本研究以「明日閱讀習慣暨家校閱讀博覽會」之閱讀習慣認證活動為場域，設計AI輔助面談系統，協助國小學生在正式面談前回顧閱讀經驗，並提供評審整理資料及準備問題時所需的參考資訊。學生端以題庫引導對話，再由AI依學生回答進行延伸追問；面談完成後，系統產生摘要與面談追問建議，供評審於正式面談前使用。正式追問與認證判斷仍由評審負責。 研究採系統設計與質性個案分析取向。初始版本實際應用於第一屆活動，共納入9位完成AI面談的學生，其中7位另有正式面談紀錄，並蒐集3位評審的使用回饋。分析資料包括學生AI面談紀錄、AI摘要、面談追問建議、正式面談資料及評審回饋。第一屆活動完成後，再依實際使用與維護過程發現的問題進行系統重構，將原本分開的學生端與評審端功能整合至既有活動平台。 分析結果發現，9位學生皆至少在一項延伸追問中補充前一回答未出現的資訊，但補充內容與程度並不相同。AI摘要能集中整理多輪對話中的閱讀資訊，面談追問建議也可提供部分備題方向；不過，摘要仍可能出現遺漏、過度概括或與原始回答不完全一致的內容，建議問題也可能較為一般。研究結果顯示，AI較適合作為正式面談前的輔助工具，由AI協助提問、整理文字與提供問題方向，評審則依學生閱讀資料及現場回答進行追問與認證判斷。改良版本已完成主要功能重構、資料串接及平台整合，其實際使用情形仍待後續驗證。;This study designed an AI-assisted interview system for an elementary school reading habit certification program. Before the formal interview， the system used a question set and AI-generated follow-up questions to help students recall reading experiences. It then produced summaries and suggested questions for reviewers. Formal questioning and certification decisions remained the reviewers’ responsibility. The study adopted a system design and qualitative case analysis approach. The initial version was used by nine students， seven of whom also had formal interview records. Feedback was collected from three reviewers. The data included AI interview records， summaries， suggested questions， formal interview records， and reviewer feedback. After the first program cycle， the student-side and reviewer-side functions were rebuilt and integrated into the existing platform. All nine students added new information during at least one follow-up exchange， although the content varied. The summaries and suggested questions helped reviewers organize information and prepare questions， but some summaries omitted details or overgeneralized student responses. The findings indicate that AI is more suitable as a pre-interview support tool than as an evaluator. System and data integration have been completed in the revised version， but its practical use still requires further evaluation.
