What is PG2 in Sengkang | Sengkang Education AI Literacy, Code for Fun and Deepfake Safety 2027

What changes for PG2 pupils in 2027 AI education?

Will PG2 students in Sengkang learn about artificial intelligence in 2027? They can encounter AI literacy through school learning, cyber wellness and the updated Code for Fun programme. In March 2026, MOE announced that Code for Fun would be updated to strengthen AI skills and made available to all primary and secondary schools in 2027. This is a policy about programme availability; it is not a promise that every pupil will have exactly the same AI lesson or receive a particular app.

PG2 means Posting Group 2, the pupil’s route for Secondary 1 admission. It is not a separate AI curriculum, and it does not decide whether the child has the ability to learn programming, question a machine-generated claim or identify misleading information. Those skills can be developed through appropriately designed lessons, projects and independent practice, with school expectations matched to the actual student’s needs.

Families in Compassvale, Rivervale, Fernvale and Anchorvale should begin with three practical questions: What AI or digital-literacy activities does the assigned secondary school actually offer? What are its rules for using AI in homework and projects? Can the pupil explain, verify and improve a result without simply copying the output? This guide provides a clear pathway from first principles to responsible use, with worked examples, a family safety checklist and links to current MOE announcements.

The 2026 MOE announcement, stated precisely

MOE’s 2026 Committee of Supply announcements described three related developments: defined AI-literacy milestones throughout education; an updated Code for Fun programme for primary and secondary schools made available to all schools in 2027; and updated Cyber Wellness lessons within Character and Citizenship Education (CCE), including validation of generative-AI information and recognition of deepfakes.

These measures should not be collapsed into a headline such as ‘Every Secondary 1 PG2 student must code an AI model by January 2027’. That is not what the announcement says. A programme being available to schools does not establish the exact timing, cohort participation, contact hours, teaching software or assessments at any one Sengkang campus. Parents should check the current school’s programme information.

The announcement also did not say that AI will replace teachers, that AI-generated essays automatically count as the child’s work, or that pupils should upload private school materials to public websites. The more useful interpretation is that digital tools are becoming part of a learning environment in which critical judgment, safety, explanation and human responsibility remain essential.

Parents may hear very different claims from social media, a tuition advertisement and the school website. Prefer the official MOE statement for the national policy and the actual school notice for implementation. If a detail is missing, ask the school rather than treating another campus’s promotional video as proof of a universal requirement.

AI literacy is not the same as knowing how to type prompts

AI literacy involves understanding what a system can do, where its outputs may be unreliable, when it is useful, what information must be protected and how humans remain accountable for decisions. Writing a prompt can be one practical skill, but generating a neat paragraph without checking its truth is not a complete demonstration of understanding.

A pupil can show AI literacy by comparing an answer to a primary source, noticing when a model has invented a reference, asking whether the data behind a claim are representative, or recognising that an edited image does not prove an event occurred. Those habits matter in English, Science, Humanities, Mathematics and daily life.

An everyday example is a student asking an AI assistant to explain why a metal object feels colder than wood in the same room. The output might be fluent, but the pupil should check the relevant heat-transfer concept, distinguish temperature from rate of heat transfer and test whether the explanation agrees with classroom learning. The quality of the learning lies in the check and explanation, not the smoothness of the generated words.

Likewise, an AI answer can be useful for brainstorming a topic or identifying questions for further reading, but the student needs to decide what is relevant and correct. The child should not mistake the tool’s confidence for evidence. An uncertain answer labelled carefully is often better than an invented precise number.

What is Code for Fun?

Code for Fun is an MOE-related programme that supports students in experiencing coding and computational thinking, with offerings and school implementation that can vary. MOE’s 2026 announcement says the programme will be updated to strengthen AI skills and made available to all schools in 2027. Parents should regard the school’s own programme page as the guide to whether their child will attend a particular module.

Coding is the act of creating and testing instructions that a computer can execute. Computational thinking includes decomposing a problem, recognising patterns, expressing a sequence and identifying mistakes. AI systems may use additional approaches that learn patterns from data. These topics are related, but a pupil should not be told that every simple conditional program is ‘artificial intelligence’.

A useful first project may involve giving a robot or on-screen character a sequence of instructions. If the character goes in the wrong direction, the student traces the steps and finds where the logic failed. That is a genuine learning task even without a large language model or neural network. Later activities may explore classification, patterns in training examples or responsible use of AI tools, subject to the school’s implementation.

Families should not rush to buy a subscription or external coding course because a 2027 policy mentions AI. Ask what the school already provides, whether it is a curricular activity or optional enrichment and what prior knowledge is actually needed. A thoughtful beginner project completed well can build more confidence than an expensive course that hides the underlying reasoning.

Six terms every new secondary student should understand

This vocabulary is a starting point, not a complete computer science syllabus. The appropriate depth depends on the learner’s age and the school’s lesson. A twelve-year-old can learn to recognise uncertainty and protect privacy long before they can explain how a large machine-learning model is trained.

Do not confuse the words ‘data’ and ‘fact’. A spreadsheet can contain wrong numbers; a training dataset can omit important groups; a model can produce confident conclusions based on patterns that do not support the claim. Asking where evidence came from is the central habit.

Term	A simple meaning	An important limitation
Algorithm	A defined set of steps or rules for a task	An algorithm is not necessarily AI.
Data	Recorded examples, measurements or information	Data can be incomplete, biased or incorrect.
Model	A representation used to make predictions, classify or generate output	Models can be wrong outside familiar cases.
Prompt	An instruction or question supplied to an AI tool	A clever prompt cannot guarantee truth.
Hallucination	A plausible-sounding output that states unsupported or fabricated content	Fluent wording is not evidence.
Deepfake	Media manipulated or generated to misrepresent someone or something	A convincing video does not prove it is authentic.

Worked example: a classifier can be wrong for good reasons

Imagine a simple demonstration program trained to classify photographs as ‘cat’ or ‘not cat’. The students show it cats, dogs, shoes and plants. It gets many classroom examples correct. The next photograph is a cat hidden behind a chair, and the system gives a wrong answer. This does not mean the photograph changed reality; it means the classification did not generalise well to that situation.

The pupil should ask which examples were used, whether training and testing cases are sufficiently varied, and what information is missing. Was the model relying on the background rather than the animal? Did it encounter unusual lighting? Was the input too blurry to determine? These are questions about evidence, limitations and generalisation.

Parents can discuss an everyday equivalent without technology: a child who has only seen red apples may initially assume that every apple is red. A green apple challenges the rule. Updating the rule after new evidence is a useful reasoning move. The analogy is simple, but it illustrates why one correct observation cannot establish a universal claim.

A classroom AI activity should use approved data, permissions and age-appropriate tools. Do not upload real classmates’ faces, names or photographs to an unknown online classifier merely to demonstrate the concept. If the lesson uses images, use school-approved or appropriately licensed examples.

Worked example: an AI-generated summary with a false source

A pupil asks a generative AI tool to summarise a historical event and receives a neat paragraph that includes a book title and page number. The student is tempted to use the citation in an assignment. But when the pupil checks the library catalogue or original source, that exact title or page cannot be found. The model may have invented a reference.

The correct action is not to ‘fix’ the paragraph by inventing another reference. Remove or qualify the unsupported statement, find a reliable source and rewrite the explanation based on verified information. If the class requires acknowledging AI use, disclose it honestly according to the teacher’s instructions.

An effective exercise is to mark every factual assertion in the generated paragraph and ask whether it can be checked. Distinguish a primary document, official statistics, a scholarly explanation and an unsourced online claim. The goal is not to forbid every tool; it is to prevent polished output from bypassing the work of finding and understanding evidence.

The pupil can also compare two sources that disagree. What date was each published? Are they describing the same period? Does a later source correct the earlier one? These questions are useful in History, Geography, Social Studies and news literacy, independent of any software.

Worked example: checking a mathematical answer without trusting tone

Suppose a student asks an AI system to solve 2(x + 4) = 18, and it responds with x = 13. The answer sounds confident, but substitution reveals the error: 2(13 + 4) = 34, not 18. The student should solve the equation themselves: x + 4 = 9, so x = 5. Checking gives 2(5 + 4) = 18.

This worked example shows an important principle: an answer is not correct because a machine presents it elegantly. The student should know which operation was applied, what remains equal and how to check the result. AI may be helpful in suggesting a different explanation, but the child must be able to justify the mathematics independently.

For a more advanced problem, the pupil might compare several methods and ask whether each works under the stated assumptions. A model can omit units, round too early or solve an equation that does not match the original word problem. The most robust habit is to return to the question and test feasibility.

Teachers can set their own rules for when AI help is permitted. A student should not enter an assessed question into a public tool when doing so violates school instructions. The aim of technology is to support authentic understanding, not replace it.

Worked example: a school notice that looks official but may be fake

Imagine a message circulating in a class chat that claims a Sengkang secondary school will change reporting time tomorrow. It includes a school logo and a screenshot of a supposed announcement. The pupil should not automatically forward it, because logos and screenshots can be copied or altered.

Check the official school channel, notice system or general office. Does the message appear in Parents Gateway or the school website? Does the date match? Is the instruction consistent with the normal reporting process? A screenshot without a verifiable source is weaker evidence than a current official notice.

If the school has not confirmed the message, the family can politely say ‘I cannot verify this yet; please check the school notice’ rather than accuse a classmate of deliberately lying. This models cautious communication. If there is a genuine emergency, follow verified official alerts and appropriate urgent instructions.

Students who learn to verify a fake school notice are practising the same skill needed to evaluate deepfake media, advertising claims and misleading AI-generated summaries. Critical checking is not a specialised computer-science trick; it is a general civic skill.

Deepfakes: why seeing is not always believing

A deepfake is content generated or manipulated to make an image, recording or video appear to depict something that did not occur as presented. Some manipulations are obvious, while others are convincing. A student should not be taught that one visual clue, such as a strange hand or imperfect lip movement, is a guaranteed deepfake detector. As tools improve, those clues can disappear.

A safer method starts with provenance and context. Who published the content? Is there an original source? Does it appear on a credible official channel? Is there independent corroboration? Could the clip have been cropped, relabelled or taken from another event? Can the claimed location or date be verified?

Consider a fictional video claiming that an MRT station in Sengkang has been closed indefinitely. The responsible response is to check official transport notices rather than repeatedly repost the alarming clip. A family may choose a precautionary travel option while seeking verification, but it should label uncertainty accurately.

If an altered image involves a real pupil or causes harassment, preserve enough evidence for a report but do not spread the image further. Notify a trusted school adult and follow the school’s safeguarding process. For serious threats or criminal concerns, seek appropriate official assistance. The lesson is not that all media are false; it is that trust should be based on evidence.

PG2 and digital ability: no permanent label

Some parents worry that a student who entered through PG2 must be less capable with technology than a pupil from PG3. That is an unsupported stereotype. Posting Groups are based on Secondary 1 admission eligibility. They do not measure a child’s creativity with code, ability to debug, cybersecurity awareness or judgment in evaluating evidence.

Under Full SBB, pupils may study individual subjects at different G levels and share form-class experiences. An interest in technology can develop through suitable school programmes, CCA activities and personal exploration, subject to access and safety. Students should not be pushed into technical enrichment simply to prove that a Posting Group label is wrong.

A pupil who enjoys sketching a user interface may contribute strongly to a group project. Another may reason well about algorithms but need help communicating decisions. Another may excel at questioning a source. These are complementary strengths. Teachers can support them through explicit tasks and feedback.

Parents can help by asking what the child tried, which error occurred and how they checked the result. A learner who can explain a debugging process has developed a useful skill regardless of whether a final programme looked impressive.

Privacy: what must not be pasted into a public AI tool

A sensible default is do not upload personal or sensitive school information to an unapproved public AI service. This includes classmates’ personal details, school IDs, private family information, medical reports, photographs of other minors without appropriate permission, private school notices, confidential assessments, passwords and one-time login codes.

Public tools may have different terms, retention settings and age requirements. Parents should check the school’s acceptable-use policy and the service’s conditions before allowing independent use. A student’s desire for a quick answer does not override another person’s privacy or the school’s information-handling rules.

If an exercise requires an example dataset, use fictitious names or school-approved anonymous information. Do not construct a demonstration by uploading a class list. If the pupil accidentally discloses personal information to an unfamiliar service, stop further sharing and seek advice from a parent or school staff member about appropriate next steps.

Use an analogy a young learner can remember: a public AI chat is not necessarily the same as a private notebook. Before pasting, ask whether the material belongs to you, whether it contains someone else’s information and whether you are authorised to share it.

Academic integrity: help versus doing the homework

AI can assist learning in several ways: explaining a definition in simpler language, suggesting a practice question, checking for missing steps or helping the pupil brainstorm topics. The boundary depends on the assignment. If the teacher requires independent writing, using AI to generate the complete response and submitting it as personal work may violate school rules.

A useful decision test has four questions: Was the tool allowed? Did the child understand the result? Did the final work remain their own where required? Was use disclosed if the teacher requested it? If any answer is unclear, ask the teacher before submission.

For a composition task, the pupil might write a draft first and use an approved tool to ask which sentence seems unclear. Then they revise and explain the change. That preserves authorship more effectively than submitting a fully generated essay that the learner cannot discuss.

Even when AI use is permitted, the student should check quotations and facts, read the original sources and avoid invented citations. A clear acknowledgement of tools is not an admission of failure; it is part of responsible communication when required by the assessment.

How Code for Fun fits with English, Mathematics and Science

English: A pupil can compare two explanations and test which is better supported by textual evidence. The transferable skill is close reading and clear reasoning, not merely generating fluent paragraphs.

Mathematics: A student can inspect an algorithm and trace why a sequence fails on one input. The transferable skill is logical structure and checking, useful for equations, data and graph interpretation.

Science: A pupil can examine how data quality affects a prediction, identify uncontrolled variables and distinguish correlation from cause. The lesson can be taught at an age-appropriate level without claiming that students must build an advanced statistical model.

Humanities and CCE: Students can compare claims, identify missing context, explore digital ethics and consider the consequences of misinformation. The focus is fair evaluation and social responsibility.

Art and D&T: Pupils may design a useful object or interface and evaluate whether a technological solution addresses the real user’s problem. The goal is a thoughtful design process, not adding AI to a project simply to make it sound modern.

Ten questions for assessing an AI answer

Parents can use two or three questions at a time instead of administering a ten-question oral examination after every homework task. The aim is to build habits. When a learner says ‘the source is missing’, that is a useful observation. When a learner checks a number and notices a unit error, that is progress.

An important part of critical thinking is knowing when to stop. A factual claim with no verifiable support should not become more believable merely because the pupil asks an AI system to rephrase it several times. Search for a reliable primary source or ask the teacher.

• Who or what produced this answer, and what task was it given?
• What evidence is provided for the most important factual claim?
• Can the cited source actually be opened and checked?
• Is the information current for the year and location discussed?
• Is the answer mixing a fact with an opinion or prediction?
• What assumptions are required for the answer to be valid?
• Are there people, groups or cases missing from the examples?
• Could the output reveal or misuse somebody’s private information?
• Can the student independently explain the result in their own words?
• If the answer is wrong, who is responsible for correcting it before use?

A safe family learning project: verify one AI claim

Choose a claim that is harmless, publicly checkable and connected to school learning—for example, the year a public library opened or a simple scientific fact. Ask the student to write the claim before consulting an AI tool. Then compare the response to an official or other reliable reference. Record whether the tool was accurate and how you established that.

Do not choose a classmate’s personal information, a medical diagnosis or an allegation about a local business. The goal is evidence checking, not gathering private facts or spreading rumours. A parent can provide a small set of trusted sources if the learner is new to independent research.

After the check, ask the pupil to write three short lines: Claim; source; conclusion. If the AI output was wrong, describe the error precisely. If it was right, explain why the independent evidence supports the claim. The verification process should remain the educational achievement, not scoring a point against the machine.

A student who completes this exercise carefully can later apply the same method to a Science result, an economic statistic, a news story or a dramatic video. The technique generalises beyond AI because the underlying skill is disciplined evaluation of evidence.

A safe family learning project: debug a simple sequence

Write an everyday algorithm for packing a school bag: read tomorrow’s timetable, select the required books, check attire, confirm assignments, place essentials in the bag and check the reporting time. Now introduce an error: the algorithm checks yesterday’s timetable instead of tomorrow’s. Ask the pupil which step fails and what consequence follows.

This is computational thinking without any special software. The student can propose a corrected sequence and test it across two different timetables. The exercise illustrates why an apparently sensible procedure can fail if the input is wrong.

Next introduce a conditional step: If tomorrow includes PE, pack the required attire; otherwise, check whether another activity requires special equipment. The pupil learns that decision rules need accurate conditions. A digital version might later be created with blocks or simple code in an approved programme.

The parent should avoid writing the whole solution. Ask the pupil to explain the logic and test a changed case. This is a useful bridge into Code for Fun while recognising that the school’s specific programming module may be entirely different.

A safe family learning project: a small data-quality experiment

Prepare a fictitious list of daily travel times for five school journeys: 18, 21, 20, 22 and 19 minutes. The mean is 20 minutes. Ask the student what happens if the sixth journey takes 50 minutes due to a disruption. The new mean is 25 minutes, showing why one unusual observation can alter a small dataset.

Now ask whether it would be accurate to claim that the journey ‘always takes 25 minutes’. Of course not; the observations vary. The student should distinguish a summary statistic from a guaranteed time. This is useful preparation for interpreting real reports and AI-generated predictions.

Do not use this fictional example as a travel estimate for an actual route in Sengkang. To plan a real school journey, use current transport information and test the practical route. The classroom lesson is about data interpretation and uncertainty, not giving directions.

Students can discuss what additional information would improve the estimate: more days, time of travel, rain, CCA dismissal and variability. They learn that a prediction requires both data and a statement of its limits.

Seven-day 2027 AI literacy starter routine

This is an optional family learning routine, not the national 2027 Code for Fun syllabus. Each activity can take a short conversation or brief practice task. Families should follow the school’s actual requirements, age rules and access policies for any digital application.

After the week, ask the pupil which habit was most useful. Do they check dates more carefully? Are they less likely to copy an unsupported answer? Can they name a privacy boundary? The evidence of progress is a more informed decision, not a certificate from a commercial AI platform.

Day	Short activity	What success looks like
1	Explain the difference between an algorithm and an AI model	Student gives one accurate example of each
2	Read an AI-generated explanation and identify a fact to verify	Student finds a suitable source
3	Solve a Maths example and check the answer by substitution	Student catches a plausible wrong answer
4	Compare a real official notice with a suspicious screenshot	Student checks provenance and date
5	Design a safe rule for school bag packing	Student tests and repairs a sequence
6	Discuss a deepfake scenario without forwarding the media	Student proposes a responsible reporting route
7	Review privacy and acceptable-use rules	Student can name information never to paste into public tools

What families in Sengkang should ask their schools

School administrators may not have a final 2027 lesson-by-lesson schedule available months in advance. That does not mean the announced national direction is false. It means detailed implementation belongs to each school and should be checked at the relevant time.

A parent can ask for the school’s official written acceptable-use guidance rather than rely on informal advice from another family. The answer should be specific enough for a pupil to understand what is allowed during study and what is not allowed in assessed work.

• What does the school currently mean by AI literacy for Secondary 1?
• Is Code for Fun part of the student’s school programme, an elective, or an additional opportunity?
• Which platforms are school-approved and what are the minimum age or consent requirements?
• How are AI tools permitted or prohibited in homework, projects and weighted assessments?
• How does the school teach verification, deepfake awareness and privacy?
• What support is available to students who lack a suitable device or reliable home internet?
• How do teachers respond when a pupil submits an AI-generated answer they do not understand?
• Which staff member should families contact about cyber wellness or unsafe online material?

When AI use becomes a wellbeing or safety issue

If a pupil uses AI tools late at night to avoid difficult homework, the problem may be an unresolved learning gap, anxiety or poor boundaries rather than enthusiasm for technology. Ask what part of the task feels impossible and help the child seek teacher feedback. Limiting screen exposure may help, but the educational problem also needs diagnosis.

If another student creates humiliating synthetic images, impersonates someone online or makes threats, preserve relevant evidence privately and inform a trusted adult or the school. Do not repost the material to collect votes on whether it is ‘real’. Serious threats require appropriate official assistance.

Some children may believe that an AI companion or bot is always trustworthy because it responds warmly. Teach that a program does not replace trusted adults and may present incorrect or unsafe guidance. Keep open conversations about online experiences so the child can ask for help without fear of immediate punishment.

The principle is consistent with MOE’s Cyber Wellness direction: technology needs judgment, empathy and responsibility. The objective is not to frighten students away from useful tools. It is to give them the ability and confidence to recognise limits and protect themselves and others.

Frequently asked questions

Does PG2 mean my child cannot join Code for Fun?

No. Posting Group 2 is an admission grouping. MOE announced that the updated programme would be available to all primary and secondary schools in 2027; individual school arrangements determine participation.

Will every Secondary 1 student be required to build an AI chatbot?

The 2026 policy announcement does not establish that universal requirement. Check the school’s actual programme and lesson objectives.

Is every algorithm AI?

No. An algorithm is a defined procedure; AI systems may use algorithms, but many ordinary algorithms do not learn from data or generate predictions.

Can students use AI for assessed homework?

Follow the teacher’s instructions. If independent authorship or tool disclosure is required, AI-generated work cannot simply be submitted as entirely the student’s own.

How can my child check an AI answer?

Use reliable primary sources, test calculations, verify dates and citations, and explain the result independently.

Can an AI model reliably identify every deepfake?

No universal detection method should be assumed. Source verification, context and independent corroboration remain essential.

Is the school’s PLD the same as an AI programme?

No. The PLD is a managed device. AI literacy is a set of knowledge and skills that may be taught through different approved resources.

Should parents buy a paid AI subscription for 2027?

Do not assume it is necessary. Ask the assigned school what is provided and what tools are permitted before spending.

What if the child shares private information in an AI chat?

Stop further disclosure and seek appropriate guidance from a parent or school adult. Do not share login credentials or other pupils’ private data.

Official sources and next Sengkang learning routes

Editorial verification: 6 November 2026. The claims about the 2027 Code for Fun availability and Cyber Wellness changes come from MOE’s 2026 announcement. All classroom projects, example prompts, school messages, pupil stories and hypothetical numbers are illustrations rather than a published syllabus or account of a named Sengkang pupil. Schools determine their 2027 programme, assessment and acceptable-use requirements.

• MOE 2026 Committee of Supply announcements — AI literacy, Code for Fun for 2027 and updated Cyber Wellness lessons.
• MOE Education Conversations — national curriculum and future-ready competency direction.
• MOE Full Subject-Based Banding — the meaning of Posting Groups and subject levels.
• Singapore Personal Data Protection Commission — wider context about respecting personal information.
• Cyber Security Agency of Singapore — national cyber-safety resources.
• eduKate Sengkang: Common Curriculum and CCE — shared subject context for cyber wellness.
• eduKate Sengkang: Home-Based Learning and SLS — managing independent digital work.
• eduKate Sengkang: PLD, Edusave and digital safety — the separate device and payment questions.
• eduKate Sengkang: bullying and cyberbullying reporting — action when online harm occurs.
• The Complete Sengkang Guide — estate, family and education references.

Continue the PG2 Sengkang learning systems and awards guides: Common Curriculum: Art, D&T, FCE, Music, PE and CCE · HBL, SLS and independent study · Edusave EAGLES, Merit Bursary and Good Progress awards · The Complete Sengkang Guide. These are independently researched education guides; consult school notices for individual eligibility and schedules.