大專校院學生人工智慧素養指標建構之分析研究

Abstract

隨著人工智慧（artificial intelligence, AI）技術迅速發展並廣泛應用於教育領域，人工智慧素養（artificial intelligence literacy, AIL）已成為高等教育課程設計與政策規劃中不可忽視的核心議題。為因應數位轉型與AI驅動之未來社會需求，本研究旨在建構一套適用於大專校院學生之人工智慧素養（AIL）指標架構，作為未來課程設計、教學實施與素養評量之依據，期能促進學生在人工智慧相關知識、技能與態度等面向之整體素養發展，並提供教育實務與政策制定之參考。本研究採用結構方程模型進行驗證，結果顯示，整體模型之配適度達學術標準（GoF = .652）。在六項研究假設中，有五項獲得支持。首先，AI素養對AI自我效能具有正向影響，且對AI自我管理能力亦呈正向影響。其次，AI自我效能對創造AI具正向作用，並且對AI自我管理能力具有正向影響。相對地，AI素養對創造AI的直接效果未達顯著，顯示學生的AI知識理解並不必然轉化為創造性AI應用的實作表現。此外，創造AI對AI自我管理能力呈顯著負向影響（路徑係數 = -0.166，t = 2.332），顯示大專校院學生在投入AI創造任務時，可能因資源配置過度集中於技術執行，反而犧牲了學習歷程中的自我調節與管理能力。綜合上述，本研究同時發現AI自我效能在AI素養與創造AI，以及AI自我管理能力之間扮演中介角色，突顯學生之自我效能與信念是影響AI應用表現與自我管理能力的核心驅動因素。本研究所建構之AI素養指標，不僅符應教育部與國際組織對AI教育政策之推動，亦可作為各大專校院發展AI課程設計、學生能力評估與素養認證之實務參考，具高度應用價值與發展潛能。

In response to the rapid advancement of Artificial Intelligence (AI) and its growing integration into education, AI literacy has become a critical issue in higher education curriculum development and policy planning. This study aimed to construct an AI Literacy (AIL) indicator framework for postsecondary students, encompassing the core dimensions of apply AI, understand AI, evaluate AI, AI ethics, and create AI, as well as related psychological competencies including AI self-efficacy and AI self-management.Using structural equation modeling (SEM), the study validated a conceptual model based on 262 valid responses. The results showed an acceptable model fit (GoF = .652), with five out of six hypotheses supported. AI literacy had a significant positive effect on both AI self-efficacy and AI self-management. AI self-efficacy further positively influenced both create AI and AI self-management. Additionally, create AI significantly impacted AI self-management. However, the direct path from AI literacy to create AI was not statistically significant, suggesting that students’ understanding and application of AI knowledge may not automatically translate into creative AI production. Moreover, create AI showed a significant negative effect on AI self-management (β = -0.166, t = 2.332), implying that high levels of engagement in AI creation may hinder self-regulatory capabilities under certain conditions. AI self-efficacy was found to mediate the relationships among AI literacy, create AI, and AI self-management, highlighting its role as a central driver of students’ AI performance and self-regulation.The AIL indicator framework developed in this study provides both theoretical insight and practical guidance for AI-related curriculum design, competency assessment, and literacy certification in higher education institutions.