期刊目錄列表-第三十九期【教育評鑑研究期刊】(115年6月出版)

提示詞素養的概念建構、層次化框架與制度化培育意涵：回顧式概念分析取向 [詳全文]

作者：張明文、陳國生、陳建志

刊期：第39期
日期：2026年6月
頁碼：1-28
DOI：10.6866/JEAE

摘要：

隨著生成式人工智慧與大型語言模型快速進入教育場域，自然語言互動已成為影響學習歷程與知識建構的重要介面。作為人機互動的核心媒介，提示詞不僅影響生成內容的方向與品質，也關係到生成式AI在教育中能否被負責任地使用。然而，現有研究對提示詞素養的概念界定、能力內涵與發展層次仍呈現高度分散，致使相關成果不易累積，也限制其在課程、評量與教師專業發展上的應用。本研究採回顧式概念分析取向，整合數位素養、AI 素養與人機互動研究，以釐清提示詞素養的理論定位、核心能力構面與層次化理解。研究結果顯示，提示詞素養宜定位為建立在數位素養與AI 素養之上的人機協作導向高階素養，其核心能力可操作化為提示設計、互動調節與批判評估三個相互連動的構面。進一步而言，本研究提出三層次提示詞素養框架，用以說明基礎素養、互動操作與高階反思之間的動態關係，並據此分析其在課程設計、教師專業發展、評量機制與治理支持上的制度化培育意涵。本研究提供一個具理論整合性的概念框架，可作為後續研究與教育實務之分析基礎。

關鍵詞：人工智慧素養；人機協作；大型語言模型；生成式人工智慧；提示詞素養

Journal of Educational Administration and Evaluation, June 2026, Vol. 39

Conceptualizing Prompt Literacy: A Layered Framework and Implications for Institutional Cultivation—A Conceptual Review Approach

Author: Ming-Wen Chang, Kuo-Sheng Chen & Chien-Chih Chen

Vol.&No.: Vol. 39
Date: June 2026
Pages: 1-28
DOI: 10.6866/JEAE

Abstract:

This study examined prompt literacy in educational contexts shaped by generative artificial intelligence and large language models. It adopted a conceptual review approach and integrated scholarship on literacy, digital literacy, AI literacy, and human-AI interaction to clarify the concept's theoretical position, core competence dimensions, and layered development. The review found that prompt literacy was not merely a technique for composing longer or more detailed prompts. Rather, it functioned as an advanced, human-AI-collaboration-oriented literacy built upon digital and AI literacy. The analysis identified three interrelated competence dimensions: prompt design, interaction regulation, and critical evaluation. Prompt design referred to translating learning goals into workable task specifications; interaction regulation referred to strategic adjustment across multiple rounds of dialogue; and critical evaluation referred to examining, revising, and taking responsibility for AI-generated outputs. Based on these findings, the study proposed a three-level framework that connected foundational digital and AI literacy, generative-AI interaction and prompt operation, and higher-order reflection. The framework showed that the levels were not a linear hierarchy but a dynamic relationship in which foundational understanding supported interaction, interaction generated evidence for evaluation, and higher-order reflection fed back into subsequent prompt design and AI understanding. The review also distinguished horizontal competence dimensions from vertical developmental levels, thereby reducing conceptual overlap among prompt design, AI literacy, and broader digital literacy. This distinction was intended to strengthen its use as a shared reference for educational research and practice. The study further discussed implications for institutional cultivation in curriculum design, teacher professional development, assessment, and governance support. It suggested that prompt literacy should be cultivated through subject-based
tasks, teacher-guided reflection, process-oriented assessment, and equitable governance conditions rather than through isolated prompt-writing tips. The contribution of this study lay in offering an integrated conceptual framework that could support future empirical research, curriculum development, and educational policy discussion on responsible generative AI use.

Keywords: AI literacy; human-AI collaboration; large language models; generative artificial intelligence; prompt literacy