本研究旨在透過系統性文獻分析，回顧過去三個主要科學教育期刊International Journal of Science Education (IJSE)、Journal of Research in Science Teaching (JRST)及Science Education (SE)在人工智慧相關研究的發展趨勢，分析主要分成整體發展趨勢、實證研究發展趨勢、及非實證研究所探討的議題。本研究經有系統的篩選後針對25篇論文分析，結果顯示這些論文的時間分布大致反映出AI的浪潮。而25篇中的15篇有實證研究，實證研究發展趨勢主要依人工智慧應用類型、使用的人工智慧技術或工具、研究設計、研究方法、教育程度面向分析及報告，結果顯示人工智慧在學習者側寫及預測、評量及評估兩個應用類型最多，自然語言處理及決策樹為最常使用的人工智慧技術，實驗設計為最常使用的研究設計，傳統測驗為最常使用的研究方法，而國中為最常被調查的教育程度。而25篇中的10篇沒有實證研究，這10篇文章主要依以下三個主題討論及報告：人工智慧對科學教育評量的潛在挑戰與省思、人工智慧對科學教學與學習的潛在挑戰與省思、人工智慧對科學教育學術研究和期刊的潛在挑戰與省思。經由實證研究及非實證研究論文的分析及討論，本研究最後亦針對未來研究與科學教育方面提供幾點參考建議。

This study aims to understand the research trends of artificial intelligence in science education by analyzing publications in three selected journals: International Journal of Science Education (IJSE), Journal of Research in Science Teaching (JRST) and Science Education (SE). The analysis is mainly divided into overall development trends, the trends of empirical research, and issues discussed in non-empirical research. This study analyzed 25 papers after systematic screening. The results showed that the time distribution of these papers roughly reflects the wave of AI. Fifteen of the 25 articles have empirical research. The development trend of empirical research is reported mainly based on the type of artificial intelligence application, the artificial intelligence technology or tools used, research design, research methods, and education level. The results show that profiling and prediction, and assessment and evaluation are two most common applications. Natural language processing and decision trees are the most commonly used artificial intelligence technologies. Experimental design is the most commonly used research design. Traditional testing is the most commonly used research method. Junior high school is the most commonly surveyed education level. Ten of the 25 articles have no empirical research. These 10 articles mainly discuss and report on the following three themes: potential challenges and reflections of artificial intelligence on science education evaluation, potential challenges and reflections of artificial intelligence on science teaching and learning. Potential challenges and reflections of artificial intelligence on science education academic research and journals. Through the analysis and discussion of empirical research and non-empirical research papers, this study finally provides some suggestions for future research and science education.

人工智慧、文獻回顧、科學教育

Artificial Intelligence, Literature Review, Science Education