Zero-Shot Chain-of-Thought with 'Let's Think Step by Step'
What You'll Learn
You'll discover how adding a simple trigger phrase can make language models reason through problems step-by-step without providing any examples.
The Magic Phrase
In the previous lesson, you learned about Chain-of-Thought reasoning—the idea that models perform better when they decompose complex problems into intermediate steps. But what if you don't have good examples to show?
Zero-Shot Chain-of-Thought is remarkably simple: just append the phrase "Let's think step by step" (or similar variants) to your prompt. That's it. No examples needed.
How It Works
When you add this trigger phrase, the model shifts into a deliberate reasoning mode. Instead of jumping directly to an answer, it generates intermediate reasoning steps—effectively creating its own Chain-of-Thought.
Without the trigger:
Prompt: "If John has 17 apples and gives away 9, then buys 5 more, how many does he have?"
Output: "13 apples"
With the trigger:
Prompt: "If John has 17 apples and gives away 9, then buys 5 more, how many does he have? Let's think step by step."
Output: "First, John starts with 17 apples. Then he gives away 9, so 17 - 9 = 8 apples remaining. Then he buys 5 more, so 8 + 5 = 13 apples. Therefore, John has 13 apples."
Why This Works
The phrase acts like a cognitive nudge. During training, models encountered countless examples where "let's think step by step" preceded methodical reasoning. This creates a strong association between the phrase and structured problem-solving behavior.
Think of it like saying "Once upon a time..." immediately puts a storyteller in narrative mode—the trigger phrase activates a reasoning pattern.
Key Takeaway: Zero-Shot Chain-of-Thought uses simple trigger phrases like "Let's think step by step" to elicit structured reasoning without any examples, making models decompose complex problems automatically.