🎯 Lesson Objectives

By the end of this lesson, you will:

Know the 8 most common prompting mistakes
Understand why each mistake causes problems
Have practical fixes for each mistake
Use a pre-send checklist to avoid errors

📖 Learn From Mistakes

The fastest way to improve at prompting is to learn what NOT to do. Here are the 8 most common mistakes and how to fix them.

❌ Mistake 1: Being Too Vague

The Problem:

"Help me with code"
"Make something for my website"
"I need a function"
"Create an app"

Why It Fails:

AI has no idea what you actually want. It will either:

Ask clarifying questions (wastes time)
Guess wrong (frustrating)
Give generic, unhelpful response

Real Example:

What You Said	What AI Made	What You Wanted
"Make a form"	Random contact form	Login form with email, password, validation

The Fix:

❌ Vague

✅ Specific

Help me with code

Create a JavaScript function that validates email addresses. Return true for valid (containing @ and domain), false for invalid. Include 3 examples.

Quick Check: Before sending, ask yourself: "Could AI interpret this 5 different ways?" If yes, add more detail.

❌ Mistake 2: Information Overload

The Problem:

"Create a website with nav bar with dropdowns and hero section 
with slideshow and contact form with validation and footer with 
social links and make it responsive with blue/white scheme and 
add animations and make it fast and accessible and add blog 
section with categories and tags and comments and auth..."

Why It Fails:

AI gets overwhelmed and:

Misses requirements buried in text
Makes random priority decisions
Produces incomplete output
May hit token limits

The Fix:

Break complex projects into focused prompts.

❌ Everything at Once

✅ Step by Step

One massive prompt with 15+ requirements

Prompt 1: 'Create responsive nav bar with logo left, 4 links right, hamburger for mobile'

Prompt 2: 'Add hero section with heading, subtitle, CTA button'

Prompt 3: 'Create contact form with validation'

The 3-5 Rule: Each prompt should focus on 3-5 related things maximum.

❌ Mistake 3: Missing Context

The Problem:

"Add a button"
"Fix this error"
"Make it better"
"Change the style"

Why It Fails:

AI doesn't know:

What project, language, or framework you're using
What already exists in your code
What "better" or "fix" means for your situation

The Fix:

❌ No Context

✅ With Context

Add a button

I'm building a React signup form with Tailwind CSS. Add a 'Clear Form' button that resets all fields. Style it gray to differentiate from the blue Submit button.

Context Checklist — Before prompting:

Technology/language mentioned?
Project type described?
What currently exists?
Specific outcome defined?

❌ Mistake 4: Assuming AI Remembers

The Problem:

"Now add the other thing we talked about"
"Use the same style as before"
"Fix that issue I mentioned"
"Continue from where we left off"

Why It Fails:

AI conversations can lose context over many messages
"Before" and "earlier" are vague references
You might be in a new conversation

The Fix:

❌ Vague Reference

✅ Explicit Details

Use the same style as before

Use same style: blue buttons (#4A90D9), white text, 8px border radius, Inter font

Pro Tip: For long projects, keep a note of:

Color codes used
Components completed
Key decisions made

Then reference this note in prompts when needed.

❌ Mistake 5: One-and-Done Mentality

The Problem:

Send prompt → Accept first result → Move on

Why It Fails:

First responses are rarely perfect
AI might misinterpret something
Small issues compound into bigger problems

The Fix:

Always review and iterate. Plan for 2-4 exchanges minimum.

Review Checklist — After every AI response:

❌ Mistake 6: Vague Feedback

The Problem:

"This isn't right"
"Make it better"
"Fix it"
"I don't like it"
"Try again"

Why It Fails:

AI doesn't know:

WHAT specifically is wrong
HOW you want it changed
What "better" means to you

The Fix:

❌ Vague Feedback

✅ Specific Feedback

This isn't right

The button should be on the right, not left. Also text is too small — make it 16px instead of 12px.

Make it better

Improve design: add 40px spacing between sections, make heading larger and bolder, add subtle shadow to card

Feedback Formula: [What's wrong] + [What I want instead] + [Specific details]

❌ Mistake 7: Changing Too Much at Once

The Problem:

"Make it bigger, change color to red, add animation, 
fix spacing, add shadow, change font, make responsive, 
add hover effect"

Why It Fails:

Hard to verify each change worked
If something breaks, which change caused it?
AI might miss some requests
Changes can conflict

The Fix:

Change 1-3 related things at a time.

The Grouping Principle — Group RELATED changes:

✅ Size + Spacing (both layout)
✅ Color + Shadow (both visual)
✅ Hover + Animation (both interaction)

Don't mix unrelated changes:

❌ Color + Responsive + Database + Font (too scattered)

❌ Mistake 8: Ignoring Edge Cases

The Problem:

"Create a login form"
→ Works with correct credentials
→ But what about wrong password? Empty fields? Long input?

Why It Fails:

AI often creates the "happy path" — code that works when everything goes right. But real users:

Make typos
Leave fields empty
Enter unexpected data
Try to break things

The Fix:

Explicitly ask for edge case handling:

Edge Case Handling Example

Create a calculator with error handling: - Division by zero: Show 'Cannot divide by zero' - Empty input: Treat as 0 - Numbers too large: Show 'Number too large' - Clear button: Reset to 0

ChatGPT Claude

Common Edge Cases:

Input Type	Edge Cases to Test
Text fields	Empty, very long, special characters
Numbers	Zero, negative, very large, decimals
Email	Invalid format, empty, very long
Passwords	Too short, empty, special characters
Forms	Submit empty, submit partial, rapid submit

📋 The Complete Pre-Send Checklist

Clarity Check:

✅ Before sending a prompt:

0/4

Expectation Check:

✅ Am I ready to receive output:

0/4

📊 Mistake Severity Guide

Mistake	How Often	Impact
Too vague	Very common	🔴 High — wrong results
Information overload	Common	🔴 High — incomplete
Missing context	Very common	🟡 Medium — generic
Assuming memory	Common	🟡 Medium — wrong details
One-and-done	Very common	🟡 Medium — suboptimal
Vague feedback	Common	🔴 High — random changes
Too many changes	Occasional	🟡 Medium — hard to verify
Ignoring edge cases	Common	🔴 High — broken features

🏆 The 80/20 of Prompting Mistakes

If You Remember Only Three Things:

1. Be Specific Most problems come from vagueness. When in doubt, add more detail.

2. Iterate Prompting is a conversation, not a vending machine. Plan to refine.

3. Test Don't assume it works. Check the result, try edge cases, verify everything.

🎉 Module 2 Complete!

🏆 Achievement Unlocked: Prompt Master

You've completed Module 2: Speaking the Language of AI!

Skills Mastered:

✅ The Art of the Prompt
✅ The C.I.C. Framework
✅ Role Playing
✅ Reading AI-generated code
✅ Iterative prompting
✅ Practical challenges
✅ Avoiding common mistakes

✅ Lesson Summary

The 8 Common Mistakes

Being too vague — Add specific details
Information overload — Break into 3-5 item prompts
Missing context — Include tech, project, situation
Assuming memory — Restate important details
One-and-done — Plan to iterate 2-4 times
Vague feedback — What's wrong + what you want
Changing too much — 1-3 related changes at a time
Ignoring edge cases — Ask for error handling

📝 Mini Quiz

📝 Check Your Understanding

1/3

What is the '3-5 rule' for prompts?