AI在教育研究領域的應用系列（十八）：運用Elicit增益系統性文獻回顧之PRISMA建構流程

Applications of AI in Educational Research Series (XVIII): Using Elicit to Enhance the PRISMA Construction Process for Systematic Reviews

系統性文獻回顧作為實證研究整合的重要方法論取徑，其建構流程需耗費大量時間與人力成本。PRISMA 2020聲明雖提供國際標準化框架，但文獻檢索、篩選、資料萃取與報告撰寫等環節仍高度仰賴人工操作。本文以PRISMA 2020流程圖的辨識、篩選與納入三大區段為分析架構，探討Elicit的Gather功能增益各階段流程之應用，並逐步示範研究問題界定、多策略並行搜尋、布林邏輯檢索、AI輔助篩選、全文評估、資料萃取及PRISMA報告自動生成等具體操作程序。本研究認為，Elicit的自動化文件功能、查詢優化引導機制與句子層級引文設計，可有效降低教育研究者建構PRISMA流程的技術門檻；同時，現有實證支持AI工具作為輔助角色而非替代方案的定位，研究者在關鍵判斷節點仍需保留專業決策權。

Systematic literature review (SLR), as an essential methodological approach for integrating empirical research evidence, requires substantial time and human resources. Although the PRISMA 2020 statement provides an internationally standardized framework, key stages– such as literature searching, screening, data extraction, and report writing– still rely heavily on manual work. Using the three main sections of the PRISMA 2020 flow diagram (identification, screening, and inclusion) as an analytical structure, this paper examines how Elicit’s Gather feature can enhance each stage of the process. It also offers a step-by step demonstration of concrete procedures, including research question formulation, parallel multi-strategy searching, Boolean logic retrieval, AI-assisted screening, full-text assessment, data extraction, and automated generation of PRISMA reports. This study argues that Elicit’s automated document functions, query-optimization guidance mechanisms, and sentence level citation design effectively lower the technical barriers for educational researchers in implementing the PRISMA workflow. At the same time, existing empirical evidence supports positioning AI tools as assistive rather than substitutive solutions; researchers should retain professional decision-making authority at critical judgment points.