A concrete example
I type "solve this problem step by step, show me your reasoning before you answer." The quality jumps up a notch.
Why it matters
This is the technique that unlocked models on math, code, and logic. Every reasoning model rests on it.
You mostly hear it around logic, planning, math, and reasoning models.
Don't mix it up with
Reasoning model: A reasoning model takes more time to think before answering a complex task.
Prompt engineering: Prompt engineering means phrasing a request with enough context, examples, and constraints to get a useful answer.
Common mistakes
- Assuming every task benefits from chain of thought.
- Forgetting it burns a lot more output tokens.
- Thinking a model that reasons is automatically more accurate (sometimes it reasons badly).
Quick checklist
- First I check whether the word names a concept, a tool, a risk, or a metric.
- I tie it to a concrete case: I type "solve this problem step by step, show me your reasoning before you answer." The quality jumps up a notch.
- I keep the main trap in mind: Assuming every task benefits from chain of thought.
Quick questions
What is Chain of thought in AI?
Chain of thought is the step-by-step reasoning a model can work through before it answers.
Where will I run into Chain of thought?
You mostly hear it around logic, planning, math, and reasoning models.
Which word should I read next?
Start with Reasoning model, Prompt engineering, Reasoning effort.