What is AI? How it works, what it can and cannot do
AI is already part of your everyday life, often without you noticing.
- Explain in plain words what AI is and where you meet it every day
- Describe how AI learns from examples instead of hand-written rules
- List what AI does well and what it cannot do
- Write a clear prompt with a task, context and an example
Read the lesson
AI is already part of your everyday life, often without you noticing. In this lesson you find out what AI really is, how it learns from examples, and what it can and cannot do. You also learn how to ask AI a good question.
01What is artificial intelligence?
AI is not magic. It is a computer system that uses the data it gets to produce an output.
- Predicts. Estimates what comes next: the next word you type, or next week's demand for a product.
- Recommends. Suggests options: films to watch, routes to take, replies to send.
- Generates. Creates new content, such as text, images or summaries, from a short prompt.
Plain-language version of OECD (2024), Updated OECD Definition of an AI System.
02AI is already part of your day
You already rely on AI, often without noticing. Here is where it quietly shows up.
- At home and on your phone. Voice assistants, predictive text, spam filters, map routes and bank fraud alerts are AI making fast guesses.
- When you shop or stream. Online shops and streaming apps match your past choices to new suggestions.
- At work. Email sorting, translation, meeting transcripts and customer chatbots handle routine tasks.
03How AI learns: from examples to answers
Modern AI is 'trained'. It studies examples until it can handle new cases on its own.
- Collect examples. Developers gather lots of data, such as text, images or numbers, that show the task the AI must do.
- Find the patterns. During training, the system adjusts itself again and again until it links inputs to the right outputs.
- Build the model. The result is a 'model': the trained system that has captured the patterns it found.
- Make predictions. You give the model something new. It uses the patterns to predict, recommend or create an answer.
Key idea: more good examples usually means a more reliable AI.
04The big idea: learning, not rules
Older software followed rules a person wrote. Modern AI works out the rules itself from data.
Old way: hand-written rules
- A person had to foresee every situation
- Each situation needed its own instruction
- It broke down on messy tasks like understanding language
New way: learning from data
- The system is shown thousands of examples
- It works out the patterns by itself
- This is why AI can now write, translate and answer questions
AI learns from human-made data, so it can copy human mistakes and bias. Always check its answers.
05Kinds of AI you will meet
Everyday AI
- Recommender AI. Suggests what you might like next, based on what you and others chose before.
- Predictive AI. Forecasts an outcome: demand next week, whether an email is spam, the fastest route.
- Generative AI. Creates new text, images or audio from a prompt. ChatGPT belongs here.
- Assistant AI. Combines the others to help you draft, summarise, answer and organise.
06Generative AI and chatbots
Tools like ChatGPT are 'generative'. Here is the simple idea behind them.
- They predict the next word. A chatbot writes by guessing the most likely next word again and again, learned from huge amounts of text.
- They sound fluent and human. They learned from human writing, so their answers read naturally, even when they are wrong.
- They do not truly understand. They match patterns in language. They do not 'know' facts the way a person does.
- Tools, not oracles. Used well, they save hours of drafting and explaining, as long as you check the result.
A chatbot predicts a likely answer. Likely is not the same as correct.
07What AI can and cannot do
AI is powerful, but it is a tool with real limits.
What AI does well
- Spot patterns in huge amounts of data in seconds
- Draft text, translate and summarise documents
- Answer common questions and automate repetitive tasks
- Work around the clock without getting tired
What AI cannot do
- Truly understand meaning, context or your intentions
- Guarantee its answers are correct
- Replace human judgement, ethics and empathy
- Know anything it was never trained on
Rule of thumb: AI drafts, a human decides. Always check important outputs.
08AI can be confidently wrong
AI can state false information as if it were true. This is called a 'hallucination'.
It predicts likely-sounding text and does not check facts. Verify facts, dates and numbers with a trusted source.
09How to write a good prompt
A 'prompt' is what you ask the AI. Better prompts get better answers.
- Be specific. Say what you want, for whom, and how long. 'Write a 5-line friendly reply to a customer asking for a refund' beats 'write a reply'.
- Give context. Add the background the AI needs: your role, the audience, the tone and any facts it must include.
- Show an example. Paste a sample of the style or format you want. Examples guide the AI better than adjectives.
- Refine step by step. Treat it as a conversation. Ask for changes like 'shorter' or 'more formal' until it fits.
If the answer misses, do not restart. Tell the AI what to fix.
10Smart habits when using AI
Use AI as a helpful assistant, not an authority you never question.
- Verify before you trust. Check facts, figures and names against a reliable source before you use them.
- Protect private information. Never paste passwords, customer data or confidential files into public AI tools.
- Keep your judgement. AI drafts and suggests. You make the final decision and you are responsible for it.
- Give it feedback. Tell the AI what to improve. Going back and forth is normal and gives a better result.
Free tools to try: ChatGPT, Google Gemini, Microsoft Copilot and DeepL Translate.
AaKey terms (6)
Artificial Intelligence (AI)
Software that learns patterns from data to predict, recommend or create, instead of following hand-written rules.
Model
The trained system that has 'learned' from examples and can now handle new inputs.
Generative AI
AI that creates new content, such as text, images or audio, from a prompt. ChatGPT is an example.
Prompt
The instruction or question you give an AI. Clear prompts get better answers.
Hallucination
When an AI states something false as if it were true, because it predicts likely text, not checked facts.
Training data
The examples an AI learns from. Their quality shapes how good and how fair the AI is.
Practise
?Check your understanding
Answer these in your own words — out loud, in a notebook, or with a study partner.
- In one sentence, how would you explain what AI is to a friend who has never heard of it?
- Why can an AI chatbot sound completely confident and still be wrong?
- Name two tasks AI does well and one important thing it cannot do.
- What is one habit you will use every time you rely on an AI answer?
Key takeaways
- AI is software that learns patterns from examples. It is a tool, not magic.
- It predicts, recommends and generates, but it does not truly understand.
- AI can be fluent and confident yet wrong, so always check important outputs.
- Clear, specific prompts with context and examples get far better results.
- You stay in charge: AI drafts and suggests, a human decides and is responsible.
Your next steps
- 1
Lesson quiz
5 questions, 70% to pass, retake as often as you like.
Create a free account to take the quiz - 2
Hands-on exercise EX 1.1
Find five ways AI is already part of your daily life or your work sector, and explain one that surprised you. This shows you that AI is not far away: it is already around you.
Open the exercise →