Good Instructions, Good Results - The Easy Way to Better ChatGPT Answers

Many people try ChatGPT and are often disappointed. The answers seem superficial or inappropriate. The problem often lies not only with the AI, but mostly with the instruction: Those who ask vaguely get vague answers. Good results are not a matter of luck, but of clarity.

In this part of the AI Academy, we show how to control ChatGPT using simple, precise prompts. You will learn how the model interprets inputs, what distinguishes good from bad instructions, and how to achieve better results step by step.

How ChatGPT Thinks

ChatGPT is not a lexicon. The model analyses each input, recognises patterns and statistically calculates which word is likely to fit next. When a request is unclear, the AI has to guess and quickly produces general or unhelpful texts. It cannot read minds and works solely with what is stated in the prompt.
Despite the large amount of data in training, only one thing counts at the moment of inquiry: the information you give the AI. The more precise this is, the more accurately the model can capture your intention.

A good instruction works like a briefing. Clear doesn’t necessarily mean short, because specific details help the AI respond more accurately. The principle is: The input controls the output.

Poor Practices – Avoid Common Mistakes

Many mistakes are repeated in practice, the indications are far too vague:

• Too General: “Tell me something about technology in real estate.”

• Ambiguous: “Summarise that better.” (What does “better” mean?)

• No Goal: “Explain that.”

• Multiple Tasks at Once: “Make a list, write an introduction and explain the rest to me.”

• No Context: “Create a nice letter for the owners.”

• Unclear Tone: “Make it sound nicer.” (Friendly? Casual? Formal?)

These formulations make the AI guess, and this often goes wrong.

Good Instructions – How It Works Better

A “prompt” is the instruction you use to give ChatGPT a task. It can be a question, a command, or a brief description of what the AI should generate. No one gets the perfect wording on the first try. It is normal to start with a rough idea, check the answer and then precisely refine missing details. ChatGPT is designed for such step-by-step work.

Strong prompts follow a clear scheme. The main components are:

1. Role: Give the AI an identity at the start, e.g. “You are an experienced manager...”

2. Goal: What should the AI do, like explain, structure or rephrase.

3. Format: List, Email, short note, Python code, explanation in simple words.

4. Tone: Factual, friendly, polite, neutral or casual etc.

5. Examples: Templates for responses or short text snippets help the model to match style and structure.

For more complex tasks, it is worth asking the AI for a transparent derivation. An instruction like “First explain your thought process, then provide the solution” ensures clearer and more understandable results.

Work in several steps: Check answers, refine, add examples or have the text rephrased. For example: “Please shorten this.” or “Add an example for each point.” ChatGPT can also critically assess and improve its own text on request.

If your question is based on certain information, provide the relevant context directly, for example by including a short text section or writing “Use the following data as a basis.” This reduces misunderstandings and ensures precise answers.

Before – After: Examples from Everyday

The 4-Point Rule: Simple Structure for Clear Prompts

Almost every good instruction can be divided into four components:

1. What should be done? - write, explain, list, shorten...

2. What is it about? - Topic, occasion, reference

3. How should it sound? - Tone, format, audience

4. What helps the AI? - Role, examples, constraints (e.g. length)

Example: “You are an experienced caretaker. Explain to an owner how to ventilate properly to prevent mould growth. Use clear language and examples, maximum 150 words.”

The GUIDE Method

As an addition to the Four Point Rule, GUIDE can help ensure nothing important is forgotten. It serves as a short checklist:

G = Goal: Aim of the request
U = User: Role or audience
I = Instructions: Specific task
D = Details: Tone, length or specifics
E = Examples: an example or short excerpt

A quick look at these five points ensures that the request is complete, and the AI receives all the necessary clues.

Basic Tools: Ready-to-Use Instructions

1. Summarize: “Summarize this text in 3 bullet points: [Insert text].”

2. Rephrase style: “Rephrase the following message factually, but friendly: [Insert text].”

3. Explain complex things: “Explain this topic in simple words so that a 10-year-old can understand it and with a suitable everyday example: [Topic].”

4. Develop ideas: “Name five concrete suggestions on how you can think creatively about this topic: [Topic].”

5. Shorten text: “Shorten the following text to under 100 words without losing important information: [Insert text].”

6. Create image: “Create an image with the following motif and style specifications: [Motif, style, details].”

7. Translate: “Translate the following paragraph into English, clearly and professionally: [Text].”

8. Create comparison: “Create a clear table of pros and cons on the following topic: [Topic].”

9. Draft example: “Create a short sample text with three to four sentences for the following occasion: [Occasion].”

10. Request analysis: “Analyse this section and list the main statements: [Text].”

Conclusion

Good instructions create clarity and bring out the best in the AI. Those who specify the target, tone, and format receive responses that are more precise, understandable, and truly helpful in everyday life. In the end, it’s clear: The better your instruction, the better the AI’s response. With a bit of practice, you will notice how significantly the quality of results improves and how naturally ChatGPT becomes an effective tool.