作為軟體開發的基礎學科，統計學雖至關重要，但傳統教學經常面臨理論與實踐脫節之挑戰。對此，目前仍少有研究探討如何將生成式AI系統性地整合於專題導向學習中，以解決統計程式開發的高認知門檻問題。為因應AI時代的教育需求，本研究以「統計方法與軟體」課程為場域，導入專題導向學習（project-based learning, PBL）、統計軟體競賽與小組互評機制，並核心融入生成式AI技術，擬探討之具體問題為：整合生成式AI技術、專題導向學習與競賽機制的教學模式，是否能顯著提升學生的統計軟體開發實作能力與課程投入度？本研究對象為修習該課程的學生，最終分析納入22份有效配對樣本。研究採混合研究方法，於為期18週的學期中進行實證，運用前、後測問卷、課程回饋單、小組互評及業界專家講評等多元工具蒐集數據。量化分析結果顯示，教學介入後學生在統計學基本認知、軟體操作與程式基礎及AI技術認知與課程投入三個能力構面上有顯著提升（p < .05）。其中，「軟體操作與程式基礎」構面進步最為顯著，效果量達到非常大的程度（Hedges’ g = 1.040）。學生對課程整體滿意度達4.86分，且從期末產出之「樣本統計量與圖表生成器」與「資料視覺化工具」中可觀察到學生已能運用AI生成複雜程式碼並解決真實問題，此具體作品成果佐證了AI輔助成功降低了程式學習障礙，而非僅止於主觀滿意度的提升。總結而言，本研究不僅證實生成式AI作為認知鷹架能有效降低程式學習門檻，更填補了新興科技融入傳統統計課程的實證缺口，確立了AI賦能PBL教學模式的有效性。

While statistics is foundational to software development, traditional instruction often suffers from a gap between theory and practice. Furthermore, limited empirical research addresses how to systematically integrate Generative AI into project-based learning (PBL) to lower the high cognitive barrier of statistical programming. This study integrated Generative AI, PBL, statistical software competitions, and peer assessment into a university-level “Statistical Methods and Software” course to evaluate its impact on students’ practical development capabilities and engagement. Using a mixed-methods approach over an 18-week semester, data were collected from 22 paired student samples using questionnaires, feedback forms, peer assessments, and industry expert evaluations. Quantitative analysis revealed significant postintervention improvements (p < .05) across three competency dimensions: basic statistical cognition, software operation and programming foundation, and AI technology cognition and engagement. Notably, the programming foundation dimension showed the greatest improvement with a very large effect size (Hedges’ g = 1.040). Overall course satisfaction reached 4.86 out of 5. Final student projects, including “sample statistic and chart generator” and “data visualization tools”, demonstrated that AI assistance effectively reduced programming learning barriers. This study confirms that Generative AI serves as an effective cognitive scaffold in PBL, bridging the empirical gap in applying emerging technologies to traditional statistics education.

生成式AI、專題導向學習、統計軟體競賽、統計教育

generative AI, project-based learning, statistical software competition, statistics education