Library 001: AI Terms You Actually Need to Know

You don’t need to learn an entirely new technical language to use AI well.

These are the terms that are actually useful for understanding what you are looking at, where your work lives, and what the AI can or cannot see.

App / Platform / UI

The app or platform is the AI product you are using, such as ChatGPT, Claude, Gemini, or Copilot.

The UI, or user interface, is the part of that product you can see and interact with: the chat window, sidebar, buttons, menus, settings, projects, and other controls.

For practical everyday use, you do not need to worry much about the technical distinction. If someone says the UI changed, they usually mean the product looks or behaves differently on the screen.

Model

The model is the AI doing the language, reasoning, or other processing underneath the product.

ChatGPT, Claude, Gemini, and Copilot are products. Those products can use different models behind the scenes.

Changing the model can change how the AI responds even when you are using the same app.

Project / Project Space / Workspace

A project, project space, or workspace is an area where related work can be kept together.

Depending on the product, it may contain conversations, files, instructions, reference material, or other resources.

Projects can be useful when you have ongoing work that should not be scattered across unrelated chats.

Exactly what a project can remember, share, or access varies by product.

Thread / Chat / Conversation

These words usually describe one ongoing conversation with an AI.

A conversation can contain much more than a single question and answer. Depending on the product, it may include:

• your messages
• the AI's responses
• images
• uploaded files
• links
• tool results
• instructions
• generated work
• decisions and context built up over time

That does not mean the AI can perfectly recall every part of an extremely long conversation forever.

What exists in the conversation and what the model can actively use at a particular moment are not always the same thing.

Pinned Thread / Pinned Chat

A pinned thread is a conversation you choose to keep easy to find instead of letting it disappear into a long list of older chats.

Pinning changes where the conversation appears in the interface.

It does not give the AI better memory or make everything inside that conversation permanently available to the model.

The exact pinning behavior varies by product.

Archived Chat

An archived chat is a conversation you remove from your normal chat list without deleting it.

The conversation still exists and can usually be accessed again through the product's archive or account controls.

Archiving is an organization feature. It does not give the conversation special memory, context, or status.

Chat Box / Message Box

The chat box or message box is the place where you type what you want the AI to respond to or do.

That can be a question, instruction, correction, request, pasted text, or follow-up.

This is where you tell the AI what you want.

You do not need special prompt language to use it.

Context

Context is the information the AI currently has available to help it answer you.

That might include the messages immediately above, earlier parts of the conversation, uploaded files, instructions, information from a project, or information retrieved through tools.

When an AI seems to have “forgotten” something, the information may still exist where you can scroll up and see it, but it is no longer available in its active context.

Context Window

The context window is the amount of information a model can work with at one time.

Think of it less like storage and more like the size of the desk the AI can work on.

You may own an entire filing cabinet of information, but only so much can be spread across the desk at once.

A large context window lets the model work with more material at the same time. It does not automatically mean perfect memory.

Memory

Memory is information that a system can retain beyond the immediate conversation.

Memory and conversation history are not the same thing.

Something appearing earlier in a chat does not necessarily mean it has been saved as long-term memory, and something stored in memory may be available in another conversation depending on the product and your settings.

Prompt

A prompt is simply what you tell the AI.

It can be a carefully written set of instructions.

It can also be:

Explain this in normal English.

That is a perfectly good prompt.

You do not need to learn a special prompt-writing formula before you are allowed to use AI.

Hallucination

A hallucination is when an AI gives you information that is wrong, unsupported, or invented, often while sounding completely confident about it.

AI-generated information should not automatically be treated as verified information just because the answer sounds polished.

The more important the answer is, the more important it is to verify it.

Agent

An agent is an AI system that can do more than simply respond with text.

Depending on the product and permissions, an agent might search for information, use tools, work through several steps, interact with software, or take actions for you.

An agent is still limited by the tools, permissions, information, and environment it has access to.

Connector

A connector lets an AI product access information from another service.

For example, a connector might allow an AI tool to search your Google Drive, read your calendar, or work with your email.

Connecting a service does not necessarily mean the AI can access everything inside it.

Access depends on the connector, permissions, product, and specific action being performed.

When the AI Starts Speaking Alien

You are allowed to tell the AI that you do not understand it.

You do not need to decode technical language before you can continue the conversation.

Try:

Explain that again in plain English. Assume I don't know the technical terminology.

Or:

Stop using AI and technical jargon. Tell me what this means in normal workplace language.

Or simply:

I don't understand any of that. Start over.

You can also interrupt with:

• Too technical.
• Shorter.
• Give me an example.
• What does that word mean?
• Why does this matter to me?
• What part of this do I actually need to know?
• Explain it without using that terminology.

You do not have to understand the AI's answer before you are allowed to correct it.

The AI is supposed to help you understand the subject.

You are not required to learn how to speak like the AI first.