Prompting
Zero-shot Learning
Asking an AI model to perform a task with no examples in the prompt — relying entirely on its general training.
01 ——
In plain English
Zero-shot learning is when you ask an AI model to do a task without giving it any examples — just instructions. Modern LLMs are surprisingly good at this because they've seen so many similar tasks during training.
Example: Just write: "Translate this English sentence to French: 'I went to the store yesterday.'" — no examples needed. The model understands and translates.
When zero-shot works well:
- Tasks the model has clearly seen many times before (translation, classification, summarisation)
- Tasks that can be unambiguously described in plain language
- Quick experiments before deciding whether to add examples
When to switch to few-shot:
- The output format matters — examples lock it in
- The task is unusual or domain-specific
- You're getting inconsistent results
Most simple consumer AI interactions are zero-shot. Production AI systems often add few-shot examples for reliability.
02 ——
Related terms
Few-shot Learning
A prompting technique where you include a handful of examples in the prompt so the AI learns the pattern you want it to follow.
Prompt Engineering
The practice of crafting inputs to an AI model carefully to get better, more reliable outputs.
Chain of Thought
A prompting technique where you ask the AI to "think step by step" before giving an answer — usually leading to better reasoning.
LLM
Large Language Model — the type of AI behind tools like ChatGPT and Claude, trained to understand and generate text.
Back to glossaryLast reviewed May 2026