1. Home
  2. Glossary
  3. Zero-shot prompting
AI glossary · Prompting

Zero-shot prompting

Zero-shot prompting: Zero-shot prompting is asking an AI model to do a task with instructions only, without showing it any examples of the output you want. It works well for common tasks the model has seen many times, such as summarizing or translating.

'Shot' is jargon for an example. A zero-shot prompt gives zero examples: 'Summarize this contract in five bullets' or 'Classify this support ticket as billing, technical, or account.' You are relying entirely on the model's general training to understand what a good answer looks like.

This is how most people prompt most of the time, and for familiar tasks it is fine. Modern models are strong zero-shot performers on summarizing, rewriting, explaining, translating, and answering questions. The better you describe the task, audience, and format, the better the result.

Zero-shot breaks down when your definition of 'good' is specific to you. If your team's meeting notes follow a particular structure, or your classification categories have subtle boundaries, the model will guess at your conventions and often guess wrong. That is the signal to switch to few-shot prompting and show it examples.

Example at work

A project manager types: 'Rewrite this status update for the executive steering committee. Lead with the decision needed, keep it under 100 words, and use no jargon.' No examples are provided. Because executive summaries are a common task, the result is usable after a light edit.

Why it matters

Knowing the term helps you diagnose a weak result. If a zero-shot prompt keeps missing your standard, the fix is usually not a longer description of what you want. It is one or two real examples of it.

Related terms