AI-101

Lesson 8

The Feedback Loop: Iterating on Responses

AI-generated

Learning Objectives
  • Understand that first responses are rarely perfect
  • Know how to give effective feedback to AI
  • Use follow-up prompts to refine, expand, or redirect
  • Recognize when to iterate vs. when to start over
  • Build confidence in steering the conversation
Introduction

You wrote a good prompt. AI responded. And the response is... fine. Not bad. But not quite what you needed.

What now?

Many people start over with a new conversation. That is usually the wrong move. The better approach is iteration: refining the response through follow-up prompts until it matches what you actually need.

This lesson teaches you how to iterate effectively. By the end, you will see AI conversations as collaborative drafting sessions, not one-shot guesses.

First Draft, Not Final Answer

Here is a mindset shift that will change how you use AI:

Treat every AI response as a first draft, not a final answer.

Writers know that first drafts are never perfect. They are starting points. The real writing happens in revision.

AI works the same way. The first response captures the general direction. Follow-ups refine it into something useful.

Why First Responses Are Imperfect

  • AI cannot read your mind about preferences you did not state
  • Your original prompt might have been missing context
  • AI defaults to certain styles that may not match your needs
  • Complex tasks benefit from back-and-forth refinement

This is not AI failing. This is how collaboration works.

Feedback That Works: Be Specific, Be Direct

The key to good iteration: tell AI exactly what to change.

Vague feedback gets vague revisions. Specific feedback gets precise improvements.

Examples of Weak vs. Strong Feedback

Weak FeedbackStrong Feedback
"Make it better""The second paragraph is too vague. Add specific numbers or examples."
"That's not right""The tone is too formal. Rewrite like I'm texting a friend."
"Try again""You focused on benefits but I need features. List the technical specs instead."
"I don't like it""Remove the joke in the opening. Keep everything else."

The Magic Phrases

These follow-up patterns work consistently:

  • "Keep the structure but make the tone more casual."
  • "Good content but cut this in half. Remove the least important points."
  • "I like the first two sections. Rewrite only the third section to focus on [X]."
  • "Add three specific examples to support the main argument."
  • "The facts are good but the formatting is hard to read. Use bullet points instead of paragraphs."
Common Follow-Ups and When to Use Them

Length Adjustments

When You GetSay This
Too long"Cut this to [X] words. Keep only the essentials."
Too short"Expand this with more detail. Especially the section about [X]."
Uneven sections"The first half is too detailed, the second half too brief. Balance them."

Tone Adjustments

When You GetSay This
Too formal"Make this conversational, like explaining to a friend."
Too casual"More professional please. This is for a business context."
Too generic"Add personality. Be more opinionated about what works best."
Too enthusiastic"Dial back the excitement. Just the facts, neutral tone."

Content Adjustments

When You GetSay This
Missing information"Add a section about [specific topic]."
Wrong focus"Less about [X], more about [Y]."
Too abstract"Replace the general claims with specific examples."
Too detailed"Remove the technical details. Focus on the main points."

Format Adjustments

When You GetSay This
Wall of text"Break this into sections with headers."
Too many bullet points"Convert the bullets into flowing paragraphs."
Missing structure"Organize this as: Problem, Solution, Next Steps."
Wrong output type"Actually I need this as a numbered list, not paragraphs."
When to Iterate vs. When to Reset

Iteration works well when:

  • The general direction is right but details are wrong
  • You are making incremental improvements
  • AI understood your intent but missed specific requirements
  • You are building on context from earlier in the conversation

Reset (start a new conversation) when:

  • AI fundamentally misunderstood your request
  • The conversation has gotten confused or contradictory
  • You want to try a completely different approach
  • The context has become so long that responses are degrading

The 3-Iteration Rule

A practical guideline: if you have iterated 3 times without getting closer to what you need, reset and try a different approach. Either your original framing was wrong, or you are making the same unclear request repeatedly.

Building on Previous Responses

The power of iteration comes from cumulative context. Each exchange adds to what AI knows about your needs.

Good Conversation Flow

You: Write a product description for wireless earbuds. Target: college students.

AI: [Writes description]

You: Good start. But make it shorter and add the price point of $49.

AI: [Revises with shorter length and price]

You: Perfect length. Now make it sound less like advertising and more like a friend's recommendation.

AI: [Revises tone while keeping length and price]

Notice how each follow-up builds on what worked and fixes what did not. You never had to re-explain the basic task.

Reference Previous Parts

You can point to specific parts of AI's response:

  • "Keep the bullet points from your last response but rewrite the opening paragraph."
  • "The example about the restaurant was perfect. Add two more examples like that one."
  • "Your three options are good. Expand on option 2 only."
Key Takeaways
  • First responses are drafts: Expect to iterate. That is normal.
  • Specific feedback wins: Tell AI exactly what to change.
  • Use proven patterns: Length, tone, content, and format adjustments have reliable follow-up phrases.
  • Know when to reset: 3 unsuccessful iterations means try a different approach.
  • Build on context: Reference previous responses to refine efficiently.
Try It Yourself

Practice the iteration loop:

  1. Ask AI to write a short bio for you (real or made up): "Write a 100-word professional bio for someone who works in [your field] and is applying for [type of opportunity]."
  2. Read the first draft. Identify one thing to change.
  3. Give specific feedback. For example: "Good structure but the tone is too formal. Make it sound confident but approachable."
  4. Read the revision. Identify another improvement.
  5. Give another round of feedback. For example: "Better tone. Now add a specific accomplishment with a number (like increased X by 20%)."
  6. Compare version 4 to version 1.

You will see dramatic improvement. This is the power of iteration.

Sources