AI Glossary

Systems where AI agents autonomously break down goals, plan steps, and execute tasks using tools without continuous human intervention.

A prompting technique where an AI model explicitly generates a step-by-step reasoning path before providing a final answer.

The maximum amount of text (measured in tokens) an AI model can process and remember at any single moment during an interaction.

Complex capabilities that spontaneously appear in large AI models as they scale up, which were not explicitly programmed or predicted.

The process of taking a pre-trained AI model and further training it on a specific, smaller dataset to adapt it to a specialized task.

When an AI model generates false, fabricated, or logically inconsistent information while presenting it as factual.

A highly efficient technique for fine-tuning large models without altering all of their original billions of parameters.

The degradation of an AI model's predictive performance over time as real-world data begins to differ from its training data.

A framework that improves AI responses by dynamically retrieving facts from an external knowledge base before generating an answer.

The neural network architecture introduced in 2017 that revolutionized AI by processing entire sequences of data simultaneously using 'attention'.

A specialized database designed to store and quickly search high-dimensional numerical representations of data (embeddings).

Asking an AI model to perform a task without providing it any prior examples or demonstrations of how to complete it.