AI Terms, Decoded

Plain-English definitions for the AI words you keep seeing — no jargon.

AI Agent: An AI system that can take multiple steps and use tools on its own toward a goal — not just answer one question, but plan, act, check its work, and adjust. A chatbot talks; an agent does.
API (Application Programming Interface): A defined way for two pieces of software to talk to each other automatically. When an app "connects to ChatGPT," it's usually talking to OpenAI's API behind the scenes.
Chatbot: A conversational AI that responds to messages in a chat interface. The simplest, most common form of AI most people interact with — the AI Prompt Tool on this site helps you talk to one more effectively.
Context Window: How much text an AI model can "see" and remember at once — your messages, its replies, any documents you've shared. Once a conversation gets longer than the window, the model starts forgetting the earliest parts.
Fine-tuning: Taking an already-trained AI model and training it further on a smaller, specific dataset so it gets better at one particular task or sounds like a particular voice.
Hallucination: When an AI states something false, made-up, or unsupported as if it were a confirmed fact — a fake citation, a wrong number, a person who doesn't exist. The single biggest reason to double-check anything important an AI tells you.
LLM (Large Language Model): The type of AI model behind tools like ChatGPT, Claude, and Gemini — trained on huge amounts of text to predict and generate language. "AI" in most everyday conversation actually means "an LLM."
MCP (Model Context Protocol): An open standard that lets AI models connect to real external tools and data — like a universal plug that lets an AI actually check your calendar, read a file, or use an app, instead of just talking about it.
Model: The underlying AI system itself — GPT-4, Claude, Gemini, Llama, and so on. "Which model are you using?" is asking which specific AI is doing the work behind an app or chat.
Prompt: The instructions or question you give an AI. A well-written prompt — specific, with context and a clear ask — reliably gets a better answer than a vague one. This is exactly what the AI Prompt Tool on this site helps you write.
Prompt Engineering: The skill of writing prompts that reliably get good results — giving an AI the right role, context, and format instead of just typing the first thing that comes to mind.
RAG (Retrieval-Augmented Generation): A technique where an AI looks up real, current information (a document, a database, a search result) before answering, instead of relying only on what it memorized during training. Reduces hallucinations for fact-based questions.
Temperature: A setting that controls how random or predictable an AI's output is. Low temperature gives safer, more consistent answers; high temperature gives more varied, creative (and occasionally stranger) ones.
Token: The small chunk of text an AI actually reads and writes in — roughly a word or part of a word. AI usage is usually priced by the token, which is why longer conversations and documents cost more.
Training Data: The massive collection of text (and sometimes images, audio, or code) an AI model learned from before it was released. What's in the training data shapes what the model knows — and what it doesn't.
Zero-shot / Few-shot: Zero-shot means asking an AI to do a task with no examples given. Few-shot means giving it a couple of examples first to show the pattern you want. Few-shot prompts are usually more reliable for anything format-specific.