AI Glossary: Key Terms Explained Simply
A plain language AI glossary covering prompts, tokens, hallucinations, image and voice terms, with simple examples so you use AI tools with confidence.
This AI glossary explains the terms you keep running into, such as prompt, token and hallucination, in plain language with everyday examples. We grouped them by area: basics, text and chat, images, voice and video, and a few more advanced ideas. You do not need to memorise everything at once; read the section you need and come back for the rest.
What Are the Basic AI Terms?
- Artificial intelligence (AI): the field of making computers perform tasks that used to need human thinking, like understanding language or recognising images.
- Machine learning: computers learning patterns from many examples instead of following hand written rules.
- Generative AI: AI that creates new content such as text, images, audio or video from a description. ChatGPT and Midjourney are well known examples.
- Model: the trained "brain" that produces results. The app is the interface; the model does the work inside.
- Training: the stage where a model learns from huge amounts of data before release.
Which Terms Matter for Chat and Text Tools?
You will meet these with assistants like ChatGPT and Claude. To compare the assistants themselves, read ChatGPT vs Claude vs Gemini.
- Large language model (LLM): a model trained on vast text so it can understand questions and answer in natural language.
- Prompt: the instruction you type. Clearer prompts give better results.
- Prompt engineering: the skill of writing prompts that specify role, goal, audience and format.
- Token: the unit models split text into, a word or part of a word. Limits are usually counted in tokens.
- Context window: how much of the conversation the model can keep in mind. In very long chats it may lose early details.
- Hallucination: when a tool states something false with confidence, such as a source that does not exist.
- Temperature: a setting in some tools controlling how creative or predictable answers are.
- System prompt: base instructions that shape the assistant's behaviour before the chat starts.
The easiest way to learn a new term is to ask the assistant: "Explain [term] like I am a beginner, give an everyday example and tell me when I would need it."
Which Terms Matter for Image Generation?
Using Midjourney, Leonardo or Ideogram? These terms make a real difference. Practise them with our ready prompts for professional AI images.
- Text to image: generating a picture from a written description.
- Diffusion model: a model that starts from random noise and refines it step by step into an image matching the prompt.
- Negative prompt: things you do not want in the image, in tools that support it.
- Aspect ratio: the image shape, square, portrait or landscape.
- Seed: a number controlling randomness; reusing it with the same prompt gives similar results in some tools.
- Inpainting: selecting part of an image and changing only that area.
- Outpainting: extending an image beyond its original edges.
- Upscaling: increasing resolution and detail for print or large screens.
- Vectorisation: converting an image into a vector file that scales to any size, useful for logos and icons.
To apply upscaling, background removal and vectorisation, see our image quality guide.
Which Terms Matter for Voice and Video?
- Text to speech: a tool reads your text in a natural voice; ElevenLabs is a well known example.
- Speech to text: transcribing recordings, useful for captions and lecture notes.
- Voice cloning: a tool learning a specific voice. Only use your own or one with explicit consent.
- Deepfake: fake but realistic video or audio of a person. Using it to deceive is harmful and may be illegal.
- Text to video: generating a short clip from a description.
Any AI tool can be wrong. Never rely on a single answer for health, legal or financial decisions; verify with a reliable source or professional.
What Are a Few More Advanced Terms?
- Fine tuning: extra training on specific data so a model specialises.
- Multimodal: a model that understands several content types together.
- AI agent: a system that carries out a sequence of steps toward a goal.
- Retrieval augmented generation (RAG): the model searches given sources before answering.
- API: a way for developers to connect a model to their apps.
- Open source models: models you can download, run and adapt within their licence.
How Can You Learn These Terms Step by Step?
- Pick one section per week.
- Try each term in a real tool, like changing the aspect ratio or adding a negative prompt.
- Write a short note in your own words.
- Ask a chat assistant to quiz you.
- After a month, reread the glossary and notice how much easier it feels.
Give me a ten question multiple choice quiz on prompt, token, context window, hallucination and negative prompt. After I answer, correct me and explain each mistake.
To try these tools without subscribing to each one, ZTOOLZ offers shared access to a range of them through ZTOOLZ's own shared accounts; available tools change over time. Shared access is great for learning but gives no licence or commercial rights.
FAQ
What is the difference between AI and generative AI?
AI is the broad field; generative AI is the part that creates new content such as text, images and audio.
What does it mean when AI "hallucinates"?
It states false or invented information confidently. Verify anything important with a trusted source.
Do I need to know all these terms to use AI tools?
No. Prompt, hallucination and aspect ratio are enough to start; the rest comes with practice.
Which term matters most for beginners?
Prompt, because result quality depends heavily on how clearly you ask.
