5 questions · pass at 70% to earn the journey badge. You can retake it anytime.
01 What is "zero-shot" prompting?
Giving the model several worked examples Asking the model to perform a task with instructions but no examples Fine-tuning the model on your data Disabling the model’s output
02 What does "few-shot" prompting add?
Random noise A handful of input/output examples that demonstrate the desired pattern A larger model Internet access
03 When does chain-of-thought prompting help most?
Simple lookups Multi-step reasoning tasks, by asking the model to reason step by step Reducing the prompt length Image generation only
04 Which elements most improve a production prompt’s reliability?
Vague, open-ended wording Clear role/context, explicit constraints, examples, and a defined output format Maximum ambiguity to encourage creativity Removing all instructions
05 How does prompting connect to AI agents?
Agents never use prompts Agent steps (planning, tool selection, reflection) are driven by carefully designed prompts Prompts only matter for the first message Agents replace prompting with SQL