Prompting Course: Basics · 2 min read
What is prompting and why does it matter?
Prompt basics, how language models interpret text, and why clarity determines quality.
Section 1: What is prompting and why does it matter?
Section titled “Section 1: What is prompting and why does it matter?”Learning objectives
- Understand what prompting is and how models interpret instructions
- Recognize why the same question can give entirely different results
- Know why prompting is an important skill
What is a prompt?
A prompt is the instruction you give a language model – a question, a task, or a description of what you want.
A well-formed prompt is like a compass: it points the model toward the right result. Without a compass, the model guesses direction.
A prompt can be a simple question, a long system instruction – or a full block of context and rules.
How does a language model "understand"?
Language models are pattern recognizers, not mind readers.
- They’ve seen enormous amounts of text and learn probable continuations – they don’t guess what you mean between the lines.
- Ambiguities are filled with what’s statistically likely, not necessarily what you needed.
Example: “Help me with this text” – should the model summarize, proofread, translate, shorten, or rephrase? Without precision the answer becomes a guess.
Why does prompting matter?
Without a clear instruction you risk vague, wrong, or off-tone responses.
Good prompting gives you:
- More relevant and useful answers
- Less back-and-forth and rephrasing
- Better control over format and tone
Same question – different results
AI can’t read minds – it follows what you write.
Poor: “Do something with this report.”
Better: “Summarize the report in exactly three bullet points. Focus only on financial KPIs (e.g. ROI, revenue growth) and include one recommendation for next quarter. Use a professional tone.”
Precision in = quality out. Tip: show the prompt to a colleague without context – do they understand the task?
Concept clarification: system prompt vs user message
System prompt (or the equivalent in assistant settings) sets the hard boundaries: role, language, what is and isn’t allowed. In Intric this is where you typically build and fine-tune as an assistant creator.
User message is what the end user types in chat. It is not the same as the full system prompt – it operates within the boundaries set above.
The system prompt is hierarchically superior: the user can ask for shorter answers or a different tone in chat, but shouldn’t be able to break the fundamental rules you’ve set.
Summary
Section titled “Summary”- The prompt is your compass – task, context, and expected result.
- Clarity determines quality – the model interprets literally what you give it.
- If a person without context doesn’t understand the instruction, the model usually won’t either.
Test your knowledge
Question 1 of 3
What is prompting?