See the Difference

Same problem. One asks AI to find similar examples first.

Direct Question
You: "A store sells apples in bags of 6. If I need at least 50 apples for a party, how many bags should I buy?"

You need to buy 9 bags. 50 ÷ 6 = 8.33, and since you can't buy a partial bag, you round up to 9 bags, which gives you 54 apples.

Correct, but AI jumped straight to the answer.

Generate Examples First
You: "Recall 3 similar problems and how they were solved. Then solve: A store sells apples in bags of 6. If I need at least 50 apples for a party, how many bags should I buy?"

Similar problem 1: "Eggs come in cartons of 12. Need 100 eggs. 100÷12=8.33 → buy 9 cartons."

Similar problem 2: "Buses hold 40 people. Need to transport 150. 150÷40=3.75 → need 4 buses."

Similar problem 3: "Tiles cover 4 sq ft each. Need 30 sq ft. 30÷4=7.5 → buy 8 tiles."

Pattern: Divide, then round UP since partial units aren't possible.

Solution: 50÷6=8.33 → buy 9 bags (54 apples).

AI recognized the pattern. More confident, more thorough.

Why This Works

This mirrors how humans solve problems. When faced with something new, we think "Have I seen something like this before?" Recalling similar problems activates relevant knowledge and reveals the underlying pattern.

AI has encountered countless problems during training. Asking it to recall similar ones brings that knowledge into context, creating a richer foundation for solving your specific problem. The generated examples act like a warm-up that focuses AI's reasoning.

How to Prompt It

Different from "Show by Example"

Show by Example: YOU provide examples to teach AI your style or format. Best for writing, tone, formatting.

Generate Examples First: AI creates its own examples to warm up its reasoning. Best for problem-solving, math, coding — when you don't have examples to give.

Works Best For

The Technique

Before asking AI to solve your problem, ask it to recall 3 similar problems and their solutions. The generated examples prime better reasoning — especially for math, coding, and logic tasks where you don't have examples to provide yourself.

When to Use This

When to Skip This