Generative AI means AI that creates new content, such as text, pictures, video, music or voices, when you describe what you want. ChatGPT writing an email, or an image maker drawing a picture from a sentence, are both generative AI.
Older AI was mostly a sorter: is this email spam or not, is this photo a cat or a dog. Generative AI is a maker. Think of the difference between a librarian who files books and an author who writes a new one in the style you ask for.
In a sentence: “Our school lets students use generative AI to brainstorm ideas, but the final essay has to be their own writing.”
What the word means
The plain point: “generative” just means it generates, or produces, something new.
NIST, the US government’s standards agency, quotes a US executive order’s definition: AI that copies the structure of what it learned from to make new, synthetic content.1
Synthetic here means machine-made rather than recorded from the real world.
A simple example
The plain point: you describe it, it makes it.
- Text: “Write a birthday message for my grandmother, warm and short.” A chatbot writes it.
- Images: “A watercolour of a lighthouse at sunset.” An image model paints it.
- Video and sound: short clips from a description, a voice reading your text aloud, or a song.
Each time, the result is new: it was not copied from one existing file, but built from patterns the model learned.
Generative AI makes things that look real. That is its power, and the reason to check where a picture, video or voice came from.
Why it matters to you
The plain point: it is useful for drafts and ideas, and it brings new risks.
- It can be wrong. NIST lists confident false answers, which people call hallucinations, as a risk of generative AI. See hallucination meaning.2
- It can fake people. Voices and faces can be copied. How to spot fake images, video and voices and deepfake self-defence help.
- Ownership is not simple. A tool letting you sell what it made is not the same as the law saying you own it. Who owns what you generate explains.
Often confused with
- AI in general: generative AI is one branch. Spam filters, face unlock and recommendation feeds are AI too, but they sort or predict rather than create.
- LLM: a large language model is the text-writing kind of generative AI. See LLM meaning.
- Agentic AI: AI that takes actions towards a goal. It usually uses generative AI inside it. See agentic AI meaning.
Questions people ask
Is ChatGPT generative AI?
Yes. ChatGPT runs on OpenAI’s GPT models, which IBM describes as foundation models that “power ChatGPT and other generative AI applications”. The G in GPT even stands for “generative”.3 See GPT meaning.
What is the difference between AI and generative AI?
Generative AI is the part of AI that makes new things. The definition NIST uses is AI models that “generate derived synthetic content”, such as “images, videos, audio, text, and other digital content.”1 Other AI sorts or predicts instead, like a spam filter deciding which emails to hide.
Can you trust what generative AI makes?
Not without checking. NIST lists “Confabulation: The production of confidently stated but erroneous or false content” among the risks of generative AI.2 Treat its work as a draft: check facts, names and numbers before you rely on them.
Learn more on this site
- Pictures and video: AI images and video for beginners.
- The different kinds: Understanding AI model types and the eight kinds of AI you already use.
- Labels on AI content: How AI content gets labelled.
- The basics: the AI glossary and AI explained for complete beginners.
- Related: Types of AI, Explained.
- Related: GPT meaning.
- Related: Generative AI vs Predictive AI.
QUICK CHECK · 3 QUESTIONS
Pick an answer to see if you got it. Nothing is recorded.
1. What does generative AI do?
- Sorts things, like spam or not spam
- Creates new content, such as text, pictures or voices, from your description
- Only unlocks your phone with your face
Show the answer
Answer: Creates new content, such as text, pictures or voices, from your description. Generative AI is a maker: you describe what you want and it produces new text, pictures, video, music or voices.
2. Which of these is sorting AI rather than generative AI?
- A spam filter deciding which emails to hide
- A chatbot writing a birthday message
- An image model painting a watercolour
Show the answer
Answer: A spam filter deciding which emails to hide. Spam filters sort or predict; generative AI creates something new, like a message or a picture.
3. How does the page suggest treating what generative AI makes?
- As finished work you can always trust
- As something copied from one existing file
- As a draft: check facts, names and numbers first
Show the answer
Answer: As a draft: check facts, names and numbers first. Generative AI can state false things confidently, so the page says to treat its work as a draft and check it before relying on it.
Related terms: Artificial intelligence (AI) · Chatbot / AI chatbot · Deep learning · GPT · Hallucination · LLM (large language model)
- 1NIST, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile (NIST AI 600-1), July 2024, read at source 1 Oct 2026“the class of AI models that emulate the structure and characteristics of input data in order to generate derived synthetic content. This can include images, videos, audio, text, and other digital content.” ↑a“the class of AI models that emulate the structure and characteristics of input data in order to generate derived synthetic content. This can include images, videos, audio, text, and other digital content.” ↑b
- 2NIST, NIST AI 600-1, July 2024, read at source 1 Oct 2026“Confabulation: The production of confidently stated but erroneous or false content”. ↑a“Confabulation: The production of confidently stated but erroneous or false content”. ↑b
- 3IBM, What is GPT (generative pretrained transformer)?, read at source 12 Oct 2026“Developed by OpenAI, these foundation models power ChatGPT and other generative AI applications” and “GPTs generate new data by applying the patterns and structure of their pretraining data to user inputs.” ↑
The comparisons on this page (sorter and maker, librarian and author) are this site’s own ways of explaining, not claims from this source.
Free printable AI cheat sheets, and AI news every week.
Sixteen one-page AI guides and checklists in one free PDF, no sign-up. And if you want to keep up, the newsletter covers what changed in AI each week, in plain English, with the sources.
SUBSCRIBE FREE ↗ GET THE CHEAT SHEET PACK