Chain of Thought (படிப்படியாக சிந்தித்தல் / Step-by-Step Logic)

Let's write the rough work on the side! (ரஃப் வொர்க் பண்ணுவோமா!)

Technical Meaning: சிந்தனைச் சங்கிலி (Sinthanai Sangili) - Forcing the AI to break down a complex problem into step-by-step logical reasoning before giving the final answer.

The Core Idea

When an AI tries to solve a complex math or logic problem in a single leap, it often makes stupid mistakes (Hallucinations). Chain of Thought (CoT) prompting is a technique where you explicitly instruct the AI to "think step by step" and write out its intermediate reasoning before outputting the final conclusion.

The Origin Story

In 2022, researchers at Google Brain discovered a massive flaw in LLMs: they failed at elementary school math word problems. But they found a magical fix. By simply appending the phrase "Let's think step by step" to the prompt, the models' accuracy jumped from 17% to 78%! Forcing the model to generate intermediate text gives it the "computational space" needed to process logic.

The Tamil Analogy

Think back to your Math Exam in School (ஸ்கூல் கணக்கு பரீட்சை).

If a complex 10-mark question appears, and you just stare at it and immediately write the final answer "X = 42" at the bottom of the paper, you will probably make a mental calculation mistake and get it wrong.

Instead, the teacher always told you: "Write the rough work on the side of the paper!"

1. Step 1: Write the formula.
2. Step 2: Substitute the values.
3. Step 3: Solve the bracket.
4. Final Answer: X = 42.

By writing down the "Chain of Thought" explicitly step-by-step, you prevent silly mistakes. Forcing an AI to show its "rough work" makes it vastly more logical and accurate!

Try It Yourself