Prompt Words: Context, Persona, Constraint, Few-Shot and More

You open an AI course, and in the first five minutes you meet words like "context", "persona", "few-shot" and "constraint". Nobody stops to explain them. You feel lost, even though the ideas behind them are simple.

This course fixes that. Each word gets a plain meaning, an everyday example and a note on where you will see it. You do not need to remember everything. Come back to this course whenever a word stops you.

This first lesson covers the words people use when they talk about prompts: the messages you type to an AI chatbot like ChatGPT, Gemini or Claude.

What You'll Learn

• What a prompt is, and the words used to describe a good one
• What context, persona and constraint mean
• The difference between zero-shot and few-shot
• What task decomposition and iteration mean
• Why "vague" and "hallucination" come up so often

Prompt

Meaning: the message you type to an AI. A question, an instruction or a request.

Example: "Write a two-line birthday message for my sister who loves cricket."

You'll see it: everywhere. "Prompt engineering" just means getting better at writing prompts. If you want to go deeper later, see Prompt Engineering.

Context

Meaning: the background information you give the AI so it understands your situation.

Example: "Write an email asking for leave" is a prompt with no context. "I am a first-year nursing student. Write a short, polite email to my professor asking for two days of leave for my cousin's wedding" has context: who you are, who it is for and why.

You'll see it: in almost every prompt lesson. More context usually means a better answer.

Persona (or role)

Meaning: telling the AI to act as a certain kind of person.

Example: "Act as a friendly maths teacher" or "You are a career coach for engineering students."

You'll see it: in prompt templates. A persona changes the tone and the level of detail. It does not give the AI real expertise, so you still need to check facts.

Task, format and tone

These three often appear together in prompt "formulas".

• Task: what you want done. "Summarize this article."
• Format: the shape of the answer. "As 5 bullet points" or "as a table with two columns."
• Tone: the feeling or style. "Friendly", "formal", "simple enough for a 12-year-old."

If a course says "add the format part", it means: tell the AI how the answer should look.

Constraint (and negative constraint)

Meaning: a limit or rule the answer must follow.

Example: "Under 100 words." "Use simple English." "Only use the facts I gave you."

A negative constraint is a rule about what not to do: "Do not use technical words." "Do not include prices."

You'll see it: in lessons on improving prompts. Constraints stop the AI from rambling or guessing.

Vague vs specific

Meaning: a vague prompt leaves out key details, so the AI has to guess. A specific prompt says exactly what you want.

Example: "Tell me about marketing" is vague. "Give me 3 cheap ways a small bakery in my town can get more customers on Instagram" is specific.

You'll see it: when a course explains why an answer was "generic". A generic answer is one that could fit anyone, because the prompt was vague.

Zero-shot and few-shot

These sound technical but are simple.

• Zero-shot: you ask without giving any examples. "Write a product description for a steel water bottle."
• One-shot: you give one example of what you want first.
• Few-shot: you give a few examples first, so the AI copies the pattern.

Example of few-shot:

Turn these into catchy one-line slogans.
Product: umbrella. Slogan: Rain, meet your match.
Product: running shoes. Slogan: Your morning, faster.
Product: steel water bottle. Slogan:

The AI sees the pattern and writes a slogan in the same style.

You'll see it: in prompt engineering courses. "Shot" here just means "example".

Task decomposition

Meaning: breaking a big task into smaller steps and asking for one step at a time.

Example: instead of "Write my whole project report", you ask for an outline first, then the introduction, then each section.

You'll see it: in lessons on hard tasks. Smaller steps give better results and are easier to check. A related idea is prompt chaining: using the answer from one prompt as the input to the next.

Iteration

Meaning: improving something in rounds. You try, look at the result, adjust and try again.

Example: your first prompt gives a long, boring answer. You reply "Make it shorter and add one example." That second message is an iteration.

You'll see it: a lot. Good AI use is almost always iterative. The first answer is a draft, not the final result.

Hallucination

Meaning: when an AI says something false but sounds confident, like a made-up fact, a fake quote or a book that does not exist.

Example: you ask for sources and the AI lists a research paper with a real-sounding title that was never written.

You'll see it: in every responsible AI lesson. It is the main reason you should check important facts. AI Essentials explains why it happens.

Citation

Meaning: a reference to where a piece of information came from, like a link, a book or a paper. To cite something means to give its source.

Example: "According to the World Health Organization (link), adults should..."

You'll see it: in research and study lessons. Some AI tools, like Perplexity or ChatGPT search, show citations. Always click one or two to check they say what the AI claims.

A quick reference

Key Takeaways

• A prompt is your message to the AI; context, persona, task, format, tone and constraints are the parts of a good one.
• Zero-shot means no examples, few-shot means a few examples first.
• Task decomposition breaks big jobs into small steps, and iteration improves the answer in rounds.
• A vague prompt gives a generic answer; a specific prompt gives a useful one.
• Hallucinations are confident mistakes, so check important facts and citations.