AI Glossary

Every AI term you'll meet — explained without jargon.

Large Language Model (LLM)
A neural network trained on massive text data that can read, reason and generate language.
Retrieval-Augmented Generation (RAG)
A pattern that grounds an LLM in your own documents by retrieving relevant passages at query time.
Embeddings
Numeric vector representations of text, images or audio that capture semantic meaning.
Fine-tuning
Continuing to train a base model on your own examples so it adopts a style or skill.
Prompt Engineering
The craft of writing instructions that consistently produce the output you want from an LLM.
AI Agents
Systems where an LLM plans a goal, calls tools, observes results and iterates until done.
Vector Database
A database optimized for storing and searching embeddings by similarity.
Context Window
The maximum number of tokens an LLM can read and remember in one request.
Tokens
The chunks of text (usually sub-words) that LLMs read and bill on.
Multimodal AI
Models that understand and generate across text, images, audio and video.
Chain of Thought
Prompting technique where the model reasons step-by-step before answering.
Reasoning Models
LLMs that spend extra compute thinking before they answer.
Diffusion Models
Generative models that create images by iteratively denoising random noise.
Fine-tuning vs RAG
Two ways to teach an LLM about new information.
Hallucination
When an LLM confidently produces false information.
Mixture of Experts (MoE)
An architecture that routes each token to a small subset of specialized expert networks.
Open Weights
Models whose weights are downloadable and runnable locally.
Open Source AI
AI tools whose source code is openly available under a permissive license.
Semantic Search
Search that ranks results by meaning instead of keyword overlap.
Tool Use
When an LLM calls external functions or APIs to take action in the real world.