跳到主要內容區
教學發展中心
  1. 首頁
  2. AI專區 AI Hub

AI教學指引 AI Teaching Guidelines

AI 教學指引

AI 教學指引AI TEACHING GUIDELINES

整理自本校《AI 倫理聲明與學習評量實施指引》（2026.09.01）。Summarised from the NTHU Implementation Guidelines on AI Ethics Statements and Learning Assessment (1 September 2026).
透明TRANSPARENCY
教師在課程中訂明 AI 使用規範；師生使用 AI 時誠實揭露。
負責RESPONSIBILITY
AI 只是素材來源之一。師生要能檢核內容的正確性，並對自己產出的內容負責。
有意識的設計教學INTENTIONAL DESIGN
依預計學習成果設計教學活動與評量，而不是沿用既有慣例。
▲ 三項核心原則 Three core principles
1
避免過度依賴 AIAvoid over-reliance on AI
用「活動覺察、產出評估、能力維持」三個判準，檢視學生怎麼用 AI。
Check how students use AI with three criteria: activity awareness, output assessment, competence maintenance.
看詳細 More ↓
2
AI 時代的評量設計Assessment in the age of AI
評量雙軌：控制型限制用 AI，非控制型可用但要說明。不靠 AI 偵測工具。
Two lanes: controlled (AI restricted) and uncontrolled (AI allowed, with explanation). Don’t rely on AI detectors.
看詳細 More ↓
3
大綱寫入倫理聲明Put an AI statement in the syllabus
完全開放、有條件開放、禁止使用，三選一寫進大綱。附三份範本可參考修改。
Choose unrestricted, conditional or prohibited use. Three templates included.
看詳細 More ↓
4
教師角色轉變The instructor’s changing role
從「提供答案」轉向「示範判斷」，讓學生看見專業判斷如何形成。
From giving answers to modelling judgement: show students how expert judgement is formed.
看詳細 More ↓
AI 時代的教學，是讓學生用了 AI 也真的學到東西，而非禁止學生使用 AI。Teaching in the age of AI means helping students genuinely learn even when they use AI, rather than banning it.
1

避免過度依賴 AI 造成認知卸載AVOID COGNITIVE OFFLOADING

學生需要判斷「何時該自己完成、何時可以尋求 AI 協助」。教師不只教學生怎麼用 AI，也要引導他們理解何時用、為何用，以及何時停下來自己判斷。
Students need to judge when to do the work themselves and when to seek AI help. Instructors teach not only how to use AI, but when and why to use it, and when to stop and decide for themselves.

學生與 AI 的三種協作模式Three modes of student–AI collaboration

AI 主導監控AI-led monitoring
AI 提供大量回饋與引導，學生接受並調整。
AI gives extensive feedback; students accept and adjust.
留意：學生可能逐漸依賴 AI。Watch for: growing dependence on AI.
學生主導、AI 輔助Student-led, AI-assisted
學生設定目標、主動決策；AI 負責檢核、回饋與挑戰。
Students set goals and decide; AI checks, gives feedback and challenges.
留意：需要較高的後設認知。Watch for: requires strong metacognition.
動態協商Dynamic negotiation
依學習階段，AI 與學生交替承擔不同角色。
AI and students swap roles as learning progresses.
留意：需要教師有意識地設計歷程。Watch for: needs intentional design by the instructor.
三種模式不互斥。目標是讓學生從「接受 AI 給的答案」，轉向「用 AI 協助自己判斷」，並對學習目標、重要判斷與最終成果負責。
The modes are not mutually exclusive. The goal is for students to move from accepting AI’s answers to using AI to support their own judgement, while remaining responsible for their goals, key decisions and final work.

有意識使用 AI 的三個判準Three criteria for intentional AI use

1 活動覺察Activity awareness
分清楚哪些步驟是自己完成、哪些由 AI 協助；可請學生說明 AI 在構思、分析、寫作或修訂中的角色。
Know which steps you did and which AI assisted; students can explain AI’s role in ideation, analysis, writing or revision.
2 產出評估Output assessment
不照單全收：查證資訊、找出錯誤或缺漏、比較不同產出，判斷是否適用於情境。
Don’t accept answers as-is: verify, find errors and gaps, compare outputs, and judge whether they fit the context.
3 能力維持Competence maintenance
即使 AI 能代勞，仍保留獨立思考、實作與能力驗證的機會。
Even when AI can do it, keep opportunities for independent thinking, hands-on practice and verification.

過度依賴會扭曲學生對自己能力的認知Over-reliance distorts students’ sense of their own ability

AI 安慰劑效應AI placebo effect
在 AI 協助下完成任務後，學生高估自己的能力，把 AI 的協助誤認為自己的進步。
After finishing tasks with AI, students overestimate their own ability and mistake AI’s help for their own progress.
AI 代筆效應AI ghostwriter effect
即使知道 AI 貢獻很大，學生仍把產出當成自己的成果，對內容的所有權、責任感與批判檢視都降低。
Even knowing AI contributed heavily, students still treat the output as their own, weakening ownership, responsibility and critical review.
長期下來會形成「技能侵蝕的惡性循環」A vicious cycle of skill erosion
認知工作交給 AIThinking handed to AI→技能退化Skills erode→更依賴 AIMore reliance on AI→無法判斷 AI 產出好壞Can no longer judge AI output

從「使用 AI」轉向「有意識的使用 AI」From using AI to using AI intentionally

使用Use→評估Evaluate→修正Revise→負責Own it
學習不只是「學生用 AI 完成任務」，而是學生使用 AI、評估 AI、修正 AI，並對最終判斷負責。這也能避免 AI 即時又流暢的答案，讓學生省略必要的深度思考。
Learning is not just students completing tasks with AI, but using, evaluating and revising AI output and owning the final judgement. This keeps fast, fluent AI answers from replacing the deep thinking students need.
2

AI 時代的學習評量設計ASSESSMENT DESIGN IN THE AGE OF AI

課後問題集、程式作業、書面報告，AI 都能產出可信度相當高的答案。教發中心建議參考奧克蘭大學、雪梨大學的評量雙軌模式（Two-Lane Approach）：教師有意識地決定，在哪種評量情境下、學生可以用 AI 到什麼程度。
AI can produce credible answers to problem sets, coding tasks and written reports. CTLD recommends the two-lane approach used at the Universities of Auckland and Sydney: instructors decide deliberately how far students may use AI in each assessment.

評量雙軌模式The two-lane approach

Lane 1 控制型評量CONTROLLED
AI 使用AI use 原則上限制使用Generally restricted
目的Purpose 驗證學生能獨立展現學習成果，維護學位的可信度Verify independent achievement and protect the credibility of degrees
評量重點Focus Authenticity & Assurance
典型形式Typical formats 監考考試、口試、現場實作、即席提問的簡報、受監控的數位考試Invigilated exams, orals, live practicals, presentations with Q&A, supervised digital exams
Lane 2 非控制型評量UNCONTROLLED
AI 使用AI use 可以使用、不設限，但必須說明怎麼用Allowed without limits, but students must explain how they used it
目的Purpose 培養與 AI 協作的能力，建立判斷力Build the ability to work with AI and exercise judgement
評量重點Focus Learning & AI capability
典型形式Typical formats 實驗報告、作品集、案例分析、簡報、海報、設計作品、實務模擬Lab reports, portfolios, case analyses, presentations, posters, design work, simulations
同一門課、甚至同一項學習設計，都可以搭配不同條件的評量。
Different lanes can be combined within one course, or even within one learning design.

不要把「增加實體考試」當成唯一解Don’t make more in-person exams the only answer

增加課堂內、限時的高風險評量，雖能降低把思考外包給 AI 的風險，卻也可能減少學生投入問題集與專題所需的長時間思考。本校不建議只用增加監考考試、限時寫作來因應生成式 AI，應兼顧控制型評量、歷程性評量、專題、實作、作品集與口頭答辯。
More timed, high-stakes in-class assessment can reduce cognitive outsourcing, but may also cut the sustained thinking that problem sets and projects require. NTHU does not recommend relying solely on more invigilated exams or timed writing; balance controlled assessment with process-based assessment, projects, practicals, portfolios and oral defences.

避免使用第三方 AI 偵測工具Avoid third-party AI detection tools

✕ 誤判率高High false positives
學生多半只用 AI 構思、修改或潤飾，仍可能被誤判。
Students often use AI only to brainstorm, edit or polish, yet may still be flagged.
✕ 公平性疑慮Fairness concerns
對非英語母語、不同社經與文化背景的學生，可能造成差別影響。
May disproportionately affect non-native English speakers and students from different backgrounds.
✕ 隱私風險Privacy risks
部分工具會蒐集與處理學生資料及可識別資訊。
Some tools collect and process student data and identifiable information.
✕ 傷害互信Erodes trust
結果不宜單獨當作學術不誠信的證據；過度依賴會增加師生間的不信任。
Results should not be the sole evidence of misconduct; over-reliance breeds distrust.
✓ 改為掌握學習歷程Look at the learning process instead
分階段繳交、定期專題討論或個別會談、課堂內寫作與實作、口頭說明與對話，強調「草稿 → 回饋 → 修訂」；程式課可用 GitHub 等版本控制保留撰寫與迭代歷程。
Staged submissions, regular project check-ins or one-to-one meetings, in-class writing and practice, oral explanation, and draft → feedback → revision cycles. Coding courses can use version control such as GitHub to keep a record of iterations.
3

在課程大綱寫入 AI 倫理聲明PUT AN AI ETHICS STATEMENT IN THE SYLLABUS

在大綱中明訂學生使用 AI 的方式，包括如何正確引註、揭露使用歷程，師生與學校處理學術誠信問題時就有依據。
State in the syllabus how students may use AI, including how to cite it and disclose their use. This gives instructors, students and the university a shared basis for handling academic integrity.

三選一：學生可以怎麼用 AIChoose one: how students may use AI

1完全開放使用Unrestricted use
無須註明
No disclosure needed
2有條件開放Conditional use
請註明如何使用生成式 AI 於課程產出（範例一、二）
Students disclose how generative AI was used (Examples 1 and 2)
3禁止使用Prohibited
請註明相關的評量與監管機制（範例三）
State the related assessment and monitoring measures (Example 3)

三份倫理聲明範本Three statement templates

範例 1　有條件開放：註明 AI 如何用於課程產出Example 1 · Conditional use: disclose how AI is used in course outputs
生成式AI倫理聲明 基於透明與負責任原則，本課程鼓勵學生利用AI進行協作或互學，以提升課程產出品質。本課程採取「有條件開放，請註明如何使用生成式AI於課程產出」。學生須於課堂作業或報告中的「標題頁註腳」或「引用文獻後」，簡要說明AI用於議題發想、文句潤飾、結構參考等方式。若經查核使用卻未依規定標明，教師、學校或相關單位有權重新評分或不予計分。教師若於教材或學習資料使用AI，也應於投影片或口頭適當標注。修讀本課程之學生於選課時視為同意以上倫理聲明。
Generative AI Ethics Statement Based on the principles of transparency and responsibility, this course encourages students to use AI for collaboration or co-learning to improve the quality of course outputs. This course adopts a “conditional use with disclosure of how generative AI is used in course outputs” policy. Students must briefly explain, in a footnote on the title page or after the references of course assignments or reports, how AI was used, such as for topic ideation, sentence polishing, or structural reference. If AI use is identified but not disclosed as required, the instructor, university, or relevant unit may re-evaluate the work or assign no credit. If instructors use AI to produce teaching materials or learning resources, they should also appropriately indicate such use in slides or orally. By enrolling in this course, students are considered to have agreed to the above ethics statement.
範例 2　有條件開放：指定作業不得使用Example 2 · Conditional use, with designated tasks excluded
生成式AI倫理聲明 基於透明與負責任原則，本課程鼓勵學生利用AI進行協作或互學，以提升課程產出品質。學生可依規定揭露AI於議題發想、文句潤飾或結構參考等用途；但在「個人反思報告」、「小組採訪作業」等指定任務中，不得使用生成式AI撰寫。未依規定揭露者，得重新評分或不予計分；教師使用AI產製教材或學習資料時亦應適當標注。修讀本課程之學生於選課時視為同意以上倫理聲明。
Generative AI Ethics Statement Based on the principles of transparency and responsibility, this course encourages students to use AI for collaboration or co-learning to improve the quality of course outputs. Students may use AI for topic ideation, sentence polishing, or structural reference, provided such use is disclosed as required. However, generative AI may not be used to write designated tasks such as an “individual reflection report” or “group interview assignment.” Failure to disclose AI use as required may result in re-evaluation or no credit. When instructors use AI to produce teaching materials or learning resources, they should also appropriately indicate such use. By enrolling in this course, students are considered to have agreed to the above ethics statement.
範例 3　禁止使用Example 3 · Prohibited use
生成式AI倫理聲明 基於透明與負責任原則，經考量課程內容與學習目標，本課程不宜使用生成式AI，學生不得繳交使用生成式AI產出的作業、報告或個人心得。若經查核發現違規，教師、學校或相關單位有權重新評分或不予計分。修讀本課程之學生於選課時視為同意以上倫理聲明。
Generative AI Ethics Statement Based on the principles of transparency and responsibility, and after considering the course content and learning objectives, this course does not permit the use of generative AI. Students may not submit assignments, reports, or personal reflections produced using generative AI. If a violation is identified, the instructor, university, or relevant unit may re-evaluate the work or assign no credit. By enrolling in this course, students are considered to have agreed to the above ethics statement.
4

教師角色：從提供答案到示範判斷THE INSTRUCTOR’S ROLE: FROM GIVING ANSWERS TO MODELLING JUDGEMENT

學習品質越來越取決於人怎麼提問、怎麼用 AI，以及怎麼判斷 AI 的產出。教師無可替代的價值，不再是提供學生靠 AI 就能快速取得的答案，而是讓學生看見專業判斷「如何形成」。
Learning quality increasingly depends on how people ask questions, use AI and judge its output. The instructor’s irreplaceable value is no longer supplying answers students can get from AI, but showing how expert judgement is formed.

「從遊」：把判斷的形成歷程攤開“Following and swimming together”: make judgement visible

「學校猶水也，師生猶魚也，其行動猶游泳也，大魚前導，小魚尾隨，是從游也。從游既久，其濡染觀摩之效，自不求而至，不為而成。」
“The university is like water, teachers and students like fish, and their activity like swimming. The big fish lead and the small fish follow. After swimming together long enough, students absorb what they observe without seeking it and achieve it without forcing it.”
— 梅貽琦〈大學一解〉，1941Mei Yi-chi, “An Interpretation of the University”, 1941
設計判斷情境Design situations for judgement→示範判斷歷程Model the reasoning→引導學生形成判斷Guide students to their own judgement
可以採用的教學方式Suggested approaches
案例討論Case discussion問題導向學習Problem-based learning專題學習Project-based learning小組討論Group discussion實作活動Hands-on activities口頭說明Oral explanation批判比較 AI 輸出Critiquing AI outputs
瀏覽數:
友善列印
登入成功