/ZEER-oh-shot/
Asking an AI to perform a task with no examples — just the instruction. The model relies entirely on its training to understand what you want.
Zero-shot prompting means giving an AI a task with zero examples of what the output should look like. You just describe what you want and trust the model to figure it out. 'Classify this email as spam or not spam' with no examples is zero-shot. Most casual AI interactions are zero-shot by default.
Zero-shot works surprisingly well for common tasks because large language models have seen millions of examples during training. But it falls apart on novel formats, domain-specific jargon, or when your definition of 'good' differs from the training data's. That's when you graduate to few-shot prompting.
Knowing the term 'zero-shot' lets you diagnose prompt failures: 'This is failing because I'm asking zero-shot for a highly specific format — let me add examples.'
For straightforward tasks where the model likely understands the format from training — summarization, classification, translation, simple Q&A.
Understanding zero-shot versus few-shot is the most fundamental prompting skill. It's the first thing to adjust when a prompt isn't working.
Zero shots fired — you're giving the AI zero examples to aim at. It has to figure out the target on its own.
A Mac app that coaches your AI vocabulary daily