Understand the AI
that shapes your world.
A practical, bilingual guide to artificial intelligence for Hong Kong students. Learn what AI really is, how it learns, where it is used, and how to use it honestly, safely and fairly.
🇭🇰 Aligned with the Hong Kong curriculum
This hub supports the Information Technology Learning Targets, the junior secondary Computer Literacy curriculum, senior secondary Information and Communication Technology (ICT), STEM education, and the cross-curricular Values Education and Media and Information Literacy frameworks. Each module lists its curriculum links.
What you will be able to do
🧠 Explain
Define artificial intelligence, machine learning and generative AI in your own words, with accurate examples.
🔍 Evaluate
Judge whether an AI output is reliable, identify bias and hallucination, and check sources before trusting an answer.
🛡️ Protect
Recognise privacy, security and deepfake risks, and know what personal data never to share with an AI tool.
🤝 Act ethically
Use AI honestly in schoolwork, disclose your use, and respect copyright and academic integrity.
Choose your learning path
How to use this hub
Work through the modules in order, or jump to whatever you need. Use the search box (/) to find any topic instantly.
Press Mark complete at the bottom of each module. Your progress is saved in this browser and shown in the sidebar.
Sort items in the "Is it AI?" activity and build a real prompt with the Prompt Builder.
Finish with the 12-question quiz, then lock in the vocabulary with flashcards.
💡 Tip
Everything here works offline and your progress stays on your device. Toggle dark mode with the moon button in the top bar.
AI Foundations
Before we can judge AI, we need a clear picture of what it is — and what it is not.
What is artificial intelligence?
Artificial intelligence (AI) is the field of computer science that builds systems able to perform tasks that normally require human intelligence — such as recognising a face, understanding language, translating text, or making a recommendation.
Notice the wording carefully: AI does not need to be conscious or self-aware to be useful. Almost all AI you meet today is narrow AI — it is excellent at one specific task and knows nothing outside it. A chess engine that beats grandmasters cannot write your essay; a language model that writes your essay cannot drive a car.
🗣️ Say it clearly
"AI is a system that performs a specific task by finding patterns in data. It does not 'think' or 'want' anything." — a definition that will serve you well in any exam answer.
Three words people mix up
AI
The whole field: any technique that makes machines perform intelligent tasks, from simple rule-based systems to neural networks.
Machine Learning
A subset of AI. Instead of being programmed with rules, the system learns patterns from data.
Deep Learning
A subset of machine learning using neural networks with many layers. Powers image recognition, translation and today's chatbots.
AI ⊃ Machine Learning ⊃ Deep Learning ⊃ Generative AI
A short history
Alan Turing asks: "Can machines think?" and proposes a test based on conversation.
The Dartmouth workshop names the field "artificial intelligence".
IBM's chess computer defeats the world champion — using search, not learning.
Neural networks dramatically win the ImageNet image-recognition contest; the modern era begins.
An AI beats a top Go player, mastering a game long thought too intuitive for machines.
The Transformer architecture is published — the foundation of nearly every modern language model.
Chatbots and image generators become available to hundreds of millions of everyday users.
Activity: Is it AI? 是人工智能嗎？
Some everyday technologies use AI; others are simply rule-based or use basic sensors. Tap an item to select it, then tap the bin you think it belongs in. You get instant feedback on each choice.
✅ Uses AI
⚙️ Rule-based / sensor only
🇭🇰 Curriculum link
Computer Literacy (Junior Secondary): understanding the nature of information technology and its development. ICT (Senior Secondary): the role of information processing in solving problems. STEM: appreciating how technology evolves to meet human needs.
How AI Works
No magic, no mystery — just data, patterns, and a lot of arithmetic.
The learning loop
Almost every machine learning system follows the same four-step cycle. Understanding this loop is the single most useful thing you can take away from this module.
Thousands, millions or billions of examples — photos, sentences, sounds, clicks — are gathered into a dataset.
The model makes predictions, measures its errors, and adjusts its internal numbers (called parameters) to reduce those errors. Repeat millions of times.
Test the model on examples it has never seen. A model that only memorised its training data has overfitted and will fail in the real world.
Release the model, watch for errors, and retrain when the world changes. Models can drift out of date.
Neural networks in one paragraph
An artificial neural network is loosely inspired by the brain. It is arranged in layers of simple units. Each unit takes in numbers, multiplies them by weights, adds them up, and passes the result through a function that decides how strongly to fire. Early layers detect simple features (an edge, a letter shape); deeper layers combine them into complex ideas (a face, a sentence's meaning). Learning means slowly tuning millions or billions of weights.
🔢 Why "training" needs so much computing
A large language model may have hundreds of billions of parameters. Adjusting each one a tiny bit, across enormous datasets, is why training can cost millions of dollars and consume a great deal of electricity.
Three ways a machine learns
| Type | How it learns | Example |
|---|---|---|
| Supervised 監督式學習 | From labelled examples — every input comes with the correct answer. | Emails marked "spam" / "not spam"; photos labelled "cat" / "dog". |
| Unsupervised 非監督式學習 | From unlabelled data — the model finds its own groupings and structure. | Grouping customers into shopping-behaviour clusters. |
| Reinforcement 強化學習 | By trial and error, receiving rewards for good actions. | Game-playing AI; robots learning to walk. |
How generative AI writes a sentence
Text-generating AI is a next-token predictor. It breaks your prompt into pieces called tokens, then repeatedly answers one question: "Given everything so far, what is the most likely next token?" Pick one, append it, and repeat. That is the whole trick — repeated billions of times, at speed.
⚠️ The key consequence
Because it predicts what is plausible, not what is true, generative AI can state falsehoods with total confidence. This is called a hallucination. Fluent writing is not evidence of accuracy.
Why AI makes mistakes
📉 Bad or narrow data
If training data misses certain groups, the model performs badly for them.
🕰️ Outdated information
A model only knows what was in its training data, up to a cutoff date.
🎲 Probabilistic output
The same prompt can give different answers — there is no single "correct" stored answer.
🙈 No real understanding
Patterns are not comprehension. Models can miss meaning, sarcasm or context.
🇭🇰 Curriculum link
ICT: algorithms, data representation and the limits of computation. Mathematics: probability and statistics. STEM: applying computational thinking to understand real systems.
AI in Daily Life
You already use AI many times a day — usually without noticing.
Where you meet AI
🗺️ Navigation & transport
Route prediction, live traffic estimates, ride-hailing matching, and the MTR's service planning all rely on pattern analysis. Your phone predicts which route you want before you ask.
🎬 Recommendations
Streaming, music and shopping apps learn from your behaviour to suggest what keeps you watching or buying. This is powerful and designed to hold your attention.
💬 Language & translation
Instant translation, autocorrect, voice-to-text and chatbots. These use natural language processing (NLP) to model how language works.
🏥 Healthcare
Scanning X-rays and retinal images for early signs of disease, speeding up drug discovery, and triaging patients. AI assists clinicians — it does not replace them.
🎓 Education
Adaptive practice platforms adjust question difficulty, automated feedback helps with grammar, and teachers use analytics to spot students who need support.
🏦 Finance & security
Banks flag fraudulent transactions in milliseconds. The same techniques power spam filters and malware detection.
🌱 Environment
Forecasting typhoons, monitoring air quality, optimising electricity grids and tracking wildlife from camera footage.
🎨 Creative tools
Image generation, music composition, video editing and design assistance. These raise new questions about authorship and copyright.
AI in Hong Kong
Traffic-light optimisation, crowd-flow monitoring and predictive maintenance across public infrastructure.
Chatbots answering citizen queries, document processing, and data analytics supporting policy planning.
Schools adopting AI-assisted learning platforms and, increasingly, teaching AI literacy directly — which is exactly what you are doing now.
Universities and science parks developing AI for medical imaging, logistics, fintech and Chinese-language processing.
Who benefits — and who might not?
A balanced view matters. For each application, ask two questions: Who gains? and Who could be disadvantaged?
| Application | Benefit | Possible concern |
|---|---|---|
| Hiring tools | Faster screening of many applicants | May repeat historical bias against certain groups |
| Recommendation feeds | Content you actually enjoy | Echo chambers; reduced exposure to other views |
| Face recognition | Faster security checks | Mass surveillance; accuracy differences across skin tones |
| Automated grading | Immediate feedback | Struggles with creative or unusual answers |
💡 Try this
Over one day, keep a list of every AI feature you notice. Most students find 20+ interactions before dinner. Noticing is the first step to critical use.
🇭🇰 Curriculum link
Media and Information Literacy: understanding how technology shapes what we see and believe. Values Education: considering fairness and the common good. Life and Society / Citizenship: technology's impact on society and the economy.
Ethics & Safety
Powerful tools create real responsibilities. This module covers the risks you are most likely to meet.
1. Bias 偏見
AI learns from human data, and human data carries human prejudice. If a hiring model is trained on a company's past decisions, it may learn to favour the same kinds of candidates the company hired before. If facial recognition is trained mostly on one skin tone, it performs worse on others.
🚩 Remember
Bias in, bias out. AI does not remove discrimination — it can automate and scale it, while making it look objective because "the computer decided".
What to do: ask who is represented in the data, who is missing, and who bears the cost of a wrong answer.
2. Privacy & data 私隱
When you type into an AI chatbot, your words may be stored, reviewed by humans, or used to improve the model. Assume that anything you type could be retained.
⛔ Never share with an AI tool
- Your full name with your address or school, or your HKID number
- Passwords, one-time codes, or bank and payment details
- Other people's private information or photos without consent
- Confidential school documents, exam papers, or internal records
3. Deepfakes & misinformation 深偽技術與假資訊
Deepfakes are synthetic images, audio or video that convincingly imitate a real person. Combined with generative text, they make it cheap to produce convincing false content at scale.
🔎 How to check before you share
- Pause. Strong emotion — outrage or amazement — is exactly what misinformation is designed to trigger.
- Find the source. Who published it first? Is it a reputable outlet with a named author?
- Cross-check. Do at least two independent, trustworthy sources report the same thing?
- Look closely. Odd hands, mismatched audio, unnatural blinking, warped text in the background.
- Reverse-search the image, and check the date — old footage is often recycled.
- When unsure, don't share. Spreading doubt is itself a harm.
4. Over-reliance & skill loss 過度依賴
If AI writes every sentence and solves every problem, you do not build the thinking it is replacing. AI should be a scaffold for your learning, not a substitute for it. The students who gain most use AI to explain, quiz themselves, and get feedback — then do the work themselves.
5. Environmental cost 環境成本
Training and running large models consumes significant electricity and water for cooling. Responsible use includes not wasting compute on trivial tasks and being aware that "free" AI services have real-world costs.
Quick judgement check
Hallucination — it predicted a plausible title. Verify before citing.
Algorithmic bias — the training data encoded past discrimination.
Possible deepfake — check the source, the lipsync and independent reporting.
Privacy breach — you shared a third party's sensitive data without consent.
🇭🇰 Curriculum link
Values Education: integrity, respect for others, responsibility and care for the environment. Media and Information Literacy: evaluating information sources and recognising manipulation. Computer Literacy: safe, legal and ethical use of IT.
Responsible Use
How to work with AI honestly, effectively and with integrity.
The three-question test
Before using AI on any piece of schoolwork, ask yourself:
1. Is it allowed?
What does your teacher or school policy say? If unsure — ask first. Rules vary by subject and task.
2. Am I learning?
Does this help you understand, or does it replace the thinking you were meant to practise?
3. Am I honest?
Can you disclose exactly how you used it? If disclosure would be embarrassing, don't do it.
⚠️ Submitting AI text as your own is academic dishonesty
Presenting AI-generated work as your own original work — without permission or disclosure — is plagiarism. The consequence is a mark of zero, and it damages the trust your teachers place in you. It also robs you of the very skills you are at school to build.
Good uses vs. risky uses
| ✅ Usually responsible | ⚠️ Usually not responsible |
|---|---|
| Asking it to explain a concept you find difficult | Submitting its essay as your own |
| Generating practice questions to test yourself | Asking it to solve homework you then copy |
| Getting feedback on a draft you wrote | Asking it to write the draft |
| Brainstorming ideas you then research yourself | Trusting its facts without checking sources |
| Improving grammar in your own sentences | Using it to write in a style you can't reproduce in an exam |
Write better prompts
A vague prompt gets a vague answer. Strong prompts give role, task, context, format and constraints. Build one below and copy it out.
💡 Follow-up prompts that work well
- "Explain that more simply, as if to a younger student."
- "What are the strongest arguments against that answer?"
- "Which parts of your answer are uncertain, and why?"
- "Give me three practice questions on this, hardest last."
How to disclose your AI use
A short, honest statement is usually enough. For example:
"I used an AI chatbot to help me understand the water cycle and to generate practice questions. The explanation and all written work in this assignment are my own."
🇭🇰 Curriculum link
Values Education: honesty, self-discipline and responsibility for one's own learning. School-based academic honesty policy: follow your school's specific rules on AI-assisted work. Language subjects: authentic authorship and citation practice.
AI Literacy Quiz
Twelve questions covering every module. You get instant feedback and an explanation for each answer.
Ready when you are
Answer at your own pace. There is no time limit and you can retake the quiz as many times as you like — your best score is saved.
📋 What's covered
Definitions · how models learn · generative AI · bias · privacy · deepfakes · academic honesty · critical evaluation.
Flashcards
Twenty-two essential terms. Tap a card to flip it, then mark whether you knew it.
💡 How to revise effectively
Say the definition out loud before flipping. If you hesitate, mark it as not known and see it again sooner — this is called active recall, and it beats re-reading by a wide margin.
Glossary & Further Reading
Every key term in one place, plus where to go next.
Key terms
Further reading & trusted sources
Guidance on AI competency frameworks for students and teachers, and on human-centred AI in schools.
Information Technology Learning Targets, curriculum guides, and circulars on the use of AI in schools.
The Office of the Privacy Commissioner for Personal Data publishes guidance on AI and personal data protection in Hong Kong.
Fact-checking and source-verification guides for spotting misinformation and synthetic media.
The most reliable source for your specific course requirements — always check the syllabus.
⚠️ A note on sources
The field of AI changes quickly. Specific tools, laws and statistics mentioned in any resource may become outdated. Always check the date of a source and prefer official, named organisations over anonymous posts.
Keep learning
📝 Retake the quiz
Come back after a week. Spaced repetition is how knowledge sticks.
🃏 Drill the cards
Work through the deck until every term feels familiar.