AI Writing, CHATGPT, Checklist, How to Write

How to Write Iterative Chain-of-Thought Prompts (with examples)

Posted on by Alan Thompson
27
Nov

In the previous article we looked at zero-shot prompts.

The key difference between iterative chain-of-thought prompts and zero-shot prompts lies in how they handle refinement and complexity.

Zero-Shot Prompts provide a quick solution without any examples or prior context. Basically, you give one prompt, and ChatGPT produces a response based solely on its training.

In contrast, Iterative Chain-of-Thought Prompts allow you to develop, refine, or analyze ideas by building on previous outputs step by step. For instance, you ask ChatGPT to generate an initial response, then iteratively improve or expand it by providing follow-up prompts. Say you’ve drafted an important report, but it feels a little off. Maybe it’s too dry. You know what you want to say, but you’re stuck finding the right words. That’s where iterative chain-of-thought prompts is most useful.

I once used this approach to refine an email pitch for a potential client. The first draft was ok, but it lacked personality. With ChatGPT as my ‘collaborative’ co-author, I started with:

“Summarize this draft. Now make it sound more engaging.”

A few rounds (i.e. iterations) of back-and-forth later, the email became more polished, more persuasive in the use of language, and ready to send out.

What Are Iterative Chain-of-Thought Prompts?

Let’s take a step back and define this prompt type.

Definition: An Iterative Chain-of-Thought Prompts allow guides ChatGPT through a step-by-step reasoning process to refine ideas or solve complex problems. By improving responses over multiple iterations, they help you achieve deeper insights or create more refined results.

It’s a bit like an open-ended brainstorming session with a super smart friend who always has fresh perspectives. Each step builds on the last, gradually honing the output into something more interesting.

Why Use Iterative Chain-of-Thought Prompts?

It’s all about refinement.

Using this approach, you can refine ideas, drafts, and plans. Instead of settling for a single answer, you can explore, tweak, and sharpen until you achieve the results you want.

Here’s why it works:

  1. Improves Clarity: Breaks down complex thoughts into manageable steps.
  2. Enhances Quality: Fine-tunes responses with each iteration.
  3. Encourages Collaboration: Simulates brainstorming with a patient partner.

In the last section, we discussed how zero-shot prompts deliver quick solutions. Now, let’s explore how iterative chain-of-thought prompts can help refine those solutions further.

Real-Life Win with Iterative Chain-of-Thought Prompts

Iterative prompts aren’t just for editing text—they’re for refining and building your ideas.

Sample Prompt Sequence:

  1. “Explain why remote work is beneficial using three points.”
  2. “Now make the tone more persuasive and include examples.”
  3. “Summarize this in one sentence for a tweet.”

Each step added value, transforming a basic explanation into a compelling, shareable message that resonated with my audience.

How to Use Iterative Chain-of-Thought Prompts

This method is perfect for projects that require more depth and structure as opposed to one shot question and answers.

Let’s look at some use cases. Whether you’re crafting arguments, editing, or brainstorming, here’s how to use it:

  1. Start Broad
    • Prompt: “List three ways social media effects local businesses in North Carolina.”
    • Use: Gather ideas or perspectives.
  2. Add Specifics
    • Prompt: “Expand on these ideas with examples from different industries.”
    • Use: Develop detailed arguments.
  3. Polish the Tone
    • Prompt: “Make this sound formal and suited for a business report.”
    • Use: Align with your target audience.
  4. Condense the Message
    • Prompt: “Summarize these points into a two-line conclusion.”
    • Use: Create clear takeaways for busy readers.

See how it works?

Now that you know how to use this method, let’s focus on crafting prompts for iterative refinement.

Klariti Checklist for Iterative Chain-of-Thought Prompts

Apply this prompt method with these steps:

  1. Start with a Clear Task
    • Bad: “Explain something.”
    • Good: “Explain the benefits of renewable energy in three points.”
  2. Refine the Output Step by Step
    • Example: “Expand this with examples,” followed by, “Make it sound more engaging.”
  3. Focus on Specific Improvements
    • Bad: “Make this better.”
    • Good: “Rewrite this to sound more conversational.”
  4. Test for Clarity and Brevity
    • Example: “Summarize these ideas into a two-line pitch.”
  5. Iterate Until Satisfied
    • Example: “This is good, but can you add a more persuasive conclusion?”

In the next section, we’ll explore practical use cases where iterative chain-of-thought prompts can shine.

5 Use Cases for Iterative Chain-of-Thought Prompts

From refining arguments to creating structured documents, these prompts are a game-changer.

  1. Crafting Reports or Presentations
    • Example Prompt: “Outline the challenges and solutions in digital marketing. Now add relevant data points.”
    • Benefit: Ensures a comprehensive, polished result.
  2. Editing and Proofreading
    • Example Prompt: “Rewrite this paragraph to make it clearer and more professional.”
    • Benefit: Transforms rough drafts into polished content.
  3. Developing Structured Arguments
    • Example Prompt: “Explain why education reform is important in three steps, with examples.”
    • Benefit: Breaks down complex issues into logical, persuasive points.
  4. Content Creation
    • Example Prompt: “Draft a blog post introduction. Now rewrite it to be more conversational.”
    • Benefit: Produces engaging, tailored content.
  5. Marketing Campaigns
    • Example Prompt: “Write a tagline for our product. Now add a playful tone.”
    • Benefit: Delivers refined messaging that aligns with brand voice.

Challenge: Try an Iterative Prompt Today

Ok, now over to you.

Pick a task you’re working on and iterate with ChatGPT. Start with something simple:

“Create a vegan meal plan for a family of four.”

Then refine:

“Now make it interesting by adding fish recipes from South-East Asia.”

See how the process transforms your output.

Next Steps

Iterative chain-of-thought prompts are like having a brainstorming partner who always comes up with a fresh perspective. You don’t need them for every project. Rather, use this approach when you want to dig deep and really explore a specific piece of work. Be patient with the AI and learn to write super specific prompts. Remember, it can’t read your mind. You need to tell is specifically what you want to achieve.

If you ever get stuck, try asking ChatGPT to explain what it understands about your request. This can help identify gaps between what you intended and what the AI is actually focusing on, allowing you to adjust your prompt accordingly.

Alan Thompson