Concepts & Definitions
AI, decoded.
Plain-English definitions for every AI term you’ll encounter — from tokens to agents, APIs to MCP.
Machine Learning
Teaching computers to learn from examples, not rules.
Tokens
The tiny chunks of text an AI reads, one at a time.
Prompt Engineering
The art of asking AI the right question in the right way.
Large Language Model
LLMAI trained on vast text to understand and generate language.
Context Window
How much text an AI can see and remember at once.
Embeddings
Turning words and ideas into numbers that capture meaning.
Retrieval-Augmented Generation
RAGGiving the AI access to your documents before it answers.
Fine-tuning
Specialising a general AI model on your specific data and style.
AI Agents
AI that plans and takes actions without you guiding every step.
Skills
Pre-built capabilities you attach to an AI to extend what it can do.
Connectors
Bridges that let AI read from and write to your apps and services.
API
APIA standardised way for software to talk to other software.
Model Context Protocol
MCPA universal standard for AI to connect to any tool or data source.
Function Calling
Tool UseLetting an AI invoke real code and APIs mid-reasoning.
Open Source LLMs
Frontier-quality AI models anyone can download, run, and modify.
Running Models Locally
How to run AI on your own machine — private, free, and offline.
Jevons Paradox
Making AI cheaper creates more usage, not less.
AI as Amplifier
AmplifierAI multiplies what you already are — good or chaotic.
Software Ecology
The holistic study of social and technical systems that produce software.
Temperature
tempThe dial that controls how random or predictable an AI's output is.
Hallucination
When an AI generates confident-sounding facts that are simply wrong.
System Prompt
The hidden instruction that shapes how an AI behaves before you say anything.
Chain of Thought
CoTPrompting a model to reason step-by-step before giving its final answer.
Multimodal AI
AI that understands and generates more than one type of media — text, images, audio, or video.
Vector Database
A database that stores meaning, not just text — finding similar content by concept rather than keyword.
Inference
The moment a trained model actually runs and generates output.
AI Benchmark
Standardised tests that measure what a model can actually do.
AI Safety
The field working to ensure AI systems do what humans actually want — now and as they become more capable.
Transformer
The architecture that made modern AI possible — attention over sequences.
Attention Mechanism
How a model decides which words to focus on when understanding a sentence.
Mixture of Experts
MoEA model architecture that activates only a fraction of its parameters per request.
Reinforcement Learning from Human Feedback
RLHFTraining AI to be helpful and safe by learning from human preference ratings.
Constitutional AI
CAITeaching AI to critique and revise its own outputs against a written set of principles.
Quantization
Compressing a model's weights to use less memory and run faster — with minimal quality loss.
Diffusion Model
The AI behind image generation: learning to reverse the process of adding noise.
Reasoning Model
AI that thinks through a problem step by step before giving its final answer.
Few-Shot Prompting
Giving an AI 2–5 examples of the task before asking it to do yours.
Zero-Shot Prompting
Asking an AI to do a task with no examples — just instructions.
Knowledge Cutoff
The date after which an AI model has no training data — and doesn't know what happened.
Grounding
Connecting an AI's response to real, verifiable sources rather than memory alone.
Prompt Injection
A security attack where malicious instructions hidden in content hijack an AI agent.
Prompt Caching
Reusing the processed version of a repeated prompt prefix to cut cost and latency.
Structured Output
Forcing an AI to respond in valid JSON or a specific schema — every time.
Streaming
Receiving an AI's response token by token as it's generated, not all at once.
AI Agents
An AI that plans, takes actions through tools, and adapts based on what it observes.
Function Calling / Tool Use
Tool UseHow a model triggers real code — a search, a database query, a calculation — instead of just generating text.
Computer Use
AI that can see a screen and control a computer — mouse, keyboard, and all.
Multi-Agent Systems
Multiple AI agents working together, each handling a specialised role in a larger workflow.
AI Memory
How AI agents store and retrieve information across turns, sessions, and tasks.
Agent-to-Agent Protocol
A2AA standard for AI agents from different systems to communicate and delegate tasks to each other.
Red-Teaming
Adversarial testing of AI systems to find failure modes before deployment.
Model Card
A standardised document describing what an AI model can do, how it was built, and where it falls short.
Direct Preference Optimisation
DPOA simpler alternative to RLHF that trains models on human preferences without a reward model.
Context Compression
Techniques to fit more useful information into a limited context window.
Add a concept
Every concept follows a fixed schema in src/lib/concepts.ts. Add an entry to the CONCEPTS array with: id, term, category, tagline, body, analogy, icon, difficulty, related.