Reference
AI Glossary
33 AI terms explained in plain English. From tokens to transformers — everything you need to understand AI.
Jump to category
Architecture
Transformer
The neural network architecture that powers modern AI language models.
Embedding
A numerical representation of text that captures its meaning as a vector.
Vector Database
A database optimized for storing and searching AI embeddings at scale.
Diffusion Model
The AI architecture behind image generators like Midjourney and Stable Diffusion.
Attention Mechanism
The core technique that allows transformers to focus on relevant parts of the input.
Mixture of Experts (MoE)
An architecture that activates only a subset of model parameters for each input, improving efficiency.
Quantization
Compressing AI models to use less memory and run faster with minimal quality loss.
Core Concepts
LLM (Large Language Model)
A type of AI trained on massive text datasets to understand and generate human language.
GPT (Generative Pre-trained Transformer)
OpenAI's family of language models that power ChatGPT.
Token
The basic unit of text that AI models process — roughly 4 characters or 0.75 words.
AI Hallucination
When an AI model generates plausible-sounding but factually incorrect information.
Context Window
The maximum amount of text an AI model can process in a single conversation.
AI Agent
An AI system that can autonomously plan, execute tasks, and use tools to achieve goals.
Multimodal AI
AI that can process and generate multiple types of content — text, images, audio, video.
Inference
The process of running a trained AI model to generate predictions or outputs.
API (Application Programming Interface)
A way for developers to programmatically access AI models in their own applications.
Generative AI
AI systems that create new content — text, images, code, music, video.
NLP (Natural Language Processing)
The field of AI focused on understanding and generating human language.
Temperature
A setting that controls how creative or deterministic AI outputs are.
Open-Source AI
AI models with publicly available weights that anyone can download and run.
Token Limit
The maximum number of tokens that can be processed in a single API request.
Model Collapse
When AI models degrade from being trained on AI-generated content instead of human data.
System Prompt
Hidden instructions that define an AI assistant personality, behavior, and constraints.
Agentic AI
AI systems that can autonomously plan and execute multi-step tasks using tools.
Techniques
Prompt Engineering
The practice of crafting effective instructions to get better results from AI models.
RAG (Retrieval-Augmented Generation)
A technique that lets AI access external knowledge bases to provide more accurate answers.
Fine-Tuning
Training a pre-trained AI model on specialized data to improve performance on specific tasks.
LoRA (Low-Rank Adaptation)
An efficient method to fine-tune AI models using much less compute and memory.
Few-Shot Learning
Providing a few examples in your prompt to teach the AI the pattern you want.
Chain-of-Thought (CoT)
A prompting technique that asks AI to show its reasoning step by step.
Grounding
Connecting AI outputs to verified sources of information to reduce hallucinations.
Zero-Shot Learning
Asking an AI to perform a task without any examples — relying on its training alone.
Fine-Tuning vs RAG
Two different approaches to customizing AI — permanent training vs. runtime knowledge injection.