Home » AI Prompts Tutorial

Prompt Engineering Basics

Prompt engineering is the process of designing, improving, and refining prompts to communicate effectively with AI models and obtain useful results.

In this chapter, you will learn the basic concepts of prompt engineering, the main elements of a prompt, common techniques, and how to improve prompts through testing and refinement.

What Is Prompt Engineering?

Prompt engineering is the practice of creating and optimizing instructions given to an AI model.

A simple prompt may be:

Write about Python.

A more structured prompt could be:

Write a beginner-friendly introduction to Python programming.
Explain what Python is, its main features, and common uses.
Use simple English and include three examples.

The second prompt provides more information about the expected response.

Prompt engineering does not mean using complicated words. The goal is to provide the AI model with clear and useful instructions.

Why Is Prompt Engineering Important?

AI models can generate different responses based on how a task is described. A well-designed prompt can help communicate the user's requirements more clearly.

Prompt engineering is useful for:

• Content creation
• Software development
• Data analysis
• Education
• Research
• Marketing
• Customer support
• Business tasks
• Image generation
• Automation

For example, a vague prompt may produce a general response:

Explain marketing.

A more specific prompt can produce a response focused on a particular requirement:

Explain digital marketing to beginners.
Describe SEO, social media marketing, email marketing, and content marketing.
Use simple English and provide one example for each.

Basic Elements of Prompt Engineering

A prompt can contain several elements. The exact structure depends on the task.

Common elements include:

Role

A role tells the AI model what perspective or type of expertise it should use.

For example:

Act as a Python instructor and explain Python functions to beginners.

Context

Context provides background information needed to understand the task.

For example:

The audience consists of college students who have basic programming knowledge.
Explain object-oriented programming in simple English.

Instruction

The instruction describes the task the AI model should perform.

For example:

Summarize the following article in 150 words.

Input

Input is the information that the AI model needs to process.

For example:

Summarize the following text:

Artificial intelligence is being used in many industries...

Constraints

Constraints define rules or limitations.

For example:

Explain cloud computing in 300 words.
Use simple English and avoid technical jargon.

Output Format

The output format specifies how the response should be presented.

For example:

List five advantages of Python in a numbered list.

Basic Prompt Engineering Techniques

Several simple techniques can help improve AI prompts.

Be Clear and Specific

Avoid vague instructions and clearly describe the required task.

Instead of:

Write about databases.

Use:

Explain relational databases to beginners.
Describe tables, rows, columns, primary keys, and foreign keys.
Include one simple SQL example.

Provide Context

Give the AI model relevant background information.

For example:

I am creating a programming tutorial for beginners.
Explain recursion in Python using simple English.

The context helps the model understand the intended audience and purpose.

Give Examples

Examples can show the AI model what type of response you expect.

For example:

Convert statements into questions.

Example:
Statement: Python is easy to learn.
Question: Is Python easy to learn?

Now convert:
Statement: Java is object-oriented.

Specify the Output

Tell the AI model how you want the result formatted.

For example:

Compare Python and Java.
Present the comparison in a table with the following columns:
Feature, Python, Java.

Set Constraints

Constraints can control the length, style, or content of the response.

For example:

Write a 500-word article about cybersecurity.
Use simple English.
Include five common cybersecurity threats.

Zero-Shot Prompting

Zero-shot prompting means asking an AI model to perform a task without providing examples.

For example:

Classify the following review as Positive, Negative, or Neutral:

"The product arrived on time and works perfectly."

The model must understand the task based on the instruction alone.

Zero-shot prompting works well for many straightforward tasks.

One-Shot Prompting

One-shot prompting provides one example before the actual task.

For example:

Convert a formal sentence into an informal sentence.

Example:
Formal: I would appreciate your assistance.
Informal: Can you help me?

Now convert:
Formal: I would like to request additional information.

The example provides a pattern that the model can follow.

Few-Shot Prompting

Few-shot prompting provides multiple examples before asking the AI model to complete a task.

For example:

Classify the sentences as Formal or Informal.

Example 1:
"Please provide the requested document."
Classification: Formal

Example 2:
"Can you send me the file?"
Classification: Informal

Now classify:
"Could you please share the report?"

Multiple examples can help the model understand a specific pattern or classification rule.

Prompt Chaining

Prompt chaining involves dividing a complex task into multiple prompts instead of asking the AI model to complete everything in one prompt.

For example, a content creation workflow could use:

Prompt 1:
Create an outline for an article about cloud computing.

Then:

Prompt 2:
Write the introduction based on the outline.

Then:

Prompt 3:
Review the article for clarity and suggest improvements.

Breaking a complex task into smaller steps can make the workflow easier to manage.

Iterative Prompting

Prompt engineering often involves testing a prompt, reviewing the response, and improving the prompt.

For example:

Write an article about cybersecurity.

If the response is too general, the prompt can be improved:

Write a beginner-friendly article about cybersecurity.
Explain five common cybersecurity threats and provide prevention techniques.
Use simple English and include practical examples.

The process can be repeated until the output meets the requirements.

Negative Instructions

Negative instructions tell the AI model what should be avoided.

For example:

Explain artificial intelligence to beginners.
Use simple English.
Do not use complex mathematical terminology.

Negative instructions can be useful when certain content, formats, or styles are not wanted.

Structured Prompting

Structured prompting organizes the instructions into clearly separated sections.

For example:

Role:
Act as a technical writer.

Audience:
Beginner programmers.

Task:
Explain REST APIs.

Requirements:
Explain REST, HTTP methods, and status codes.
Include one practical example.

Output:
Use headings and short paragraphs.

This structure makes complex requirements easier to understand and modify.

Prompt Testing and Refinement

A prompt should be tested against the expected result.

When reviewing the output, consider:

• Did the AI complete the requested task?
• Did it follow the required format?
• Did it use the correct tone?
• Did it include the required information?
• Did it follow the constraints?
• Did it include unnecessary information?
• Is the information accurate?

Based on the results, modify the prompt and test it again.

Common Prompt Engineering Mistakes

Beginners commonly make the following mistakes:

• Using vague instructions
• Providing insufficient context
• Giving conflicting requirements
• Using unnecessary information
• Not specifying the target audience
• Not defining the expected output
• Giving too many instructions at once
• Assuming the AI model understands unstated requirements
• Not testing and refining prompts
• Trusting AI-generated information without verification

Simple Prompt Engineering Workflow

A basic prompt engineering process can be summarized as follows:

Define the Goal
      |
      v
Write the Initial Prompt
      |
      v
Generate the Response
      |
      v
Review the Output
      |
      v
Identify Problems
      |
      v
Improve the Prompt
      |
      v
Test Again

This process can be repeated until the prompt produces a useful and reliable result.

Example of Basic Prompt Engineering

Consider the following initial prompt:

Write about artificial intelligence.

This prompt does not specify the audience, length, topics, or format.

An improved version could be:

Role:
Act as an AI educator.

Task:
Write a beginner-friendly introduction to artificial intelligence.

Requirements:
Explain what AI is, how it works, and three common applications.
Use simple English and include practical examples.

Output:
Use headings and short paragraphs.
Keep the article under 800 words.

The improved prompt provides the AI model with a clear role, task, requirements, and output format.

Prompt Engineering Best Practices

When designing prompts, keep the following practices in mind:

• Clearly define the goal.
• Use specific instructions.
• Provide relevant context.
• Identify the target audience.
• Give examples when useful.
• Define the expected output format.
• Add important constraints.
• Break complex tasks into smaller steps.
• Test prompts with different inputs.
• Review and refine the generated output.
• Verify important AI-generated information.

Prompt engineering is an iterative skill. You do not always need to create the perfect prompt on the first attempt. Testing and improving the prompt based on the generated output is an important part of the process.