Crafting Your First Prompts
After understanding AI theory in the previous article, now it’s time to get hands-on. In this article, I’ll guide you through designing basic prompt structures to achieve desired results from AI.
You may have already tried asking AI a few times, but did you know that how you structure your prompts directly affects the quality of responses? A well-designed prompt helps AI understand more clearly, answer more accurately, and saves you time.
1. Basic Prompt Structures
A well-structured prompt significantly enhances response quality by ensuring clarity, specificity, and context. Below are 5 basic structures you should know:
Instruction-Based Prompts
This is the simplest and most direct type of prompt. You clearly state what AI needs to do without too much context.
Example: “Summarize the main themes of the novel Pride and Prejudice.”
This type of prompt is suitable when you want to get straight to the point and don’t need AI to infer further.
Open-Ended Prompts
With open-ended prompts, you allow AI to respond creatively and exploratively. This type is suitable when you want to receive multiple perspectives or new ideas.
Example: “Describe a futuristic city where humans and AI live in harmony.”
Open-ended prompts often yield more diverse and surprising results, but are also harder to control.
Role-Based Prompts
Asking AI to play the role of an expert or specific individual. This is a very effective way to adjust the tone and depth of responses.
Example: “You are a financial analyst. Explain the impact of inflation on the stock market.”
AI will respond with a more professional tone, using terminology appropriate to the assigned role.
Context-Providing Prompts
Include background information so AI better understands the situation and increases response accuracy.
Example: “In the context of space exploration, explain the challenges of humans going to Mars.”
Providing context helps AI filter out irrelevant details and focus on the main issue.
Step-by-Step Prompts
Guide AI through multi-step reasoning. This type of prompt is very useful when you need AI to explain a process or analyze in detail.
Example: “Explain the photosynthesis process in 5 detailed steps.”
This approach helps AI not skip important stages and makes the answer easier to follow.
2. Experimenting with Simple Queries
Creating effective prompts isn’t an overnight thing. You need to continuously experiment and refine. Below are some methods to help you improve results:
Rephrase the Prompt
Change the wording to receive different responses. Sometimes just modifying a few words significantly changes the results.
Example:
- “Explain what AI is.” → Answer might be somewhat generic
- “Explain what AI is to a person without technical knowledge.” → More specific and understandable answer
Add Context
Provide background information so AI filters out irrelevant details.
Example:
- “Write about AI history.” → Too broad
- “Write about AI history from 1950 to present, focusing on important milestones.” → Clear and focused
Use Constraints
Set limits on word count, style, or format.
Example: “Write under 150 words about AI’s impact on education.”
Constraints help AI focus on the most important content without rambling.
Request Multiple Perspectives
Encourage AI to consider different viewpoints to enrich the response.
Example: “Analyze the advantages and disadvantages of using solar energy.”
3. Do’s and Don’ts
Below are golden rules to help you optimize interactions with AI:
✅ Do’s
Be Specific The more detailed, the better the results. Instead of asking generally, clearly state the information you need.
Provide Context Helps answers be more accurate and appropriate. AI can’t read your mind.
Use Step-by-Step Instructions Improves clarity for complex tasks. For example: “Step 1, analyze the problem. Step 2, provide solutions.”
Use Examples Provide samples so AI understands the desired format or style.
Example: “Write a professional email like this: [email template]. Write a similar email to invite customers to a seminar.”
❌ Don’ts
Avoid Vague Prompts Generic queries often lead to superficial answers.
Example of vague prompt: “Tell me about technology.” → Too broad, AI doesn’t know what you want.
Don’t Assume AI Understands Implicit Context Must clearly state necessary information. AI doesn’t have human-like reasoning ability.
Avoid Overly Long and Complex Prompts Too much information can confuse the model. Break prompts into smaller parts if needed.
Don’t Expect Perfection on the First Try Need to iterate and refine to achieve desired results. Prompt Engineering is a process, not a result.
Conclusion
Prompt Engineering is a controllable skill through applying existing structures and continuously refining through trial and error.
You don’t need to be perfect from the start. What matters is that you understand each structure type, know when to use which, and are willing to experiment.
Start with simple prompts, observe results, and gradually refine until you feel more confident.
Illustrative Example
Imagine you’re asking an assistant to write a report.
If you say: “Write a report”, the result might not match your intention.
But if you say clearly: “Write a 500-word report on Q3 marketing effectiveness, focusing on social media, using professional tone”, you’ll receive exactly what you need.
AI is the same. The clearer you are, the closer the results match expectations.