Mission 6: Chain of Thought

🎯 Your Learning Goals

You understand why "step by step" produces better results

You can apply different Chain-of-Thought techniques

You use CoT for math, logic, coding, and complex decisions

🧠

The Secret of Step-by-Step Thinking

Imagine asking someone: "What is 17 × 24?" – and they immediately answer: "408". Trustworthy? Maybe. But if they say:

"So, 17 × 24... I'll calculate 17 × 20 = 340, then 17 × 4 = 68, and 340 + 68 = 408."

Now you trust them! The same applies to AI. When it shows its thoughts, its answers are not only better – but also understandable and verifiable.

⚡ The Difference: Direct vs. Step by Step

❌ Without CoT (Direct Answer)

✅ With CoT (Step by Step)

Question: "A train travels 120 km in 2 hours. How far in 5 hours?"

Answer: "300 km"

No explanation, no understanding, errors possible

Question: "Solve step by step: A train travels 120 km in 2 hours. How far in 5 hours?"

Answer:
1. Speed = 120 km ÷ 2 h = 60 km/h
2. Distance = 60 km/h × 5 h = 300 km
3. Answer: 300 km

Traceable, verifiable, trustworthy

🔗 The 4 CoT Techniques

01

Zero-Shot CoT

Simplest Entry

Simply add these magic words:

"Let's think step by step."
"Solve this step by step."
"Show your reasoning."

Works immediately – no examples needed!

02

Manual CoT

You Control the Steps

Provide the structure for the AI to follow:

"Go through these steps:\n1. Identify the problem\n2. List possible solutions\n3. Evaluate each solution\n4. Recommend the best one"

Maximum control over the thinking process!

03

Self-Consistency

Multiple Checks

Let the AI work through multiple solution paths:

"Solve this in 3 different ways step by step. Then compare the results and choose the most reliable solution."

Especially useful for complex/mathematical problems!

04

Tree of Thoughts

Branching Thinking

Explore different paths systematically:

"Think out loud: Consider different approaches to this problem. For each approach: What are the pros and cons? Which approach is most promising?"

Ideal for creative or strategic problems!

🧪 The CoT Lab

Select a problem and see how CoT improves the answer:

Select a problem:

A laptop costs $1200 including 20% sales tax. How much does it cost without tax?

Select CoT technique:

Your optimized prompt:

A laptop costs $1200 including 20% sales tax. How much does it cost without tax? Let's think step by step.

How the AI might think:

1

Understand the problem: $1200 is the gross price (incl. 20% tax)

2

Formula: Gross = Net × 1.20 → Net = Gross ÷ 1.20

3

Calculation: 1200 ÷ 1.20 = $1000

4

Answer: The laptop costs $1000 without tax

🌍 CoT in Practice

📊

For Math & Calculations

"Calculate the total profit. Show each calculation step and explain the formula."

Always add "Show your work"!

💻

For Coding & Debugging

"This function has a bug. Walk through step by step what should happen vs. what actually happens."

Catches logic errors that would otherwise be missed!

📋

For Text Analysis

"Analyze this text. Go step by step: 1) Main topic, 2) Tone, 3) Target audience, 4) Key message"

Structured analysis instead of chaotic output!

⚖️

For Decisions

"I need to choose between [Option A] and [Option B]. First list all relevant factors, then evaluate each option against them."

Rational decisions instead of gut feeling!

⚠️

Common CoT Mistakes

❌

Too many steps at once:
"Analyze this, evaluate that, recommend something, create a plan..."
→ The AI skips steps or gets lost.

❌

No numbering:
"Go through all aspects" without clear order.
→ The AI jumps between topics randomly.

❌

CoT for simple questions:
"Think step by step: What is the capital of France?"
→ Overkill! CoT is meant for complex problems.

"Chain of Thought is like a GPS: Only useful when the route is complex. For a straight road, you don't need directions."

🔥 Pro Tips for CoT

1

Combine with Roles

"You are a math teacher. Explain this calculation step by step, as if you're helping a student."

2

Request Intermediate Checks

"After each step: Briefly check if the result makes sense before continuing."

3

Use for Complex Tasks

CoT is especially worth it for: Multi-step problems, logic puzzles, debugging, mathematical calculations

4

Correct the Thinking Process

If the AI makes an error: "In step 3 you calculated X, but it should be Y. Correct that and adjust the following steps."

✅ Quick Check

1. When should you use Chain of Thought?

Always, for every question

For complex, multi-step problems

Only for math tasks

2. What is Zero-Shot CoT?

Simply adding "step by step"

Giving multiple examples

manualPrescribing every step yourself

3. What does Self-Consistency do?

speedGenerate the answer faster

multipleCheck and compare multiple solution paths

Make the answer shorter

📊 Your Mission Progress

0/3 Objectives Completed

0% Total Progress

🎯 Check all learning objectives to complete the mission!

Chain of Thought