Types of Prompts
Not all prompts are written the same way. Depending on the task and how much guidance is given to the AI, prompts can be categorized into different types. Understanding these types helps in choosing the right approach for any situation.
Each type of prompt has a specific purpose, and using the right type at the right time leads to better, more consistent results from AI models.
Overview of Prompt Types
Prompts are broadly classified based on two factors:
- How much information or examples are provided — ranging from none to several
- The nature of the task — whether it is instructional, conversational, creative, or analytical
Here are the main types of prompts covered in this course:
- Instructional Prompts
- Conversational Prompts
- Zero-Shot Prompts
- One-Shot Prompts
- Few-Shot Prompts
- Role-Based Prompts
- Chain-of-Thought Prompts
- Template Prompts
The last few types (Role-Based, Chain-of-Thought, Template) are explored in detail in the Intermediate section. This topic focuses on building a clear understanding of each type at a foundational level.
1. Instructional Prompts
An instructional prompt gives the AI a direct command or task. It is the most common type of prompt and is used when a specific action is needed.
Structure: [Action verb] + [Task] + [Optional: Format or Constraint]
Example:
"List the seven wonders of the ancient world."
"Summarize this article in two sentences."
"Translate the following text into Spanish."
Instructional prompts work best when the task is clearly defined and requires a straightforward output.
2. Conversational Prompts
A conversational prompt is phrased naturally, like something said in a normal conversation. It does not follow a rigid structure and is more open-ended.
Example:
"Hey, can you help me come up with a name for a bakery that sells traditional Indian sweets?"
"I'm trying to explain cloud storage to my grandfather. How would you put it in simple words?"
Conversational prompts are good for brainstorming, casual assistance, and exploratory tasks where an open-ended response is acceptable.
3. Zero-Shot Prompts
A zero-shot prompt gives no examples. It simply states the task and expects the AI to complete it based on what it already knows from training.
Example:
"Classify the following sentence as positive, negative, or neutral: 'The food was cold and tasteless.'"
There is no example of what a positive, negative, or neutral sentence looks like — the AI is expected to know. Zero-shot prompts work well for tasks the AI is familiar with, such as sentiment analysis, translation, and summarization.
4. One-Shot Prompts
A one-shot prompt provides one example before making the actual request. The example shows the AI the pattern or format expected in the response.
Example:
"Here is an example of a product review summary:
Review: 'The headphones have great sound but the battery dies quickly.'
Summary: Good sound quality, poor battery life.
Now summarize this review: 'The laptop is fast and lightweight, but the keyboard feels cheap.'"
With one example in hand, the AI follows the same format for the new review. One-shot prompts are useful when the output needs to follow a specific style or structure.
5. Few-Shot Prompts
A few-shot prompt provides two or more examples before the actual task. Multiple examples help the AI recognize a clearer pattern, especially for complex or unusual tasks.
Example:
"Convert each item below into a polite customer support reply:
Customer: 'Where is my order?'
Reply: 'Thank you for reaching out. We are checking on your order status and will update you within 24 hours.'
Customer: 'The product stopped working after two days.'
Reply: 'We sincerely apologize for the inconvenience. Please share your order number so we can arrange a replacement.'
Customer: 'I was charged twice for the same item.'"
The AI now has two examples to draw from and can generate a reply that matches the tone and structure shown.
6. Role-Based Prompts
A role-based prompt assigns a persona or role to the AI before making a request. This influences the tone, depth, and style of the response.
Example:
"Act as a financial advisor. Explain the difference between a savings account and a fixed deposit to someone who has never invested before."
By giving the AI a role, the response becomes more focused and appropriate for the context. Role-based prompts are explored in depth in the Intermediate section.
7. Chain-of-Thought Prompts
A Chain-of-Thought prompt asks the AI to think through a problem step by step before giving a final answer. This is particularly useful for complex reasoning tasks like math, logic, or multi-step analysis.
Example:
"A train travels 60 km in 1 hour. How far will it travel in 3.5 hours? Think through this step by step."
The instruction "think through this step by step" encourages the AI to show its reasoning, which leads to more accurate answers for complex problems. This type is covered in full detail in the Intermediate section.
8. Template Prompts
A template prompt uses a fixed structure with placeholders that can be filled in for different situations. This makes it easy to reuse the same prompt for similar tasks.
Example Template:
"Write a [type of content] about [topic] for [audience] in a [tone] tone. Keep it under [word count] words."
Filled Example:
"Write a blog introduction about electric vehicles for young adults in an engaging tone. Keep it under 100 words."
Template prompts are highly reusable and save time when the same type of content is needed repeatedly.
Choosing the Right Prompt Type
Here is a quick reference guide for selecting the right prompt type based on the task:
| Task Situation | Best Prompt Type |
|---|---|
| Quick, simple task (translation, summary) | Zero-Shot or Instructional |
| Casual brainstorming or exploration | Conversational |
| Output needs to follow a specific format | One-Shot or Few-Shot |
| Complex reasoning or multi-step problems | Chain-of-Thought |
| Expert tone or domain-specific response | Role-Based |
| Repetitive tasks with changing details | Template Prompt |
Key Takeaway
Prompts come in many forms. Instructional and conversational prompts are great for everyday tasks. Zero-shot, one-shot, and few-shot prompts differ in how many examples are provided. Role-based, chain-of-thought, and template prompts offer more control over the AI's output style and reasoning approach. Selecting the right type based on the task makes prompt writing faster and more effective.
In the next topic, we will look at Basic Prompt Structure — how to build a prompt from scratch using a reliable framework.