ChatGPT Terms: Prompts, Tokens, Context, and Models
Learn key ChatGPT terms like prompts, responses, context, tokens, hallucinations, and models in plain English for beginners.
What Prompts, Responses, Context, Tokens, Hallucinations, and Models Mean
If AI feels confusing, it is often because people throw around words like tokens and hallucinations as if everyone was born knowing them. You were not. Most people were not. So let’s translate the jargon into normal human language.
Why this matters
If you know these basic terms, ChatGPT starts making a lot more sense.
You will understand:
- what you are actually giving the AI
- what the AI is actually giving back
- why answers can change
- why it sometimes forgets things
- why it can sound confident and still be wrong
In other words, this is the difference between feeling lost and feeling like you know where the buttons are.
The idea in plain English
Here are the core terms:
| Term | Plain English meaning |
|---|---|
| Prompt | What you type or say to ChatGPT |
| Response | What ChatGPT gives back |
| Context | The information ChatGPT uses to answer, usually your current chat and instructions |
| Tokens | Small chunks of text the AI reads and writes, kind of like pieces of words |
| Hallucination | A made-up answer that sounds real but is not |
| Model | The underlying AI system doing the work |
A few quick explanations:
Prompt
A prompt is your instruction, question, request, or message.
Examples:
- “Explain this like I’m new to it.”
- “Write a polite email to my landlord.”
- “Summarize this article in 3 bullets.”
Response
This is the AI’s reply. Hopefully useful. Occasionally weird. Sometimes both.
Context
Context is the background ChatGPT uses to figure out what you mean. That can include:
- earlier messages in the same chat
- files or text you pasted in
- instructions you gave about tone, format, or goal
Without good context, the AI is guessing more.
Tokens
Tokens are how AI measures text. Think of them as small text units, not full sentences. More text means more tokens.
Why this matters: long chats, long documents, and long answers all use tokens. If too much context builds up, the model may lose track of earlier details.
Hallucinations
A hallucination is when the AI invents something, a quote, fact, source, statistic, or explanation, and presents it like it is true. This is not lying in the human sense. It is prediction gone off the rails.
Confident tone does not equal correct answer. That is the trap.
Models
A model is the version of the AI you are using. Different models have different strengths. One may be better at writing, another at coding, another at speed, another at handling longer context.
Same question, different model, different result. Like asking the same question to two smart people with different specialties.
Simple example
Say you type this prompt:
“Plan a simple weekend trip to Chicago for two people on a budget.”
Here is what happens:
- your prompt is the trip request
- ChatGPT’s travel plan is the response
- if you add, “We love museums and want to avoid expensive restaurants,” that becomes extra context
- the whole conversation uses tokens
- if the AI invents a fake hotel deal or a museum that does not exist, that is a hallucination
- the quality of the answer may depend on which model is answering
Quick tips
- Give enough context to help, but do not write a novel unless needed.
- Treat every answer as a draft, not gospel.
- Double-check names, dates, prices, quotes, and medical, legal, or financial claims.
- If the response is off, improve the prompt instead of giving up.
- If answers suddenly get messy in a long chat, start a fresh one.
Quick summary
A prompt is what you ask. A response is what you get back. Context is the background the AI uses. Tokens are the text units it processes. A hallucination is a convincing but false answer. A model is the AI engine behind the scenes.
Learn these six terms, and ChatGPT stops feeling like sci-fi and starts feeling like a tool.