions of user confidence. We validated the scale with 933 participants from Taiwan. Via exploratory and confirmatory factor analyses, a bi-factorial model: GAI-assisted Task Self-Efficacy and GAI-assisted Social Self-Efficacy were established. In addition to the GAI-SES, this study employed the General Self-Efficacy Scale (GSES), the AI Self-Efficacy Scale (AISES), and the 18-item Psychological Well-Being Scale (PWB) to examine criterion validity and associations with psychological well-being. The GAI-SES showed good internal consistency, test–retest reliability, and satisfactory construct, convergent, discriminant, and nomological validity, correlating moderately with AI self-efficacy and weakly with general self-efficacy. Crucially, our findings link GAI self-efficacy to better psychological well-being. GAI self-efficacy did not differ significantly by gender, suggesting a potentially inclusive landscape in generative AI use. This study offers a validated tool for assessing GAI self-efficacy, supporting future research on its psychological and educational impacts.</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

Development and validation of the Generative AI Self-Efficacy Scale (GAI-SES): psychometric properties, gender differences, and well-being associations

  • 期刊论文

  • 发表日期：

    2026年2月7日

查看全文
  • Sen-Chi Yu,
  • An-Chia Liu,
  • Hung-Bin Sheu

摘要

Moving beyond conventional AI efficacy measures, we constructed the Generative Artificial Intelligence Self-Efficacy Scale (GAI-SES) to specifically capture both task-oriented and social dimensions of user confidence. We validated the scale with 933 participants from Taiwan. Via exploratory and confirmatory factor analyses, a bi-factorial model: GAI-assisted Task Self-Efficacy and GAI-assisted Social Self-Efficacy were established. In addition to the GAI-SES, this study employed the General Self-Efficacy Scale (GSES), the AI Self-Efficacy Scale (AISES), and the 18-item Psychological Well-Being Scale (PWB) to examine criterion validity and associations with psychological well-being. The GAI-SES showed good internal consistency, test–retest reliability, and satisfactory construct, convergent, discriminant, and nomological validity, correlating moderately with AI self-efficacy and weakly with general self-efficacy. Crucially, our findings link GAI self-efficacy to better psychological well-being. GAI self-efficacy did not differ significantly by gender, suggesting a potentially inclusive landscape in generative AI use. This study offers a validated tool for assessing GAI self-efficacy, supporting future research on its psychological and educational impacts.