13/08/2026
Quick start: What is ChatGPT?
ChatGPT is an AI assistant that can understand instructions and generate useful responses in natural language.
You can use it for:
βοΈ Writing, rewriting, summarizing, and translating
π» Coding and troubleshooting
π Learning and teaching
π§ Brainstorming ideas
π Analyzing information and documents
π Researching current information when web access is available
π¨ Creating image concepts and visual materials
π Planning projects, courses, websites, events, and workflows
How to interact: Tell it what you want, provide relevant context, specify the desired format, then review and refine the result.
2) Benefits and limitations
Where ChatGPT excels
Quickly generating and organizing ideas.
Explaining complicated topics in simple language.
Creating first drafts and structured plans.
Adapting content to different audiences and languages.
Helping debug code and identify likely problems.
Iteratively improving an answer based on your feedback.
Where it can struggle
It can make factual mistakes or confidently invent information.
Current information may require web research.
Specialized legal, medical, financial, or technical claims should be independently verified.
Ambiguous prompts can produce generic answers.
It may misunderstand an unusual name, language, font, technical configuration, or local context without enough information.
Rule of thumb: Use ChatGPT to accelerate your thinkingβnot as an automatic substitute for verification.
3) Why prompts matter
A prompt is essentially your instruction to the AI. More useful instructions generally produce more useful results.
A strong prompt usually contains
Role + Task + Context + Requirements + Output format
For example:
Poor:
Make a website.
Better:
Act as a senior UI/UX and full-stack developer. Design a professional responsive website for a computer training center. Include Home, About, Courses, Registration, Certificate Verification, and Contact pages. Use a modern educational style, mobile-first layout, accessible typography, smooth animations, and clear calls to action. Provide the page structure, features, database requirements, and implementation plan.
Useful prompt techniques
Be specific: Say exactly what you need.
Give context: Explain who will use the result and why.
Set constraints: Budget, length, technology, audience, language, etc.
Specify format: Table, bullets, step-by-step guide, code, presentation outline, etc.
Provide examples: Show the style or result you're aiming for.
Ask for iteration: βGive me version 1, then identify weaknesses and improve it.β
Define success: Explain what a good result should accomplish.
4) Best practices for reliable results
Clarity
Instead of:
Tell me about Excel.
Try:
Explain Excel's Filter feature to a beginner. Use simple language, a practical example, and 5 practice questions.
Context
Include information that changes the answer:
Your skill level
Target audience
Software/version
Device/operating system
Country or organization, when relevant
Existing work or constraints
Iteration
Don't expect the perfect answer immediately.
A useful sequence is:
Initial request β Review β Give feedback β Refine β Verify
For example:
Make it more professional.
Then:
Keep the same structure, but make the language suitable for government/NGO documentation and reduce unnecessary wording.
Verification
For important information:
Ask for sources.
Request the model to distinguish facts, assumptions, and recommendations.
Verify critical claims independently.
For current information, ask ChatGPT to search the web.
Safety
Don't unnecessarily provide:
Passwords
Private keys
Financial credentials
Sensitive personal information
Confidential organizational information
5) Simple real-world workflow
Example: Planning a computer training project
Step 1 β Initial prompt
I want to establish a computer training program for beginners. Help me create a project plan.
Step 2 β Add context
The learners are 15β25 years old. The course will run for 3 months. Classes will be 5 days per week. We have 20 computers and one instructor. The main subjects are computer fundamentals, Windows, MS Word, Excel, PowerPoint, internet, and basic AI.
Step 3 β Request structure
Create a weekly curriculum with learning objectives, practical exercises, assessments, and required equipment.
Step 4 β Improve
Review this curriculum as an experienced training-center manager. Identify weaknesses, missing topics, unrealistic timing, and accessibility problems.
Step 5 β Finalize
Rewrite the curriculum incorporating your improvements. Make it suitable for presenting to an NGO donor.
This approach is much better than trying to write one enormous prompt immediately.
6) Common pitfalls
Pitfall Better approach
βMake it professionalβ Define what professional means
Too little context Explain the audience and purpose
Huge, confusing request Break it into stages
Assuming everything is correct Verify important facts
Asking vague questions Give measurable requirements
Not mentioning software version Include exact version
Accepting the first answer Critique and refine it
Asking for current facts without research Request web research
Giving contradictory instructions Put requirements in priority order
Sharing sensitive information Remove or anonymize it
7) Advanced prompting techniques
Role instructions
You can establish expertise:
Act as a senior web architect and accessibility specialist.
This helps establish the perspective and standards you want.
Structured instructions
A powerful template is:
Role: Senior project manager
Task: Create a project plan
Context: [background]
Audience: [users]
Requirements: [requirements]
Constraints: [limitations]
Output: [desired format]
Quality criteria: [what success looks like]
Ask for evaluation
Instead of simply asking for an answer:
Produce a solution, then check it against these requirements and identify anything missing.
Ask for alternatives
Give me three approaches. Compare their advantages, disadvantages, cost, complexity, and recommended use case.
Ask for assumptions
Before answering, identify any important assumptions you're making. If an assumption could significantly change the answer, ask me about it.
Chain-of-thought considerations
You generally don't need to ask ChatGPT to reveal its private chain-of-thought. A better approach is to request a useful, concise explanation:
Give me the conclusion, key reasoning steps, assumptions, and important checksβwithout exposing private internal reasoning.
This gives you transparency without requiring hidden internal reasoning.
β A reusable βhigh-quality promptβ template
Act as: [expert/role]
Goal: [what I want to accomplish]
Context: [important background]
Audience: [who will use/read it]
Requirements:
[requirement]
[requirement]
[requirement]
Constraints: [budget, technology, length, deadline, etc.]
Output format: [steps/table/report/code/etc.]
Quality standards: [professional, simple, accessible, accurate, etc.]
Before finalizing: Check your response against every requirement and point out anything uncertain.
The biggest improvement you can make is simple: don't just tell ChatGPT what topic you're interested inβtell it what outcome you want, who it's for, what constraints exist, and what the finished result should look like.
