# AI Glossary

Plain-English definitions of the AI words you keep hearing, each with a link to go deeper.

Source page: https://ai.justout.today/glossary/  
Last walkthrough: Nov 20, 2026, 12:03 AM PT

## Basics

### Artificial intelligence (AI)
Computer systems doing tasks that normally need human intelligence.  
Link: https://en.wikipedia.org/wiki/Artificial_intelligence  
Checked: Nov 20, 2026, 12:03 AM PT

### Machine learning
Software that learns patterns from data instead of hand-written rules.  
Link: https://en.wikipedia.org/wiki/Machine_learning  
Checked: Nov 20, 2026, 12:03 AM PT

### Neural network
Layers of simple connected units whose weights are learned from data.  
Link: https://en.wikipedia.org/wiki/Neural_network_(machine_learning)  
Checked: Nov 20, 2026, 12:03 AM PT

### Deep learning
Machine learning with many-layered neural networks.  
Link: https://en.wikipedia.org/wiki/Deep_learning  
Checked: Nov 20, 2026, 12:03 AM PT

### Large language model (LLM)
A neural network trained on huge amounts of text to predict the next token.  
Link: https://en.wikipedia.org/wiki/Large_language_model  
Checked: Nov 20, 2026, 12:03 AM PT

### Generative AI
AI that creates new text, images, audio, video or code.  
Link: https://en.wikipedia.org/wiki/Generative_artificial_intelligence  
Checked: Nov 20, 2026, 12:03 AM PT

### AGI
Artificial general intelligence: AI matching people across most tasks. Definitions vary.  
Link: https://en.wikipedia.org/wiki/Artificial_general_intelligence  
Checked: Nov 20, 2026, 12:03 AM PT

## How models work

### Transformer
The architecture behind modern LLMs, built on attention.  
Link: https://en.wikipedia.org/wiki/Transformer_(deep_learning_architecture)  
Checked: Nov 20, 2026, 12:03 AM PT

### Attention
Lets a model weigh which earlier tokens matter for the next one.  
Link: https://en.wikipedia.org/wiki/Attention_(machine_learning)  
Checked: Nov 20, 2026, 12:03 AM PT

### Token
A chunk of text (often part of a word) that models read and write.  
Link: https://platform.openai.com/tokenizer  
Checked: bot-blocked

### Context window
How many tokens a model can consider at once.  
Link: https://en.wikipedia.org/wiki/Large_language_model  
Checked: Nov 20, 2026, 12:03 AM PT

### Embedding
A list of numbers representing meaning, used for search and similarity.  
Link: https://en.wikipedia.org/wiki/Embedding_(machine_learning)  
Checked: Nov 20, 2026, 12:03 AM PT

### Diffusion model
Generates images by gradually removing noise.  
Link: https://en.wikipedia.org/wiki/Diffusion_model  
Checked: Nov 20, 2026, 12:03 AM PT

### Mixture of experts (MoE)
A model that routes each token to a few specialist sub-networks.  
Link: https://en.wikipedia.org/wiki/Mixture_of_experts  
Checked: Nov 20, 2026, 12:03 AM PT

### Multimodal
Handles more than one kind of input or output, such as text plus images.  
Link: https://en.wikipedia.org/wiki/Multimodal_learning  
Checked: Nov 20, 2026, 12:03 AM PT

### Reasoning model
A model trained to work through steps before giving a final answer.  
Link: https://en.wikipedia.org/wiki/Reasoning_language_model  
Checked: Nov 20, 2026, 12:03 AM PT

## Training & tuning

### Fine-tuning
Further training a model on your own examples.  
Link: https://en.wikipedia.org/wiki/Fine-tuning_(deep_learning)  
Checked: Nov 20, 2026, 12:03 AM PT

### LoRA
Cheap fine-tuning that trains small add-on matrices.  
Link: https://arxiv.org/abs/2106.09685  
Checked: Nov 20, 2026, 12:03 AM PT

### RLHF
Reinforcement learning from human feedback, used to make models helpful.  
Link: https://en.wikipedia.org/wiki/Reinforcement_learning_from_human_feedback  
Checked: Nov 20, 2026, 12:03 AM PT

### Distillation
Training a small model to imitate a larger one.  
Link: https://en.wikipedia.org/wiki/Knowledge_distillation  
Checked: Nov 20, 2026, 12:03 AM PT

### Quantization
Storing weights with fewer bits so models run faster and smaller.  
Link: https://huggingface.co/docs/optimum/concept_guides/quantization  
Checked: Nov 20, 2026, 12:03 AM PT

### Overfitting
Learning the training data too closely and failing on new data.  
Link: https://en.wikipedia.org/wiki/Overfitting  
Checked: Nov 20, 2026, 12:03 AM PT

### Synthetic data
Training data generated by programs or other models.  
Link: https://en.wikipedia.org/wiki/Synthetic_data  
Checked: Nov 20, 2026, 12:03 AM PT

### Open weights
Model weights you can download and run yourself.  
Link: https://en.wikipedia.org/wiki/Open-source_artificial_intelligence  
Checked: Nov 20, 2026, 12:03 AM PT

## Using models

### Prompt
The instructions and context you give a model.  
Link: https://en.wikipedia.org/wiki/Prompt_engineering  
Checked: Nov 20, 2026, 12:03 AM PT

### Chain of thought
Asking for, or training, step-by-step reasoning.  
Link: https://arxiv.org/abs/2201.11903  
Checked: Nov 20, 2026, 12:03 AM PT

### RAG
Retrieval-augmented generation: look things up, then answer with them.  
Link: https://en.wikipedia.org/wiki/Retrieval-augmented_generation  
Checked: Nov 20, 2026, 12:03 AM PT

### Vector database
A database built to search embeddings by similarity.  
Link: https://en.wikipedia.org/wiki/Vector_database  
Checked: Nov 20, 2026, 12:03 AM PT

### Function calling / tool use
A model asking your code to run a tool with structured arguments.  
Link: https://developers.openai.com/api/docs/guides/function-calling  
Checked: Nov 20, 2026, 12:03 AM PT

### AI agent
A model in a loop that plans, calls tools and acts toward a goal.  
Link: https://en.wikipedia.org/wiki/Intelligent_agent  
Checked: Nov 20, 2026, 12:03 AM PT

### MCP
Model Context Protocol, a standard plug for connecting tools to AI apps.  
Link: https://modelcontextprotocol.io  
Checked: Nov 20, 2026, 12:03 AM PT

## Risks & safety

### Hallucination
A confident answer that is false or made up.  
Link: https://en.wikipedia.org/wiki/Hallucination_(artificial_intelligence)  
Checked: Nov 20, 2026, 12:03 AM PT

### Prompt injection
Hidden instructions in content that hijack an AI's behavior.  
Link: https://en.wikipedia.org/wiki/Prompt_injection  
Checked: Nov 20, 2026, 12:03 AM PT

### Deepfake
Synthetic media that convincingly fakes a real person.  
Link: https://en.wikipedia.org/wiki/Deepfake  
Checked: Nov 20, 2026, 12:03 AM PT

### AI alignment
Making AI systems pursue the goals people intend.  
Link: https://en.wikipedia.org/wiki/AI_alignment  
Checked: Nov 20, 2026, 12:03 AM PT

