「科學(中一至中三)教案設計獎勵計劃」第二階段現已接受報名,截止日期是 2027 年 3 月 1 日,結果在 2027 年 6 月通知。每所學校只可以交一份教案,課堂內容不多於兩個課節,大約 80 分鐘,並且要配合《科學(中一至中三)課程框架》(2025)的學習單元。交件包括一份電子表格、最多六頁中文字與教學材料,以及一張由校長簽署的申請表格。The second phase of the Science (S1-S3) Teaching Plan Design Award Scheme is open, with entries due on 1 March 2027 and results announced in June 2027. Each school may submit one teaching plan, the lesson may run to no more than two periods, about 80 minutes, and it has to fit a learning unit in the Science (S1-S3) Curriculum Framework (2025). An entry consists of an electronic form, up to six pages of Chinese teaching material, and an application form signed by the principal.

電子表格分四部分:課堂基本資料、數字教育工具、八行課堂設計、評鑑方法。表格本身已經規定了大半的答案:單元、學習內容與學習成果都是下拉選單,選項只有課程框架的 13 個單元、59 個子單元與 331 條學習成果;八行課堂設計的每一行要選一個科學實踐過程、一種學習模式、一個人工智能角色,再填活動與分鐘;兩個欄位各有 150 字上限。The electronic form has four parts: class details, digital education tools, an eight-row lesson design and an evaluation method. The form has already fixed most of the answers. The unit, the learning content and the learning outcomes are dropdowns that carry only the framework's 13 units, 59 sub-units and 331 outcomes. Each of the eight lesson-design rows takes a science practice, a learning mode, an AI role, an activity and a number of minutes. Two fields are capped at 150 characters.

評審團按科學思維、創意思維與運算思維三方面評分。三種思維之中沒有一欄叫「人工智能」:表格的人工智能欄記錄工具在這一課的角色,學生的思考記錄在學習模式、活動與評鑑三欄。一份教案要拿分,靠的是這三欄填得出來;工具名稱寫得漂亮,補不上空白的一欄。The panel scores three kinds of thinking: scientific, creative and computational. None of the three has a column called artificial intelligence. The AI column records the tool's role in the lesson; the student's thinking is recorded in the learning-mode column, the activity column and the evaluation field. An entry earns its marks when those three can be filled in, and a well-written tool name cannot cover a blank column.

摘要Summary

  • 第二階段的截止日期是 2027 年 3 月 1 日,每所學校只可以交一份教案,結果在 2027 年 6 月通知。Phase two closes on 1 March 2027, each school submits one teaching plan, and results are announced in June 2027.
  • 對象是「『智』為學理」撥款計劃的參與學校;完成獎勵計劃其中一個階段,視同已發展撥款要求的一個教學例子。The scheme is open to schools in the AI for Science Education funding programme, and completing one phase of the award counts as one of the teaching examples that programme requires.
  • 電子表格的下拉選單只有課程框架(2025)的 13 個單元、59 個子單元與 331 條學習成果,教案要先落在其中一個子單元。The form's dropdowns carry only the 2025 framework's 13 units, 59 sub-units and 331 learning outcomes, so the plan has to sit inside one of those sub-units first.
  • 八行課堂設計各要一個科學實踐過程、一種學習模式、一個人工智能角色與分鐘;表格會按學習模式把分鐘加起來。Each of the eight lesson-design rows takes a science practice, a learning mode, an AI role and minutes, and the form adds the minutes by learning mode.
  • 官方的示範課全長 60 分鐘,被動 5 分鐘、主動 40 分鐘、建構 15 分鐘、互動 0 分鐘,五行之中只有一行用了人工智能。The official sample runs 60 minutes: 5 passive, 40 active, 15 constructive and 0 interactive, with AI in one of its five rows.
  • 評審只看科學、創意與運算思維;附件二把每一種寫成學生在課堂做的事,材料要顯示學生做過,評審才看得到。The panel scores scientific, creative and computational thinking, and the appendix writes each one as a student action, so the material has to show them doing it.
  • 兩個 150 字欄位問兩件事:工具解決了哪一個教學困難,以及用甚麼數據證明這一課有效。Two fields of 150 characters ask two questions: which teaching difficulty the tool solves, and what data shows the lesson worked.
  • 格式與署名規則決定材料會不會被讀:中文六頁、英文八頁、12 號字、不得出現校名與品牌、校長簽署;提交之後版權歸教育局。Format and signature rules decide whether the material is read at all: six Chinese pages or eight English ones at 12 point, no school or brand names, the principal's signature, and copyright to the Bureau after submission.

這份教案為甚麼現在要寫Why this plan is worth writing now

第二階段由現在到 2027 年 3 月 1 日[1],大約三個半月。這個窗口要做完三件事:在課堂上教一次、留下學生的作品、讓校長簽署;三件事都要排進校曆。Phase two runs from now to 1 March 2027[1], about three and a half months. Three jobs have to be done inside that window: the lesson taught once, the student work kept and the principal's signature obtained; all three go into the school calendar.

  1. 01
    2026 年 10 月:第二階段開放October 2026: phase two opens

    計劃詳情在 9 月發出,網頁在 10 月更新,第二階段現已接受報名。The scheme details were issued in September and the page was revised in October, when phase two opened.

  2. 02
    2027 年 3 月 1 日:交件截止1 March 2027: entries close

    電子表格、教學材料與校長簽署的申請表格,同日交回教育局科學教育組。The electronic form, the teaching material and the signed application form reach the Science Education Section that day.

  3. 03
    2027 年 6 月:結果通知June 2027: results

    金、銀、銅獎與證書,獲獎教師可能獲邀在教師專業培訓場合分享。Gold, silver and bronze awards with certificates, and winners may be invited to share at professional training events.

  4. 04
    2027 年 9 月 30 日:撥款總結報告30 September 2027: the grant's final report

    同一份教案可以算作撥款要求的一個教學例子,報告要交代三項承諾的完成情況。The same plan can count as one of the teaching examples the grant requires, and the report accounts for all three commitments.

四步在同一年之內:交件在 3 月,結果在 6 月,撥款總結報告在 9 月。Four steps inside one year: the entry in March, the results in June, the grant's final report in September.

撥款計劃的參與學校有 221 間(截至 2025 年 6 月,星島頭條)[10]。每所學校要發展最少兩個教學例子、舉辦最少一次公開課或示範課、最少一次經驗分享會,並在 2027 年 9 月 30 日或之前交總結報告。撥款網頁寫明:完成教案設計獎勵計劃其中一個階段,視同已發展一個教學例子[9]。交一份教案,可以同時應付兩份文件的要求。The funding programme had 221 participating schools as of June 2025 (Sing Tao Headline)[10]. Each school has to develop at least two teaching examples, hold at least one open or demonstration lesson and at least one sharing session, and submit a final report by 30 September 2027. The programme's page states that completing one phase of the award scheme counts as one teaching example developed[9], so a single entry serves both documents.

課程框架在 2027/28 學年由中一級開始實施(教育局通函第 14/2025 號),學校最早可以在 2025/26 學年試行[8]。2027 年 3 月交的教案,正好是新課程全面落地之前的一個試行版本。課堂教完,學校知道這一個單元在 40 分鐘內是否可行;這個答案比獎項耐用。The framework takes effect in Secondary 1 from the 2027/28 school year (EDB Circular 14/2025), and schools may trial it from 2025/26[8]. A plan submitted in March 2027 is therefore a trial version written just before the curriculum becomes compulsory. Once the lesson has been taught, the school knows whether the unit fits into 40 minutes, and that answer outlasts the award.

一份教案要交的,是一個課節的證據。學校不用在 3 月之前寫完一份年度計劃,只需要在 3 月之前教完一節課,並把那一節課的四份紀錄留下:教案、學生作品、人工智能的輸出、學生的修正。四份紀錄齊了,表格的八行與兩個 150 字才有內容可填。What the entry needs is the evidence of one period. A school does not have to finish an annual plan by March; it has to teach one lesson by March and keep four records from it: the plan, the student work, the AI output and the student's correction. With those four, the eight rows and the two 150-character fields have evidence to work from.

交件之前,八條規則決定這份教案寫得成甚麼樣子Before anything is written, eight rules decide what the plan can be

交件之前有八條規則,其中三條會改變教案的寫法:匿名、頁數、課堂長度。其餘五條決定交件的程序,包括誰可以交、交幾份、誰簽署、幾時截止。Eight rules stand between a lesson and a submitted entry, and three of them change how the plan is written: anonymity, page count and lesson length. The other five settle the procedure: who may enter, how many entries, who signs and by when.

規則Rule 內容What it says
對象Who may enter「『智』為學理」撥款計劃的參與學校。Schools taking part in the AI for Science Education funding programme.
名額How many每所學校只可以提交一份教案。One teaching plan per school.
課堂長度Lesson length不多於兩個課節,大約 80 分鐘。No more than two periods, about 80 minutes.
中文材料Chinese material最多六頁,12 號字,單行間距。Six pages at most, 12 point, single spacing.
英文材料English material最多八頁,12 號字,單行間距。Eight pages at most, 12 point, single spacing.
匿名Anonymity不得包含學校、公司及品牌的名稱、標籤與商標。No school, company or brand name, label or trademark.
簽署Signature申請表格由校長簽署,並蓋上學校印章。The principal signs the application form and the school stamp is applied.
截止Deadline2027 年 3 月 1 日或之前交回教育局科學教育組。The entry reaches the Science Education Section by 1 March 2027.

匿名規則把學校最想展示的名字拿走。教案與材料之內,學校、公司與品牌的名稱、標籤和商標都不可以出現,評審只看課堂。六頁上限把一個單元的內容壓到一節課;80 分鐘上限令教師選一個科學實踐過程做深,把整個單元剪碎放進去,兩邊都做不好。The anonymity rule takes away the name a school most wants to show. No school, company or brand name, label or trademark may appear in the plan or the material, and the panel sees only the lesson. The six-page cap presses a unit's content into one period, and the 80-minute cap pushes a teacher to take one science practice deep; cutting a whole unit into pieces and pushing it in leaves both halves badly done.

版權條款要早知道。作品提交之後,版權屬於教育局;教育局可以把作品的全部或部分內容,連同學校名稱,用於非牟利的教育推廣,也可以修改、翻譯、改編、複製、派發及上載至互聯網,毋須再取得教師同意或付費[2]。學校交出的是一份可以公開的教學資源,這個決定應該由科組與校長一起做。The copyright term is worth knowing early. Once the work is submitted, the copyright belongs to the Bureau, which may use all or part of it, with the school's name, for non-profit educational promotion, and may modify, translate, adapt, copy, distribute and upload it without further consent or payment[2]. A school is handing over a teaching resource that can be published, and that decision belongs to the panel and the principal together.

電子表格的四部分各有上限。第一部分記位置,第二部分記工具與困難,第三部分記八行課堂,第四部分記數據。兩部分是 150 字,其餘兩部分是選單與表格。The four parts of the electronic form each have a limit. The first records where the lesson sits, the second the tool and the difficulty, the third the eight rows of the lesson, the fourth the data. Two parts are 150 characters, and the other two are menus and a table.

部分Part 問甚麼What it asks 限制Limit
(A) 課堂基本資料(A) Class details學校、年級、班別、課堂時間、單元、學習內容、學習成果與科學學習活動。School, level, class, lesson minutes, unit, learning content, learning outcomes and the type of science activity.三個學習成果,七類活動。Three outcomes, seven activity types.
(B) 數字教育工具(B) Digital tools用了哪一類工具,以及工具解決哪一個教學困難。Which kind of tool was used, and which teaching difficulty it solves.150 字。150 characters.
(C) 課堂設計(C) Lesson design八行的科學實踐、學習模式、學生活動、人工智能角色與分鐘。Eight rows of science practice, learning mode, student activity, AI role and minutes.八行,四種學習模式。Eight rows, four learning modes.
(D) 評鑑方法(D) Evaluation用甚麼數據評鑑這一課對三種思維的作用。What data shows what the lesson did for the three domains.150 字。150 characters.

課堂基本資料要填甚麼?What does the class-details part ask for?

第一部分填的是這一課在課程裡的位置:學校、年級、班別、課堂時間、單元、學習內容、最多三條學習成果,以及這一課屬於哪一類科學學習活動。除了學校、年級、班別與課堂時間,其餘都由下拉選單決定。The first part records where the lesson sits in the curriculum: the school, level, class, lesson minutes, unit, learning content, up to three learning outcomes, and the type of science learning activity. Apart from the school, the level, the class and the minutes, every field is a dropdown.

七類科學學習活動是觀察、分類、對照實驗、尋找規律、科學建模、工程設計與探索社會性科學議題。計劃詳情文件自己的例子選了觀察與探索社會性科學議題,配一節 40 分鐘的光合作用課。選多了,等於聲稱一節課做了幾種不同性質的探究。The seven activity types are observation, classification, controlled experiment, pattern seeking, scientific modelling, engineering design and exploration of socio-scientific issues. The scheme's own example ticks observation and socio-scientific issues for a 40-minute lesson on photosynthesis. Ticking more claims that one period ran several different kinds of enquiry.

學習成果一欄是下拉選單,只可以選課程框架為該子單元列出的句子[7]。這個限制對教案有好處:成果是官方的句子,評審可以把活動與成果逐句對上;教師自己寫一句「學生會明白密度」,評審無從核對。The outcomes field is a dropdown and accepts only the sentences the framework lists for that sub-unit[7]. The limit helps the plan: an official sentence can be matched against the activity, while a teacher's own line that students will understand density gives the panel nothing to check.

表格的四部分之下,是一個已經填好選項的結構。單元選單有 13 項,學習內容選單跟隨單元,學習成果選單再跟隨學習內容。59 個子單元的學習成果由 1 條到 12 條不等:9.1 力和運動的特徵有 12 條,2.2 生命週期只有 1 條。一節 40 分鐘的課,通常只能覆蓋其中兩至三條。Under the four parts sits a structure whose options are already written. The unit menu carries 13 items, the learning-content menu follows the unit, and the outcomes menu follows the learning content. The 59 sub-units carry between 1 and 12 outcomes: 9.1, on the characteristics of force and motion, has 12, and 2.2 Life Cycles has 1. A 40-minute lesson can usually cover two or three of them.

由左至右:13 個單元收窄到 331 條學習成果,最後填入表格的只有三個欄位;9.1 力和運動的特徵有 12 條,2.2 生命週期只有 1 條。Left to right: 13 units narrow to 331 outcomes, and three fields reach the form; 9.1, on the characteristics of force and motion, has 12 outcomes, and 2.2 Life Cycles has 1.

選單影響時間分配。教師先選一個子單元,數一數它有幾條學習成果,再決定這一節課教哪兩條。選一個有 12 條成果的子單元,這一節課只能交代其中兩條,其餘十條要另找時間。The menu shapes the time. A teacher picks a sub-unit, counts its outcomes and decides which two the lesson will teach. A sub-unit with 12 outcomes leaves this period covering two of them and the other ten for another week.

一個實用的做法,是把選定的三條成果抄進備課紙,寫活動時逐條指回其中一條。三條之中有一條對不上任何活動,就要改活動或改成果,趁未教之前改。A practical habit is to copy the chosen outcomes onto the lesson-planning sheet and point each activity back to one of them. If one outcome has no activity, change the activity or the outcome before the lesson is taught.

年級與班別兩欄看似瑣碎,卻決定評審怎樣讀這份教案。中一與中三的同一個課題,學生的推理能力不同,活動的難度也應該不同;班別寫得清楚,評審知道這一份教案是為哪一群學生寫的。The level and class fields look minor, and they decide how the panel reads the plan. The same topic at Secondary 1 and Secondary 3 asks for different reasoning, and the activity should differ too; a clear class name tells the panel which group of students the plan was written for.

工具那一欄要寫的是困難,一個產品功能答不上The tool field asks about a difficulty, and a product feature cannot answer it

(B) 部分先選工具類別,再用 150 字回答一個問題:這個工具解決了教授這個課題時的哪一個困難或限制。計劃詳情文件的例子寫得具體:學生難以把「光強度與光合作用速率」這種抽象關係轉化為量化預測,所以用生成式人工智能提供模擬結果,讓學生在做實驗之前先練習分析數據。Part B asks for the tool type, then 150 characters on one question: which difficulty or limit in teaching this topic the tool solves. The scheme's own example is concrete: students struggle to turn an abstract relationship such as light intensity against the rate of photosynthesis into a quantitative prediction, so generative AI supplies simulated results and the students practise analysing data before the experiment.

這條問題要的是一個教學困難,一個產品功能答不上。寫「這個平台有 20 種模擬實驗」描述的是工具;寫「學生做對照實驗時分不清控制變量」描述的是學生,工具才有位置。The question asks for a teaching difficulty, and a product feature cannot answer it. "The platform has 20 simulations" describes the tool. "Students cannot tell the controlled variable apart in a fair test" describes the students, and only then does the tool have a place.

示範寫法,連標點 101 字A worked answer, 101 characters in Chinese with punctuation

學生看不見密度,判斷浮沉時只看重量,鋁塊比木塊重就以為一定沉。這一課讓人工智能先按重量預測三件物件的浮沉並說出理由,學生用量到的質量與體積算出密度,指出人工智能弄錯了哪一步,再把修正後的預測寫回工作表。students cannot see density, so they judge floating and sinking by weight, and an aluminium block that weighs more than a wooden one is expected to sink. In this lesson the AI predicts how three objects behave from their weight and gives its reason; students calculate density from the mass and volume they measured, show where the AI went wrong, and write the corrected prediction back onto the worksheet.

150 字的上限令答案要有一條線:困難、工具、學生的動作。三樣之中缺一樣,評審讀到的是一段產品說明,或者一段教學感想。The 150-character cap forces one line: the difficulty, the tool, the student's action. With one of the three missing, the panel reads a product description, or a reflection on teaching.

工具類別有七項:模擬實驗、線上平台、數據收集儀、流動裝置、機器學習系統、生成式人工智能與其他。一份教案可以選多於一項,選得越多,第二部分的 150 字越難寫得具體;兩個工具各自解決一個困難,比五個工具並列容易寫。The tool types are seven: simulation, online platform, data logger, mobile device, machine-learning system, generative AI and other. A plan may tick more than one, and the more it ticks the harder the 150 characters are to make concrete; two tools that each solve one difficulty are easier to write than five listed together.

八行課堂設計怎樣填?How should the eight lesson-design rows be filled?

八行對應八種科學實踐過程:提出科學問題、創建科學模型、計劃和進行探究、運算思維、分析數據、建構科學解釋、科學論証、科學傳意和評鑑。每一行要選一種學習模式、寫出學生活動、選一個人工智能角色、填上分鐘。表格會按學習模式把分鐘加起來,四種模式各佔多少,一眼看得見[3]。The eight rows correspond to eight science practices: asking scientific questions, creating scientific models, planning and carrying out investigations, computational thinking, analysing data, constructing scientific explanations, scientific argumentation, and scientific communication and evaluation. Each row takes a learning mode, the student activity, an AI role and the minutes. The form adds the minutes by learning mode, so the mix is visible at a glance[3].

四種學習模式的分別在學生做甚麼,文件寫得清楚:被動是學生聆聽教師提出的問題;主動是學生從提供的資料中選問題;建構是學生提出原創問題;互動是學生與同儕討論,提出原創問題。同一個實踐過程,四種模式是四種不同的課堂[5]。The four modes differ in what the student does, and the document is explicit: passive means the student listens to the teacher's question; active means the student chooses a question from the material provided; constructive means the student poses an original question; interactive means students discuss and pose an original question together. The same practice is four different lessons at the four modes[5].

科學實踐Practice 被動(P)Passive (P) 主動(A)Active (A) 建構(C)Constructive (C) 互動(I)Interactive (I)
提出科學問題Asking questions學生聆聽教師提出的問題。The student listens to the teacher's question.學生從提供的資料中選擇問題。The student chooses a question from the material provided.學生提出原創問題。The student poses an original question.學生與同儕討論後提出原創問題。Students discuss and pose an original question.
創建科學模型Creating models學生閱讀課本,認識一些科學模型。The student reads the textbook and meets some scientific models.學生按指示逐步建模。The student builds a model step by step on instruction.學生從自己的數據找規律並建模。The student finds a pattern in their own data and builds a model.學生以小組方式共建模型解釋現象。Students build a model together to explain a phenomenon.
計劃和進行探究Planning investigations學生閱讀實驗或探究步驟。The student reads the steps of the experiment.學生按固定步驟進行探究。The student runs the investigation on fixed steps.學生設計探究或寫出步驟。The student designs the investigation or writes the steps.學生以小組方式共同設計探究。Students design the investigation together.
運算思維Computational thinking學生聆聽教師講解如何表示與計算數據。The student hears how data are represented and calculated.學生按固定步驟表示與計算數據。The student represents and calculates data on fixed steps.學生用其他方法表示與計算數據。The student finds another way to represent and calculate the data.學生與同儕討論其他表示與計算方法。Students discuss other ways to represent and calculate the data.
分析數據Analysing data學生聆聽教師講解數據的規律與結果。The student hears the teacher explain the pattern and the result.學生按固定步驟分析數據。The student analyses the data on fixed steps.學生尋找其他資料,與自己的數據作進一步分析。The student brings in other information and analyses it against their own data.學生以小組方式合作尋找資料並分析。Students look for information and analyse it together.
建構科學解釋Constructing explanations學生閱讀課本或聆聽教師講授科學解釋。The student reads or hears a scientific explanation.學生按課業要求解釋科學現象。The student explains a phenomenon as the task requires.學生在不熟悉的情境中應用概念與數據解釋。The student applies the concept and the data in an unfamiliar situation.學生與同儕討論後,在不熟悉的情境中解釋。Students discuss and then explain in an unfamiliar situation.
科學論証Argumentation學生閱讀課本,了解議題的科學論述與證據。The student reads the textbook's account of an issue and its evidence.學生按「說法、證據、推理」的框架寫出論証。The student writes an argument on the claim, evidence and reasoning frame.學生分析數據、歸納證據並寫出論述。The student analyses the data, draws out the evidence and writes the argument.學生以小組方式分析後寫出不同論述並辯論。Groups analyse the data, write different arguments and debate them.
科學傳意和評鑑Communication學生聆聽教師講授科學結論。The student hears the teacher state the conclusion.學生按固定格式表達科學意念。The student expresses a scientific idea in a fixed format.學生創作表達科學意念,例如反思文章或海報。The student makes a reflection or a poster that expresses the idea.學生與同儕討論後創作表達科學意念。Students discuss and then make a reflection or a poster that expresses the idea.

官方的示範課把 60 分鐘分成被動 5、主動 40、建構 15、互動 0[3]。五行之中只有一行用了生成式人工智能,角色是輔助修訂。這個分配說明兩件事:一份教案不需要每一行都有人工智能,也不需要每一行都是建構。The official sample splits 60 minutes into 5 passive, 40 active, 15 constructive and 0 interactive[3]. One of its five rows uses generative AI, in the revision role. The distribution settles two questions: not every row needs AI, and not every row needs to be constructive.

八行不必填滿。官方示範填了五行,其餘三行留空。填滿八行而每一行的活動都是觀看影片,學習模式一欄會出賣它。The eight rows do not have to be full. The official sample fills five and leaves three empty. An entry that fills all eight with watching a video is given away by the learning-mode column.

每一行的人工智能角色有四種選擇,官方文件把它們寫成四種互動方式,各配一個例子。四種角色沒有優劣,分別在學生要付出哪一種思考[6]。Each row takes one of four AI roles, which the official document writes as four kinds of interaction with an example each. No role is better than another; they differ in the thinking they ask of the student[6].

人工智能角色AI role 人工智能做甚麼What the AI does 學生要展示甚麼What the student shows 學校留下甚麼What the school keeps
不涉及生成式人工智能No generative AI學生在探究過程中不使用人工智能。Students use no AI in the investigation.學生在沒有幫助下完成探究。The student completes the investigation unaided.實驗設計、數據與報告。The design, the data and the report.
輔助構思Ideation人工智能只提供意念與資料,不生成內容。The AI offers ideas and sources and generates no content.學生如何選取與改良意念。How the student selects and improves an idea.方案比較與最後採用的設計。The comparison of options and the design used.
輔助修訂Revision人工智能只改進行文表達,不生成探究內容。The AI improves the expression only and generates none of the investigation.學生如何評鑑與修正修訂稿。How the student judges and corrects the revision.學生的原稿與修訂稿。The student's draft and the revision.
完成任務讓學生評鑑AI completes, student judges人工智能生成內容並完成整個探究。The AI generates the content and completes the whole investigation.學生如何指出錯誤、要求修正並說明理由。How the student shows the error, asks for a correction and gives the reason.對話紀錄與修正後的結果。The conversation record and the corrected result.

第四種最容易出事。人工智能完成整個探究,學生要做的正是評鑑與修正;少了這一步,交上去的是一份列印本,學生沒有做過任何一種思考。附件二的三種思維,在第四種模式裡全部落在學生的評鑑上。The fourth role fails most often. The AI completes the whole investigation, and the student's job is precisely to judge and correct it; without that step the entry is a printout and the student has done none of the thinking. In the fourth mode, all three domains in the criteria rest on the student's judgement.

四種角色本身不計分。表格不會因為選了生成式人工智能而加分,也不會因為有一行不涉及而扣分。評審讀的是八行的分鐘分配與活動,看學生在哪幾行真正做了思考。The four roles carry no marks of their own. The form does not reward a row for using generative AI, and it does not punish a row for leaving it out. The panel reads the minutes and the activities across the eight rows to see where the student actually did the thinking.

示範:一節四十分鐘的密度課,八行這樣分配A worked 40-minute lesson on density, row by row

以下是一份示範,用中一單元七 7.2 密度,一節 40 分鐘。示範由我們寫,用來顯示八行可以怎樣分配;它未曾交件,也與任何一間學校的教案無關。What follows is a demonstration that uses 7.2 Density in Unit 7 at Secondary 1, in one 40-minute period. We wrote it to show how the eight rows can be allocated; it has never been submitted, and it is not any school's plan.

這一課選了 7.2 的兩條學習成果:運用公式計算密度,以及比較物體與水的密度,推斷浮沉。第三欄留空,因為一節課只教這兩條。The lesson uses two of the 7.2 outcomes: applying the formula to calculate density, and comparing an object's density with water's to infer whether it floats. The third field is left empty because one period teaches these two.

科學實踐Practice 學習模式Mode 學生活動Student activity 人工智能角色AI role 分鐘Min
提出科學問題Asking questions主動(A)Active (A)學生看木塊、鋁塊與膠匙在水中的浮沉,從三件物件中選出要解釋的一件。Students watch a wooden block, an aluminium block and a plastic spoon in water, and choose the one they will explain.不涉及None55
計劃和進行探究Planning investigations主動(A)Active (A)學生量度三件物件的質量與體積,用公式算出密度,寫下預測。Students measure the mass and volume of the three objects, calculate density and write a prediction.不涉及None1010
分析數據Analysing data建構(C)Constructive (C)人工智能按重量預測浮沉並說出理由;學生用自己算出的密度指出它弄錯了哪一步。The AI predicts floating and sinking from weight and gives its reason; students use their own density figures to show which step it got wrong.完成任務讓學生評鑑AI completes, student judges1010
建構科學解釋Constructing explanations建構(C)Constructive (C)學生寫出浮沉的解釋,人工智能修改行文,學生核對數字與因果,保留原稿。Students write the explanation, the AI improves the wording, and students check the numbers and the reasoning and keep their draft.輔助修訂Revision1010
科學論証Argumentation互動(I)Interactive (I)兩組交換預測與數據,辯論哪一件物件最難預測,並指出影響結果的變項。Two groups swap predictions and data, debate which object is hardest to predict, and name the variables that affect the result.不涉及None55

五行合共 40 分鐘:主動學習 15 分鐘、建構學習 20 分鐘、互動學習 5 分鐘,被動學習是 0 分鐘。表格右邊會按這四種模式把分鐘加起來,一看就知道這一課的思考放在哪裏。The five rows make 40 minutes: 15 active, 20 constructive, 5 interactive and no passive time. The form adds those four figures, so where the thinking sits is visible at once.

  • 主動學習 15 分鐘Active 15 minutes
  • 建構學習 20 分鐘Constructive 20 minutes
  • 互動學習 5 分鐘Interactive 5 minutes
  • 被動學習 0 分鐘Passive 0 minutes

一條 40 分鐘的課,按五行的分鐘數畫成比例:建構學習佔一半,被動學習沒有出現。The 40-minute lesson drawn to scale from the five rows: constructive work is half of it, and passive time does not appear.

這個分配把 40 分鐘的一半放在建構學習:分析數據與建構解釋,兩行都是建構,加起來 20 分鐘。學生在同一個模式裡做兩件事:先用數據反駁人工智能,再寫自己的解釋。被動是 0 分鐘,因為這一課沒有教師講解環節;需要先講一次公式的班,可以把第 2 行改為被動 5 分鐘、主動 5 分鐘,總數仍然是 40。Half of the 40 minutes sits in constructive work: analysing data and constructing an explanation, 20 minutes between them. In the same mode the students do two pieces of work: they use their data to rebut the AI, then write their own explanation. Passive time is zero because the lesson has no teacher exposition; a class that needs the formula explained first can split row 2 into 5 passive and 5 active minutes and keep the total at 40.

三種思維在這一課都有落點。科學思維在變項與公平測試:三件物件、同一杯水、同一個量度方法,學生要說明為甚麼這個比較公平。運算思維在數據表與模型:密度是一個模型,人工智能的「重量模型」與真實世界不符,學生指出它的限制。創意思維在第五行:兩組要提出另一個可以推翻預測的測試,例如同體積、不同質量的物件。All three domains have a place in this lesson. Scientific thinking sits in the variables and the fair test: three objects, one cup of water, one measuring method, and a student who can say why the comparison is fair. Computational thinking sits in the data table and the model: density is a model, the AI's weight-only model does not match the world, and the students name its limit. Creative thinking sits in row five, where each group proposes another test that could overturn the prediction, such as objects of the same volume and different mass.

這一課留下四份紀錄:八行的教案、學生的密度工作表、人工智能的預測與理由、學生寫在表上的修正。四份紀錄放進交件材料,評審看到的是一個學生做過、改過、寫過的課堂。The lesson leaves four records: the eight-row plan, the students' density worksheet, the AI's prediction and reason, and the corrections the students wrote on the sheet. Put those four records into the submission and the panel sees a lesson the students worked in, corrected and wrote about.

示範選一個學生未學過的課題,因為預測才有意義。學校選單元時,選一個學生在這個學期還未做的子單元;學生已經知道答案,人工智能的錯誤就沒有人會認真反駁。The demonstration uses a topic the students have not met, because a prediction only matters before the answer is known. A school choosing its sub-unit should pick one the class has not done this term; once students know the answer, nobody bothers to rebut the AI.

評審看哪三種思維?Which three kinds of thinking does the panel score?

評審按科學思維、創意思維與運算思維三方面評分。附件二把每一方面寫成一組教師可以安排的教學行為:科學思維七條、創意思維六條、運算思維八條[4]。附件二在每一組之前都寫明「教師可考慮」,動作的接受者是學生;材料要顯示學生做過,評審才看得到。The panel scores scientific, creative and computational thinking. The appendix turns each into a set of teaching moves a teacher can arrange: seven for scientific thinking, six for creative and eight for computational[4]. Each group begins with what the teacher may consider, and the action lands on the student; the material has to show the student doing it before the panel can see it.

運算思維最容易被寫成口號。附件二開宗明義:運算思維不限於電腦編程,指的是用系統性、數據化、模型化的方法解決科學問題。它列出的做法包括分解任務、發現並修正錯誤、整理數據找出規律與異常、畫圖與建立數據表、先提假設再用數據反證,以及比較模型與真實世界的異同、理解模型的限制。Computational thinking is the one most easily reduced to a slogan. The appendix says at the outset that computational thinking is not limited to programming: it means solving science problems in a systematic, data-based and model-based way. Its list covers breaking a task into steps, finding and correcting errors, organising data to find patterns and anomalies, drawing graphs and building tables, stating a hypothesis and testing it against data, and comparing a model with the real world to understand the model's limits.

創意思維也容易被讀成美感。附件二列的是多元與原創的意念、開放性問題、多於足夠的材料、小組比較不同方案、分析方案的優點與限制,以及解釋改良後的方案為甚麼更有效。一張畫得漂亮的海報,對不上其中一條;一個可以改的測試方案,對得上五條。Creative thinking is easily read as decoration. The appendix lists multiple and original ideas, open questions, more material than the task needs, groups comparing different solutions, analysing a solution's strengths and limits, and explaining why a revised solution works better. A neatly drawn poster matches none of them; a test design that can be revised matches five.

科學思維的七條之中,有兩條最常被跳過:搜尋和比較不同來源的資訊,以及判斷資料的可信度與偏差。這兩條正好是生成式人工智能進入課堂之後最需要的一種能力:學生要判斷一段生成內容可不可信,憑的是課程知識與自己量到的數據。Two of the seven scientific-thinking bullets are most often skipped: searching for and comparing information from different sources, and judging a source's credibility and bias. Those two are exactly the ability a generative-AI classroom needs most: a student judges a generated claim against subject knowledge and their own measurements.

評審範疇Domain 附件二列出的做法What the appendix lists 交件材料放甚麼What the material carries
科學思維Scientific thinking用貼近生活的現象引發探究、用概念解釋觀察、辨識自變量與因變量、設計公平實驗、安排重複測試、比較不同來源、判斷可信度與偏差、檢視論證中的謬誤。Starting from a familiar phenomenon, explaining an observation with a concept, identifying independent and dependent variables, designing a fair test, repeating a measurement, comparing sources, judging credibility and bias, and testing an argument for unsound assumptions.變項表與實驗設計、重複量度的數據、學生對人工智能說法的判斷。The variables table and the design, the repeated measurements, and the student's judgement of the AI's claim.
創意思維Creative thinking鼓勵多元與原創意念、設置開放性問題、提供多於足夠的材料、小組比較不同方案、分析方案的優點與限制、解釋改良後的方案為何更有效。Encouraging multiple and original ideas, setting open questions, providing more material than the task needs, having groups compare solutions, analysing strengths and limits, and explaining why a revision works better.兩至三個方案的比較、學生寫下的改良理由、被推翻的預測。A comparison of two or three options, the student's written reason for the revision, and a prediction that was overturned.
運算思維Computational thinking分解任務與設計流程、發現並修正錯誤、整理數據找出規律與異常、畫圖與建立數據表、先提假設再用數據反證、建立模型、比較模型與真實世界。Breaking a task into steps, finding and correcting errors, organising data for patterns and anomalies, drawing graphs and building tables, stating a hypothesis and testing it against data, building a model, and comparing the model with the real world.學生的數據表與圖、模型預測與實驗結果的差距、學生指出模型限制的那一段。The student's table and graph, the gap between the model's prediction and the result, and the paragraph where the student names the model's limit.

三種思維沒有一一對應三個章節。一節課可以同時有兩種:學生用自己的密度數據反駁人工智能,同時用到科學思維與運算思維;再提出另一個測試,用到創意思維。評審在材料裡找的是這些場面,找不到就當作沒有發生。The three domains do not map one-to-one onto three sections. One lesson can carry two at once: students rebut the AI with their own density data, which uses scientific and computational thinking, and then propose another test, which uses creative thinking. The panel looks for those moments in the material, and a moment that is not there is treated as not having happened.

材料六頁的空間有限,三種思維不可能平均分配。示範的密度課把重點放在科學思維與運算思維,創意思維只佔第五行 5 分鐘;這個比例要在交件時說清楚,評審知道學校選了哪一種深度。Six pages do not allow an even spread across three domains. The density demonstration puts its weight on scientific and computational thinking and gives creative thinking 5 minutes in row five; the submission should state that proportion, so the panel knows which depth the school chose.

評鑑一欄要寫的是數據,投入程度答不上The evaluation field asks for data, and engagement cannot answer it

(D) 部分只有一條問題:你會如何收集數據,評鑑這份教案如何幫助學生掌握科學思維、創意思維及運算思維。150 字。答案要有三樣:一個可以在課前課後各做一次的工具、一組可以比較的數字、一個說明數字與思維之間關係的句子。Part D asks one question: how will you collect data to evaluate how the plan helps students grasp scientific, creative and computational thinking. 150 characters. The answer needs three parts: an instrument that can be used before and after the lesson, a set of numbers that can be compared, and a sentence that connects the numbers to the thinking.

最常見的答案是「用問卷了解學生的投入程度」。投入程度可以量,卻答不上思維。附件二要的是學生能不能辨識變項、能不能比較模型與真實世界、能不能提出另一個方案;問卷要問到這些,才有用。The commonest answer is a questionnaire on how engaged students felt. Engagement can be measured, and it does not answer the question. The appendix asks whether students can identify variables, compare a model with the real world and propose another solution; a questionnaire has to ask those questions to be of use.

示範寫法,連標點 114 字A worked answer, 114 characters in Chinese with punctuation

課前用五條預測題問學生哪一件物件會浮,課後用同一組題目再問一次;另收學生的評鑑紀錄,按三項編碼,包括有沒有指出密度、有沒有引用自己量到的數據、有沒有改正人工智能的理由。兩組數字合起來,可以看到預測有沒有改變,以及改變有沒有理由。five prediction items before the lesson and the same five afterwards; the students' critique records are coded on three points, whether they named density, cited their own measurements and corrected the AI's reason. The two sets of numbers show whether the predictions changed and whether the change had a reason.

一節課的數據很薄,說得清楚的只有「這一班在這兩條成果上的表現有沒有改變」。要說得更穩,可以用兩個班:一班用第四種人工智能角色,一班不用,其餘相同。計劃沒有要求對照班;樣本小,如實寫出限制的答案,仍然比一個誇大的結論可信。One lesson yields thin data, and the honest claim is that this class changed on these two outcomes. A firmer claim needs two classes: one with the fourth AI role, one without, everything else the same. The scheme does not require a comparison class; with a small sample, an answer that states its limits is still more credible than an inflated conclusion.

附件二的三種思維是評鑑的對象,題目就要問到思維。問學生喜不喜歡這一課,答的是另一個問題;把那個答案放進評鑑欄,等於用投入程度代替學習。The appendix's three domains are what the evaluation is for, so the instrument has to ask about them. Asking whether students enjoyed the lesson answers a different question; putting that answer in the evaluation field substitutes engagement for learning.

交件時把工具一併放進材料:前測的五條題目、編碼表、學生的評鑑紀錄各一頁。評審要看到數字怎樣來,才信得過那個數字;只寫「數據顯示有進步」,沒有一頁可以翻,等於沒有數據。Put the instrument into the material: one page each for the five pre-test items, the coding sheet and a student critique record. The panel has to see where a number came from before it can trust the number; "the data shows improvement" with no page to turn to is the same as no data.

三種會被扣分的寫法,而且每一種都很常見Three ways an entry loses marks, and all three are common

第一種是人工智能代替學生思考。選了第四種角色,材料裡卻只有一份生成報告,沒有學生的評鑑與修正。附件二的三種思維在這一種寫法裡全部消失,因為學生沒有做過任何一步。The first is AI thinking for the student. The entry picks the fourth role, and the material holds only a generated report with no student judgement or correction. All three domains disappear in that version, because the student has done none of the steps.

第二種是標籤與活動不符。八行都填建構,活動卻是觀看影片、抄筆記。學習模式一欄的定義寫得清楚:建構是學生因應自己的數據尋找規律並建模。活動對不上,評審看到的是標籤,讀不到課堂。The second is a label that does not match the activity. All eight rows are marked constructive, and the activities are watching a video and copying notes. The learning-mode definition is explicit: constructive means the student looks for patterns in their own data and builds a model. When the activity does not match, the panel sees a label and cannot read a lesson.

第三種是兩個 150 字欄位寫成產品說明與承諾。(B) 寫工具的功能,(D) 寫「將會收集問卷數據」,兩欄都沒有一個數字、一個工具名稱或一個學生的動作。這兩欄是評審唯一讀到的教師自述,寫得空,前面八行也救不回來。The third is two 150-character fields turned into a product description and a promise. Field B describes what the tool does, field D says questionnaire data will be collected, and neither names a number, an instrument or a student action. Those two fields are the only first-person account the panel reads, and a blank there cannot be rescued by the eight rows above.

還有一種在評審之前就結束:材料出現校名或品牌、頁數超過上限、字號小於 12 號、一間學校交了兩份、申請表格沒有校長簽署。這些失誤與教案質素無關,卻令作品失去資格或不被讀完,交件之前要有人逐項核對[2]。One more ends before judging starts: a school or brand name in the material, more pages than the limit, type smaller than 12 point, two entries from one school, or an application form without the principal's signature. None of these is about the quality of the plan, and each can disqualify the entry or stop it being read to the end, so someone has to check the list before submission[2].

我們可以一起備課,代筆那份教案就留給科組We can plan the lesson with you, and the writing belongs to the panel

我們為學校安排到校 AI 課程,也做科學科的課堂與教師工作坊。對這份表格,我們可以做的有三件:與科組一起選子單元與學習成果、把八行設計排到真實的課堂時間、在課堂上留下學生的作品與評鑑紀錄。We run on-campus AI programmes for schools, and we teach science lessons and run workshops for teachers. On this form we can do three jobs: choose the sub-unit and the outcomes with the panel, fit the eight rows into real lesson time, and keep the student work and the critique record from the lesson.

我們不做的一件事,是替學校寫這份教案。計劃要求作品由教師原創、從未公開發表、沒有侵犯版權,提交之後版權歸教育局。一份由供應商代筆的教案,把學校的名字放在校長簽署之上,原創者卻是供應商,這個風險不應該由學校承擔。One job we do not take on is writing the entry for the school. The scheme requires the work to be the teachers' original work, never published and free of infringement, and the copyright passes to the Bureau on submission. A plan written by a supplier carries the school's name above the principal's signature without being the teachers' work, and that risk should not sit with the school.

我們也不會預測獎項。評審由教育局初步遴選,再由評審團選出金、銀、銅獎;附件二列出的是範疇,沒有計分表,我們沒有評審團的評分方法。可以保證的只有一件事:交上去的教案,八行填得出來,兩個 150 字有數據,材料留得住學生的思考。We do not predict awards either. The Bureau screens the entries and a panel selects the gold, silver and bronze awards; the appendix lists domains and not a mark scheme, and we do not hold the panel's weighting. One promise can be kept: the plan that goes in has eight rows that can be filled, two 150-character answers with data behind them, and material that keeps the student's thinking.

這一頁寫的是 2026 年 11 月 17 日讀到的文件。附件二註明 2025 年 9 月,計劃詳情文件是 2026 年 9 月;教育局保留更改、暫停或終止計劃的權利,也保留更改獎項的權利。規則一改,這一頁要跟着改。This page describes the documents as they read on 17 November 2026. The criteria appendix is dated September 2025 and the scheme details September 2026; the Bureau reserves the right to change, suspend or end the scheme and to change the awards. If the rules change, this page has to change with them.

常見問題Questions

學校沒有參加「智為學理」,可以交教案嗎?May a school outside the funding programme enter?
計劃的對象是「『智』為學理」撥款計劃的參與學校,而撥款計劃已經截止申請,其他學校目前不在對象之內。未參加的學校仍然可以用附件二的三組範疇檢視校本教案,把同一份八行設計留給校內觀課。The scheme is open to schools in the AI for Science Education funding programme, that programme has closed to applications, and other schools are outside its scope for now. A school that is not in it can still use the appendix's three domains to review its own plan and keep the same eight-row design for an in-school observation.
完成獎勵計劃其中一個階段,真的可以算一個教學例子嗎?Does completing one phase really count as a teaching example?
可以。撥款計劃網頁在「計劃要求」一項寫明:完成科學(中一至中三)教案設計獎勵計劃其中一個階段,視同已發展一個教學例子或資源[9]。學校仍然要為這個例子準備教學方案與教學資源,交總結報告時一併提交。Yes. The funding programme's page states under its requirements that completing one phase of the Science (S1-S3) Teaching Plan Design Award Scheme counts as one teaching example or resource developed[9]. The school still prepares the teaching plan and the resources for that example and submits them with the final report.
教案一定要用生成式人工智能嗎?Must the plan use generative AI?
計劃要求把數字教育工具(包括人工智能)融入課堂。電子表格的八行之中,每一行都可以選「不涉及GenAI」,四種角色也把「不涉及」列為第一種。整份教案完全不用人工智能,難以回應計劃的目的;官方的示範課五行之中只有一行用了人工智能,其餘由學生自己動手。The scheme asks for digital tools, including AI, to be built into the lesson. Every one of the eight rows can take no generative AI, and the four roles list that option first. A plan that uses no AI at all is hard to square with the scheme's purpose; the official sample puts AI in one of its five rows and leaves the rest to the students.
可以用英文交嗎?May an entry be written in English?
可以。計劃詳情文件的格式表列出中文六頁、英文八頁,兩者都是 12 號字、單行間距。每所學校只可以交一份,語言版本要在交之前選定。Yes. The format table in the scheme details allows six Chinese pages or eight English ones, both at 12 point with single spacing. Each school submits one entry, so the language is settled before submission.
一份教案可以幾位教師署名?How many teachers may be named on one entry?
申請表格有「負責教師」一欄,填姓名、聯絡電話與電郵;另有「參與教案設計的教師」欄,可以填多於一位。獎項的證書發給獲獎教師。The application form names a lead teacher with a phone number and an email, and it carries a separate list of teachers who took part in the design, which can hold more than one name. Certificates go to the teachers who win.
獲獎作品的版權誰屬?Who holds the copyright in a winning entry?
作品提交之後,版權屬於教育局。教育局可以把作品的全部或部分內容,連同學校名稱,用於非牟利的教育推廣,也可以修改、翻譯、改編、使用、複製、派發及上載至互聯網,毋須再取得教師同意或付費。教師要確保作品原創、從未公開發表、沒有侵犯版權,引用資料要註明來源。Once submitted, the copyright belongs to the Bureau, which may use all or part of the work, with the school's name, for non-profit educational promotion, and may modify, translate, adapt, use, copy, distribute and upload it without further consent or payment. The teachers have to ensure the work is original, never published and free of infringement, and that every source is acknowledged.
第一階段與第二階段有甚麼分別?How does phase two differ from phase one?
兩個階段用同一份評審範疇文件:第二階段網頁連結的附件二註明 2025 年 9 月[4]。第一階段的評審在 2026 年 3 月開始(星島頭條,2025 年 8 月 1 日);第二階段的截止日期是 2027 年 3 月 1 日,結果在 2027 年 6 月通知。Both phases rest on the same criteria document: the appendix linked from the phase-two page is dated September 2025[4]. Phase-one judging began in March 2026 (Sing Tao Headline, 1 August 2025); phase two closes on 1 March 2027, with results announced in June 2027.