The exact same model can produce a shallow, generic answer or a sharp, useful one depending entirely on how you ask. Prompting is the interface through which you access everything a model is capable of — treating it as an afterthought means leaving most of that capability on the table.
This module is hands-on: below, you'll send the same task through the model twice, once plain and once with a technique applied, and directly compare the two outputs.
Try it: naive vs technique, side by side
Same task, sent twice — once plain, once with the technique you pick applied as a system prompt.
Knowledge Check
5 questions — answer all, then submit
1. What does chain-of-thought prompting ask the model to do?
2. What is the self-consistency technique?
3. What is prompt injection?
4. Why should prompt templates be versioned like code?
5. What makes few-shot prompting more effective than zero-shot for some tasks?