Few-Shot Learning

Level 1

When you show the AI just a couple of examples and it figures out the pattern — like learning a game after watching someone play twice.

Level 2

Teaching AI to do a new task by giving it just a few examples in the prompt, instead of retraining it on thousands of examples.

Level 3

A technique where a model learns to perform a task from a small number of examples provided in the prompt, without updating its parameters.

Level 4

In-context learning from k demonstration examples prepended to the input, enabling task adaptation without gradient updates — a key emergent capability of large language models.

Level 5

Prompting with k exemplar input-output pairs that define a task distribution, leveraging the model's implicit Bayesian inference over latent task variables — performance scaling logarithmically with k and model size.

Sources

• NIST Trustworthy and Responsible AI Resource Center: Glossary — National Institute of Standards and Technology. Accessed 2026-07-20.