2024 年 10 月，教育局以初中科學科為試點推出「『智』為學理」撥款計劃，申請獲批的學校獲發一筆過 10 萬元，用來加強科學科教師運用人工智能輔助教學的能力[1]。撥款由優質教育基金支持，學校可以用到 2026/27 學年完結，未用的餘額在 2027 年 8 月 31 日之後要退回基金[1][3]。In October 2024 the Education Bureau launched the AI for Science funding scheme, a pilot for junior secondary science, and schools whose applications succeeded received a one-off HK$100,000 to strengthen science teachers' use of AI-assisted teaching[1]. The money comes through the Quality Education Fund and can be used until the end of the 2026/27 school year, and any balance left after 31 August 2027 is returned to the Fund[1][3].
真正卡住學校的是三項承諾，錢反而容易處理。通函要求學校發展最少兩個在初中科學應用人工智能的教學例子或資源、舉辦最少一次公開課或示範課、舉辦最少一次經驗分享會[1]。這三項都是教學工作，而撥款只可以支付兩類支出：科學科教師報讀專上院校或相關專業機構的短期課程，以及與專上院校或相關專業機構協作的支援計劃，後者上限 5 萬元[2]。教具、人工智能軟件與裝置都不在名單之內[2]。一間學校如果打算用這 10 萬元買工具，會發現錢用不出去，承諾卻仍然要交。The three commitments are what actually hold schools up, and the money is the easier part. The circular asks a school to develop at least two teaching examples or resources for using AI in junior secondary science, to hold at least one open or demonstration lesson, and to hold at least one experience-sharing session[1]. All three are teaching work, while the money may pay for two kinds of spending only: science teachers attending short courses run by tertiary institutions or relevant professional bodies, and a support programme run with a tertiary institution or professional body, capped at HK$50,000[2]. Teaching aids, AI software and devices are not on the list[2]. A school that plans to spend the HK$100,000 on tools finds that the money cannot go out and the commitments still have to be met.
2026 年 9 月 30 日的中期報告已經過去，下一個硬期限是 2027 年 9 月 30 日的總結報告，它要交代截至 2027 年 8 月 31 日的開支，以及三項承諾各自完成的數量[1][3]。財務及會計安排須知寫明，學校未能按時提交總結報告，教育局與基金有權收回整筆撥款[3]。交齊的關鍵落在五個位置：三項承諾、10 萬元的兩個用途、教學例子的四種人工智能用法、2027 年 9 月前的四個時段，與 50 萬元「智啟學教」的分工。The interim report fell due on 30 September 2026 and has passed; the next hard date is the final report on 30 September 2027, which accounts for spending up to 31 August 2027 and states how many of each commitment the school completed[1][3]. The notes on financial and accounting arrangements say plainly that where a school fails to submit the final report on time, the Bureau and the Fund may recover the whole grant[3]. Five places decide whether the file is complete: the three commitments, the two uses of the HK$100,000, the four modes of AI use in a teaching example, the four periods before September 2027, and how this grant divides from the HK$500,000 AI for Empowering Learning and Teaching scheme.
摘要Summary
- 「『智』為學理」撥款計劃在 2024 年 10 月以初中科學科為試點推出，申請獲批的學校獲發一筆過 10 萬元，由優質教育基金支持[1]。The AI for Science scheme began in October 2024 as a junior secondary science pilot and gives a successful applicant a one-off HK$100,000, backed by the Quality Education Fund[1].
- 撥款可以用到 2026/27 學年完結，未用的餘額在 2027 年 8 月 31 日之後要退回基金，學校也不可以把撥款或餘款調往其他帳項[1][3]。The money may be used until the end of the 2026/27 school year, any balance left after 31 August 2027 is returned to the Fund, and neither the grant nor its balance may be moved to another account[1][3].
- 三項承諾是發展最少兩個在初中科學應用人工智能的教學例子或資源、舉辦最少一次公開課或示範課、舉辦最少一次經驗分享會，後兩項可以在校內或校外舉行[1]。The three commitments are at least two teaching examples or resources for using AI in junior secondary science, at least one open or demonstration lesson, and at least one experience-sharing session, with the last two held on or off campus[1].
- 撥款只可以支付兩類支出：科學科教師報讀專上院校或相關專業機構的短期課程，以及與專上院校或相關專業機構協作的支援計劃，支援計劃上限 5 萬元[2]。The money covers two kinds of spending only: science teachers attending short courses run by tertiary institutions or relevant professional bodies, and a support programme run with a tertiary institution or professional body, capped at HK$50,000[2].
- 教具、人工智能軟件、應用程式、裝置、實驗室工程、學生或家長的課程、分享會的宣傳與飲食支出，全部列為不恰當運用[2]。Teaching aids, AI software, applications, devices, laboratory works, courses for students or parents, and spending on promoting a sharing session or on food are all listed as improper use[2].
- 總結報告在 2027 年 9 月 30 日或之前交回教育局課程支援分部科學教育組，交代截至 2027 年 8 月 31 日的開支與三項承諾的完成數量；未能按時提交，教育局與基金有權收回整筆撥款[1][3]。The final report goes back to the Science Education Section of the Bureau's Curriculum Support Division by 30 September 2027, accounting for spending to 31 August 2027 and the number of each commitment completed; where it is late, the Bureau and the Fund may recover the whole grant[1][3].
三項承諾是甚麼？What are the three commitments?
通函第 5 段列明三項承諾：發展最少兩個在科學（中一至中三）應用人工智能的教學例子或資源、舉辦最少一次運用人工智能輔助教學的公開課或示範課、舉辦最少一次經驗分享會[1]。公開課與分享會可以在校內或校外舉行，通函沒有規定聽眾的人數或身份[1]。三項都是教學工作，撥款的使用範疇裡沒有為它們設一個開支類別[2]。Paragraph 5 of the circular sets out three commitments: develop at least two teaching examples or resources for using AI in science at S.1 to S.3, hold at least one open or demonstration lesson that uses AI-assisted teaching, and hold at least one experience-sharing session[1]. The lesson and the session may be held on or off campus, and the circular sets no number or kind of audience[1]. All three are teaching work, and the funding rules create no spending category for them[2].
| 承諾Commitment | 最低要求Minimum | 要交的文件Documents | 完成期限Deadline |
|---|---|---|---|
| 學校發展在初中科學應用人工智能的教學例子或資源[1]。The school develops teaching examples or resources for using AI in junior secondary science[1]. | 學校發展兩個，對象是科學（中一至中三）的課堂。The school develops two, aimed at S.1 to S.3 science lessons. | 學校交教學方案與教學資源，教學影片屬非必須[4]。The school submits a teaching plan and the resources, and a lesson video is optional[4]. | 學校在 2027 年 8 月 31 日前完成[1][3]。The school completes it by 31 August 2027[1][3]. |
| 學校舉辦運用人工智能輔助教學的公開課或示範課[1]。The school holds an open or demonstration lesson in AI-assisted teaching[1]. | 學校舉辦一次，校內或校外都可以[1]。The school holds one, on or off campus[1]. | 學校交課堂簡報、學生問卷及分析[4][6]。The school submits the lesson slides and the student questionnaire with its analysis[4][6]. | 同上，2027 年 8 月 31 日前完成[3]。As above, by 31 August 2027[3]. |
| 學校舉辦經驗分享會[1]。The school holds an experience-sharing session[1]. | 學校舉辦一次，校內或校外都可以[1]。The school holds one, on or off campus[1]. | 學校交經驗分享會的簡報[4]。The school submits the slides from the session[4]. | 同上，2027 年 8 月 31 日前完成[3]。As above, by 31 August 2027[3]. |
兩個教學例子可以來自同一級、同一科，通函沒有要求分開[1]。放在不同的科學主題或不同的學習階段，總結報告的說明會容易寫，校內的分享也多了內容。教育局科學教育組另設科學（中一至中三）教案設計獎勵計劃，完成其中一個階段，視同已發展一個教學例子或資源[4]。已經參加獎勵計劃的學校，先點算手上已有的教案，再決定還要補幾個。The two examples may come from the same level and the same subject, because the circular does not require them to be separate[1]. Spreading them across different science topics or learning stages makes the final report easier to write and gives an internal sharing session more to say. The Bureau's Science Education Section also runs a lesson-design award scheme for S.1 to S.3 science, and completing one of its stages counts as one teaching example or resource[4]. A school that has already taken part should count the lesson plans it holds before deciding how many are still missing.
公開課與分享會的要求是次數，規模由學校決定。通函只寫校內或校外，沒有寫聽眾是誰[1]。以科務會議形式舉行的分享會符合這個寫法，前提是內容關於初中科學應用人工智能，並留下簡報與出席紀錄。公開課也一樣：一節由科組同事觀課的課堂符合條文，重要的是課堂前後有紀錄，包括教案、簡報與學生問卷[4]。The lesson and the session are counted, not sized. The circular says on or off campus and says nothing about who attends[1]. A sharing session run as a panel meeting fits that wording, provided the content covers AI in junior secondary science and the school keeps the slides and an attendance record. The same holds for the open lesson: one lesson observed by colleagues meets the clause, and what matters is the record around it, which is the plan, the slides and the student questionnaire[4].
十萬元可以用在哪裏？What may the HK$100,000 be spent on?
撥款只可以支付兩類支出。第一類是安排科學科教師報讀由專上院校或相關專業機構舉辦、關於人工智能輔助教學的短期課程，包括講座與工作坊。第二類是與專上院校或相關專業機構協作，開展人工智能輔助教學的支援計劃，這一項不應多於 5 萬元[2]。其餘支出，即使與人工智能有關，都列為不恰當運用[2]。The grant pays for two kinds of spending. The first is sending science teachers to short courses on AI-assisted teaching run by tertiary institutions or relevant professional bodies, including talks and workshops. The second is a support programme on AI-assisted teaching run with a tertiary institution or a relevant professional body, and that item should not exceed HK$50,000[2]. Everything else, even where it concerns AI, is listed as improper use[2].
| 支出類別Category | 條文允許的寫法What the rules allow | 最容易出錯的位置Where it goes wrong |
|---|---|---|
| 科學科教師報讀短期課程Short courses for science teachers | 學校安排教師報讀專上院校或相關專業機構舉辦的人工智能輔助教學課程、講座或工作坊[2]。The school sends teachers to AI-assisted teaching courses, talks or workshops run by tertiary institutions or relevant professional bodies[2]. | 學校以同一筆錢買課程用的軟件或教具，條文把採購學與教資源列為不恰當[2]。The school uses the same money to buy the software or aids used in the course, and the rules list buying learning and teaching resources as improper[2]. |
| 與專上院校或專業機構協作的支援計劃A support programme with a tertiary institution or professional body | 學校與專上院校或相關專業機構協作開展支援計劃，金額不多於 5 萬元[2]。The school runs a support programme with a tertiary institution or relevant professional body, for no more than HK$50,000[2]. | 學校把支援計劃寫成採購教學資源，或把金額開到 5 萬元以上[2]。The school writes the support programme up as a purchase of teaching resources, or lets the figure pass HK$50,000[2]. |
| 其他與人工智能有關的支出Any other AI-related spending | 條文沒有第三類，學校只可以在此列以外另找資源[2]。There is no third category, so the school has to find other resources for anything else[2]. | 裝置、軟件、應用程式、實驗室工程、學生或家長的課程、分享會宣傳、飲食，全部不恰當[2]。Devices, software, applications, laboratory works, courses for students or parents, promoting a sharing session and food are all improper[2]. |
5 萬元的上限本身就是一個提醒。10 萬元裡最多一半可以放在支援計劃，其餘要落在教師的課程。換句話說，這筆撥款有一半以上要變成教師的學習，學校的採購只佔小部分[2]。一間學校如果沒有教師培訓的計劃，10 萬元會用不完，而餘款要退回基金[3]。The HK$50,000 cap is itself a signal. At most half of the HK$100,000 may sit in the support programme, and the rest has to go on teachers' courses. Put another way, more than half of this grant is meant to become teachers' learning, and the school's purchasing is the smaller part[2]. A school with no teacher-training plan will not spend the HK$100,000, and the balance goes back to the Fund[3].
採購要跟程序。資助學校與按位津貼學校須依教育局通告第 4/2013 號「資助學校採購程序」及附載的指引，並參閱《學校行政手冊》第 6.4 章；直資學校依資助學校程序，或依校董會批准的校本採購政策；官立學校依教育局內部通告[2][3]。10 萬元的總額與 5 萬元的上限，會令報價與招標的門檻成為實際問題。先問校內的採購負責人，再決定分幾張單、每張單用哪一種程序。Purchasing follows procedure. Aided and capacity-based subsidy schools follow Education Bureau Circular No. 4/2013 on procurement procedures in aided schools and its attached guidance, together with chapter 6.4 of the School Administration Guide; direct subsidy scheme schools follow the aided school procedures or a school-based policy approved by their board; government schools follow the Bureau's internal circulars[2][3]. The HK$100,000 total and the HK$50,000 cap turn quotation and tender thresholds into a practical question. Ask the school's procurement officer first, then decide how many orders to raise and which procedure each one takes.
帳目要獨立。資助中學、按位津貼中學與直資中學須為這筆撥款另設一個獨立分類帳，記錄所有收支[3]。學校不用把發票與收據副本交教育局，但要存檔並保留不少於七年，以備查核[1][3]。餘款要退：截至 2027 年 8 月 31 日仍未使用的餘額，資助、按位津貼與直資學校要以劃線支票退回基金，支票抬頭寫「香港特別行政區政府」；官立學校的餘額按指定用戶帳號的記錄退還[3]。撥款與餘款都不可以調往其他帳項[1]。The ledger is separate. Aided, capacity-based subsidy and direct subsidy scheme secondary schools keep a separate ledger for this grant recording every receipt and payment[3]. Schools do not send invoices and receipts to the Bureau, but they keep them on file for at least seven years in case of a check[1][3]. The balance is returned: any amount unspent at 31 August 2027 goes back to the Fund by crossed cheque made payable to the Government of the Hong Kong Special Administrative Region for aided, capacity-based subsidy and direct subsidy scheme schools, while government schools return it against the designated user account record[3]. Neither the grant nor its balance may be moved to another account[1].
教學例子要達到甚麼標準？What does a teaching example have to show?
一個教學例子要交代三件事：人工智能在這一課做甚麼、學生在這一課做甚麼、學生的學習有甚麼證據[4][6]。教育局科學教育組把生成式人工智能在科學課堂的角色分成四種，由完全不使用，到由人工智能完成整個探究再交由學生評鑑[5]。這四種模式對應 Furze 等人在 2024 年提出的 AI Assessment Scale[8]。先選模式，再寫教案，比先寫好教案再想人工智能放在哪裏容易，也令兩個例子不會撞在同一種用法上。A teaching example has to answer three questions: what the AI does in the lesson, what the students do in the lesson, and what evidence of learning the lesson leaves[4][6]. The Bureau's Science Education Section sorts the role of generative AI in a science lesson into four modes, from no AI at all to AI completing the whole investigation for students to judge[5]. The four modes follow the AI Assessment Scale proposed by Furze and colleagues in 2024[8]. Choosing the mode first and writing the plan second is easier than writing a plan and then finding somewhere to put the AI, and it stops the two examples landing on the same use.
| 模式Mode | 人工智能的角色The AI's role | 學生要做的事What students do | 學校留下的證據Evidence kept |
|---|---|---|---|
| 第一種：不涉及生成式人工智能[5]Mode 1: no generative AI[5] | 學生在探究過程中不使用人工智能[5]。Students use no AI in the investigation[5]. | 學生自行選擇材料、調節斜面坡度、觀察車速變化並分析數據[5]。Students choose materials, adjust the ramp, observe how the car's speed changes and analyse the data[5]. | 學校留下實驗設計、數據與實驗報告[5]。The school keeps the design, the data and the report[5]. |
| 第二種：輔助構思[5]Mode 2: AI for ideation[5] | 人工智能只就探究提供意念與資料，不生成內容[5]。The AI suggests ideas and sources and generates no content[5]. | 學生選取並改良方案，然後在不借助人工智能的情況下完成探究[5]。Students select and improve a method, then run the investigation without AI[5]. | 學校留下方案比較與最後採用的實驗設計[5]。The school keeps the comparison of methods and the design finally used[5]. |
| 第三種：輔助修訂[5]Mode 3: AI for revision[5] | 人工智能只協助改善行文表達，不生成探究內容[5]。The AI improves expression only and generates none of the investigation[5]. | 學生評鑑人工智能改寫後的報告，發現錯誤並依實驗結果修正[5]。Students judge the revised report, find its errors and correct them against their results[5]. | 學校留下學生原稿與人工智能修訂稿，兩份一併呈交[5]。The school keeps the student's draft and the AI revision, submitted together[5]. |
| 第四種：由人工智能完成任務，交由學生評鑑[5]Mode 4: AI completes the task for students to judge[5] | 人工智能生成內容並完成整個探究[5]。The AI generates content and completes the whole investigation[5]. | 學生指出人工智能結論的不合理處，要求修正並說明理由[5]。Students show why the AI's conclusion is unsound, ask for a correction and explain the reasoning[5]. | 學校留下對話紀錄與修正後的預測結果[5]。The school keeps the conversation record and the corrected prediction[5]. |
兩個例子放在不同模式，一套組合會比兩個同類的例子有說服力。第二種與第四種是一組實用的配搭：一個展示學生怎樣用人工智能擴闊思路，再自己動手驗證；另一個展示學生怎樣判斷人工智能的錯誤，並用實驗數據說服它改正[5]。四種模式沒有優劣，分別在學生要付出哪一種思考。第一種模式沒有使用人工智能，單獨不構成一個人工智能教學例子，但它是很好的對照組：同一課在兩個班分別用第二種與第一種模式，學生問卷的分數就有了比較的基礎[6]。Two examples placed in different modes make a stronger pair than two of the same kind. Modes 2 and 4 work well together: one shows students using AI to widen their thinking and then testing it by hand, the other shows them judging an AI error and correcting it with their own data[5]. No mode is better than another; they differ in the kind of thinking they ask of students. Mode 1 uses no AI, so on its own it is not an AI teaching example, but it makes a good control: run the same lesson in two classes, one at mode 2 and one at mode 1, and the questionnaire scores have a comparison point[6].
三種寫法會在總結報告裡露出來。第一種是課堂只有教師示範工具，學生沒有動手，學生問卷答不出學習的改變。第二種是選了第四種模式，卻沒有評鑑環節，學生直接交出人工智能的答案，學校留下的只有一份列印本。第三種是問卷只在課後做一次，沒有對照班或前測，分析只能寫滿意度，寫不出學生學了甚麼[6]。Three kinds of write-up show up in the final report. In the first, the teacher demonstrates a tool and the students do nothing, so the questionnaire cannot show a change in learning. In the second, the school picks mode 4 but skips the judging step, and the students hand in the AI's answer with only a printout left behind. In the third, the questionnaire is given once at the end with no comparison class and no pre-test, so the analysis reports satisfaction and says nothing about what students learned[6].
學生問卷有 18 條題目，圍繞趣味、投入、好奇心、自行探究、概念掌握、表達、證據推理與選科意願[6]。第 6、9、10、12 條是反向題，例如「我希望教師在這堂科學課上沒有使用人工智能」與「即使在人工智能協助下，我仍覺得科學很難學」[6]。填答後要先倒轉這四條的分數，再計算平均，否則分析會把「希望老師沒有用人工智能」讀成正面結果。用教育局的分析表可以省一次換算，但要看清楚它有沒有處理反向題。The student questionnaire has 18 items covering interest, engagement, curiosity, independent enquiry, grasp of concepts, expression, reasoning from evidence and intention to choose science subjects[6]. Items 6, 9, 10 and 12 are worded in reverse, such as "I wish the teacher did not use AI in this science lesson" and "even with the help of AI, I still feel science is hard to learn"[6]. Those four scores have to be reversed before any average is taken, or the analysis reads "I wish the teacher had not used AI" as a positive result. The Bureau's analysis workbook saves the arithmetic, once it is clear whether it reverses those items.
由現在到 2027 年 9 月 30 日，學校要分四個時段完成From now to 30 September 2027, the school works through four periods
撥款可以用到 2027 年 8 月 31 日，總結報告在 2027 年 9 月 30 日或之前交回教育局課程支援分部科學教育組[1][3]。由 2026 年 11 月計起，還有九個月與一個暑假，足夠完成三項承諾，但每個時段只夠做一件主要的事。把三項承諾平均撒在整個學年，結果通常是三件都做了一半。The money may be used until 31 August 2027, and the final report goes to the Science Education Section of the Bureau's Curriculum Support Division by 30 September 2027[1][3]. From November 2026 there are nine months of term and one summer left, which is enough for the three commitments if each period carries one main task. Spreading all three across the year usually leaves all three half done.
| 時段Period | 要完成的事What gets done | 要留下的證據Evidence kept |
|---|---|---|
| 2026 年 11 月至 2027 年 1 月November 2026 to January 2027 | 學校選定兩個教學例子的課題與模式，並安排教師報讀課程[2]。The school fixes the topics and modes of the two examples and sends teachers to a course[2]. | 學校留下課題名稱、模式選擇、報讀紀錄與收據[3]。The school keeps the topic names, the mode chosen, the enrolment record and the receipt[3]. |
| 2027 年 2 月至 4 月February to April 2027 | 學校完成第一個教學例子，並在同一個課題上舉行公開課或示範課[1]。The school finishes the first example and holds the open or demonstration lesson on the same topic[1]. | 學校留下教學方案、教學資源、課堂簡報與學生問卷[4]。The school keeps the teaching plan, the resources, the slides and the questionnaire[4]. |
| 2027 年 5 月至 7 月May to July 2027 | 學校完成第二個教學例子，並舉行經驗分享會[1]。The school finishes the second example and holds the sharing session[1]. | 學校留下教學方案、分享會簡報與出席紀錄[4]。The school keeps the second plan, the slides and an attendance record[4]. |
| 2027 年 7 月至 9 月July to September 2027 | 學校結帳，填寫總結報告並在 9 月 30 日或之前交回科學教育組[1]。The school closes the accounts, completes the final report and files it with the Section by 30 September[1]. | 學校留下獨立分類帳、總結報告與校董會批核紀錄[3]。The school keeps the separate ledger, the final report and the board's approval[3]. |
支援計劃要早兩個月開始談。以我們與學校編排課堂的經驗，由確定範圍到第一節課，八至十星期是常見的前置時間：學校要定範圍、走採購程序、編時間表，機構要設計課堂與準備材料。想在 2027 年 2 月開始，2026 年 12 月就要開始談[2]。同一道理，教師課程的名額在學期中段最緊張，1 月之前報名會比 4 月容易。A support programme needs to be discussed two months early. From our own work scheduling lessons with schools, eight to ten weeks is the usual lead time from fixing the scope to the first lesson: the school sets the scope, runs its procurement, builds the timetable, and the provider designs the lesson and prepares the materials. A start in February 2027 means the conversation begins in December 2026[2]. Teacher-course places are tightest in the middle of a term for the same reason, so booking before January is easier than booking in April.
公開課要訂在學校的行事曆上，並且預留一節後備。示範課當日如果有同事觀課，學生的表現會受影響，問卷的結果要與平日的一節比較才有意義[6]。分享會可以排在第三學期的科務會議，把第二個教學例子講一次，同時完成第二項與第三項承諾。The open lesson belongs on the school calendar with a backup slot behind it. When colleagues observe, students behave differently, so the questionnaire result only reads correctly next to a normal lesson[6]. The sharing session can sit in a panel meeting in the third term, walking through the second example and closing two commitments at once.
中期報告已經過期。2026 年 9 月 30 日的限期已過，通函沒有寫補交安排[1]。未交的學校應該直接聯絡教育局課程支援分部科學教育組，電話 3698 3440，問清楚補交的方法，以及它對總結報告的影響[1]。這件事排在所有教學工作之前，因為它影響的是整筆撥款，三項承諾只佔其中一部分。The interim report is overdue. The deadline of 30 September 2026 has passed and the circular sets out no late-submission arrangement[1]. A school that has not filed should call the Science Education Section of the Curriculum Support Division on 3698 3440 and ask how to file late and what it means for the final report[1]. This comes before any teaching work, because what is at stake is the whole grant, and the three commitments are only part of it.
總結報告要填甚麼？What does the final report ask for?
總結報告有五個位置：兩類支出的金額與數量、截至 2027 年 8 月 31 日的餘款、三項承諾各自完成的數量、校長簽署的聲明，以及隨報告提交的教學文件[1]。報告在 2027 年 9 月 30 日或之前交回教育局課程支援分部科學教育組[1]。表格在通函附件四，可以和中期報告的附件三對照著填，兩個版本用的是同一組分項。The final report has five places to fill: the amounts and counts for the two spending categories, the balance at 31 August 2027, the number of each commitment completed, the principal's signed declaration, and the teaching documents filed with it[1]. It goes back to the Science Education Section of the Curriculum Support Division by 30 September 2027[1]. The form is Annex 4 to the circular and can be filled alongside Annex 3, the interim report, because both use the same sub-items.
| 報告部分Part of the report | 要填的內容What goes in | 常見的遺漏What gets missed |
|---|---|---|
| 分項(i) 教師短期課程Item (i), teacher courses | 學校填報讀課程的教師人數、課程數目與總支出，並說明課程是否由附件二名單內的提供者舉辦[1]。The school states the number of teachers, the number of courses and the total spent, and says whether the courses were run by providers listed in Annex 2[1]. | 學校只填金額，沒有填人數與課程數目[1]。The school enters the amount and leaves out the teacher and course counts[1]. |
| 分項(ii) 支援計劃Item (ii), support programme | 學校填支援計劃的數目與總支出，金額不多於 5 萬元[2]。The school enters the number of support programmes and the total spent, within the HK$50,000 cap[2]. | 學校把支援計劃與其他開支混在同一張單據裡[3]。The programme is mixed with other spending on the same invoice[3]. |
| 餘款The balance | 學校填已全數用完，或寫明餘款金額；資助、按位津貼與直資學校的餘款須退回基金[3]。The school states that the grant is fully spent or gives the balance, and for aided, capacity-based subsidy and direct subsidy scheme schools the balance is returned to the Fund[3]. | 學校沒有在 2027 年 8 月 31 日前用完，餘款變成退款[3]。The money is not spent by 31 August 2027, and the balance becomes a repayment[3]. |
| 三項承諾的完成數量Counts for the three commitments | 學校填已發展的教學例子數目、公開課次數與分享會次數[1]。The school enters the number of examples developed, lessons held and sessions held[1]. | 學校填的次數與附上的文件對不上[4]。The counts do not match the documents attached[4]. |
| 聲明與簽署Declaration and signature | 校長簽署、學校印鑑，並聲明開支符合運用原則、使用範疇與採購程序[1]。The principal signs and the school stamps, declaring that spending followed the principles, the permitted scope and the procurement rules[1]. | 報告未經校董會、法團校董會或學校管理委員會批核[1]。The report goes out without approval from the school's board or management committee[1]. |
隨報告交的文件有六份，其中教學影片屬非必須：教學方案、教學資源、教學影片、課堂簡報、學生問卷及分析、經驗分享會簡報[4]。教學方案範本是一個試算表，學生問卷是一個 PDF，分析表是另一個試算表，三份檔案都可以在計劃網頁下載[4]。用它們由第一日開始做紀錄，總結報告的說明就不用重寫。Six documents go with the report, one of them optional: the teaching plan, the resources, the lesson video, the lesson slides, the student questionnaire with its analysis, and the sharing-session slides[4]. The plan template is a spreadsheet, the questionnaire is a PDF and the analysis is another spreadsheet, and all three can be downloaded from the scheme's page[4]. Keeping records in them from the first day means the final report's narrative does not have to be written twice.
報告要進入學校報告。通函第 10 段要求學校把中期報告與總結報告的內容，包括資助項目、開支與評估結果，載於該學年的學校報告，經校董會、法團校董會或學校管理委員會批核，並把批核後的周年計劃與學校報告上載校網[1]。教育局科學教育組會定期探訪學校，並與教師舉行焦點小組會議，檢視計劃的實施情況[1]。探訪與焦點小組看的，是同一組文件：兩個教學例子，與學生的學習證據。The report also enters the school report. Paragraph 10 of the circular requires the content of the interim and final reports, including the funded activities, the spending and the evaluation results, to be carried in that year's school report, approved by the school's board or management committee, and the approved annual plan and school report to be posted on the school website[1]. The Section visits schools from time to time and holds focus groups with teachers to review how the scheme is running[1]. The visits and the focus groups look at the same set of documents: the two teaching examples and the evidence of student learning.
與「智啟學教」可以同時用嗎？Can a school use both grants?
可以，兩筆撥款的用途幾乎互補。10 萬元的「智為學理」支付科學科教師的短期課程與支援計劃，50 萬元的「智啟學教」支付裝置、訂閱、學與教資源與校本人工智能方案，並把資助教師或家長修讀人工智能課程列為不恰當運用[2][7]。同一項開支只可以在一個計劃報銷，兩筆錢要分開記帳[3][7]。Yes, and the two grants are close to complementary. The HK$100,000 AI for Science grant pays for science teachers' short courses and for a support programme, while the HK$500,000 AI for Empowering Learning and Teaching scheme pays for devices, subscriptions, learning and teaching resources and school-based AI solutions, and lists funding teachers or parents to attend AI courses as an improper use[2][7]. One item of spending may be claimed once, so the two pots are kept in separate accounts[3][7].
| 項目Item | 「智為學理」10 萬元AI for Science, HK$100,000 | 「智啟學教」50 萬元AI for Learning and Teaching, HK$500,000 |
|---|---|---|
| 科學科教師報讀人工智能課程Science teachers attending AI courses | 可以，這是撥款的主要用途[2]。Allowed, and it is the grant's main use[2]. | 不恰當，附件二明確排除[7]。Improper, and Annex 2 excludes it explicitly[7]. |
| 與專上院校或專業機構協作的支援計劃A support programme with a tertiary institution or professional body | 可以，上限 5 萬元[2]。Allowed, capped at HK$50,000[2]. | 可以，屬於校本人工智能方案[7]。Allowed, as a school-based AI solution[7]. |
| 購置裝置、訂閱工具與學與教資源Devices, subscriptions and learning resources | 不恰當，附件二把採購列為不恰當[2]。Improper, as Annex 2 lists purchasing[2]. | 可以，這是撥款的主要用途[7]。Allowed, and it is the scheme's main use[7]. |
| 用款與報告的期限Spending and reporting deadlines | 2027 年 8 月 31 日用盡，2027 年 9 月 30 日交總結報告[1][3]。Spend by 31 August 2027 and file the final report by 30 September 2027[1][3]. | 2028 年 8 月 31 日用盡，2028 年 9 月 30 日交總結報告[7]。Spend by 31 August 2028 and file the final report by 30 September 2028[7]. |
兩份通函都要求獨立帳目與校董會批核，也都禁止把撥款調往其他帳項[1][3][7]。同時持有兩筆撥款的學校，最容易出錯的地方是把教師培訓的單據放入 50 萬元的帳，因為那張單看起來屬於人工智能培訓。在「智啟學教」的運用指引裡，這正是不恰當運用[7]；同一張單放在「智為學理」的帳，就完全正確[2]。Both circulars require a separate ledger and board approval, and both forbid moving the money to another account[1][3][7]. For a school holding both, the likeliest error is putting a teacher-training invoice into the HK$500,000 account, because the invoice looks like AI training. Under the AI for Learning and Teaching rules that is improper use[7], while the same invoice in the AI for Science account is exactly right[2].
兩筆撥款的時間表相差一年。10 萬元在 2027 年 8 月 31 日截數，50 萬元在 2028 年 8 月 31 日截數[3][7]。同一年入帳的開支，要分別貼上兩個標籤；2027/28 學年仍然有 50 萬元可用，但科學科那一筆已經結帳，之後的教師培訓要另找資源。The two timetables are a year apart. The HK$100,000 closes on 31 August 2027 and the HK$500,000 on 31 August 2028[3][7]. Spending booked in the same year carries two different labels, and while the 2027/28 school year still has the HK$500,000, the science grant is closed by then and later teacher training needs another source.
有些部分我們幫得上，有些部分留給科組Some of this we can carry, and some of it stays with the panel
我們幫得上的部分，集中在兩個教學例子與它們的證據：選定模式與課題、寫教學方案、準備課堂資源、到校主持公開課或示範課、整理學生問卷與分析，以及把六份文件整理成一個可以交的檔案[4]。這些工作與我們平日設計課堂、帶學生做專題的做法一樣，分別只在題材由科學科決定。同一套設計方法也用於我們的到校 AI 課程，分別在支援計劃只覆蓋一個課題，課程覆蓋一個學期或以上。What we can carry sits around the two teaching examples and their evidence: choosing the mode and the topic, writing the teaching plan, preparing the lesson resources, running the open or demonstration lesson on site, handling the student questionnaire and its analysis, and assembling the six documents into a file that can be submitted[4]. The work is the same as our ordinary lesson design and project work with students; the only difference is that the science panel sets the subject matter. The same design method runs through our on-campus AI programme, except that a support programme covers one topic while the programme runs a term or longer.
留給科組的部分同樣清楚。科學科的課程內容與教學判斷屬於科組，包括哪一級、哪一個課題、哪一個探究適合用人工智能。支援計劃是否符合「專上院校或相關專業機構」的寫法，由學校與校董會判斷，教育局說明沒有指定課程提供者[4]。採購程序由學校的採購負責人決定，我們只提供報價與服務範圍[2]。What stays with the panel is just as clear. The science content and the teaching judgement belong to the panel, including which level, which topic and which investigation suits AI. Whether a support programme meets the wording "a tertiary institution or relevant professional body" is a judgement for the school and its board, and the Bureau states that it designates no course provider[4]. The procurement route belongs to the school's procurement officer; we supply a quotation and a scope of service[2].
限制也寫清楚。我們的紀錄只覆蓋我們上過的課，學校其他科組的教學例子要由科組自己補。一個支援計劃的上限是 5 萬元，這個上限決定了合作的長度，通常是一個課題、兩個級別、一個學期之內，做不到一年的全程陪伴[2]。把兩個教學例子留在科組手上，第二年才有可以重複的課堂；外判一份文件不會令三項承諾變成學校的能力。The limits are as plain. Our records cover the lessons we taught, and teaching examples from other panels have to come from those panels. The HK$50,000 cap on a support programme sets the length of the work, usually one topic, two levels and one term, and not a year of continuous support[2]. Keeping the two examples with the panel is what gives the school a lesson it can run again next year, and outsourcing a document does not turn three commitments into a capability.
