Glossary · Generative AI
What is generative AI? The difference between AI, generative AI and a language model
By Kristian Stig Henriksen · Dear Future
Generative AI is artificial intelligence that creates new content: text, images, code or audio, from the patterns it has learned from large amounts of examples. It predicts what most likely comes next, and does it well enough that the result looks like something a person made. ChatGPT and Gemini are generative AI. They are also what your customers ask for advice.
What is AI, briefly
Artificial intelligence is an umbrella term for programs that solve tasks which otherwise require human intelligence: recognising, predicting and deciding. Most of the AI you have met over twenty years is not generative. It sorts spam, recommends a film, reads a number plate or predicts when a machine will fail. It classifies and predicts, but it does not create anything new.
What makes generative AI different?
Generative AI produces. Where a classic model answers yes or no or 73% likely, a generative model answers with a finished text, an image or a program. Technically it is still prediction, only repeated: the language model guesses the next word, then the next and the next, until the answer is written. The breakthrough in 2017 was an architecture, the transformer, that made it possible to train on enormous amounts of text and keep track of context across long passages.
The everyday consequence is that the interface is language. You do not have to learn a system; you describe what you want and get it back in fluent form. That is why adoption has moved faster than any earlier technology, and why buyers have started asking an assistant before they search.
AI, generative AI and language model: three circles inside each other
Artificial intelligence is the big circle. Generative AI is a circle inside it: the part that creates content. A large language model, an LLM, is a circle inside that again: the kind of generative AI that works with text, and the engine in ChatGPT, Gemini, Claude and Copilot. Image models are generative AI, but not language models.
The assistants are then a layer around the language model: the chat experience, the web search, the memory and the tools. When someone says ChatGPT is an LLM, that is almost right; ChatGPT is an application built on an LLM.
How ChatGPT and Gemini find their answers
A generative model has two sources. The first is training: everything it read before it was finished, compressed into patterns. It can reproduce that, but it cannot look it up, and it ends on a fixed date. The second is retrieval: with web search switched on, the assistant fetches current pages, reads them and writes its answer with references to them.
It is the second source that makes AI visibility something you can influence. What the model learned in training you cannot correct. Which pages it finds when it searches depends on whether your pages exist, can be fetched and answer clearly. Our guide to getting recommended by ChatGPT goes into exactly that.
What it means for how customers find you
When a buyer asks a generative assistant who can help with a task in Denmark, she gets one answer with a few names, not ten blue links. Either you are one of the names or you are not in consideration. That is the shift GEO and LLM SEO are about, and it is what we measure daily: who is named, who is cited, and in what tone.
Read on: what an LLM is, what GEO is and the LLM SEO guide.
From concept to numbers
What does generative AI say about your brand?
Get an AI visibility analysis: we measure daily who ChatGPT and Gemini name and cite in your category, and where you stand.
Get the analysis →Frequently asked questions
What is the difference between ChatGPT and AI?
AI is the whole field of programs that solve tasks which otherwise require human intelligence. ChatGPT is one product within it: an assistant built on a generative language model that writes answers in natural language. All of ChatGPT is AI; most AI is not ChatGPT.
Are Google AI and ChatGPT the same?
No. ChatGPT is OpenAI's assistant. Google's counterpart is Gemini, and Google Search additionally has AI Overviews and AI Mode, which build answers on Google's search index. They are all generative AI, but with different models, sources and indexes, and they do not name the same brands.
What is the difference between AI and generative AI?
AI covers all programs that recognise, predict or decide. Generative AI is the part that creates new content: text, images, code or audio. A spam filter is AI; ChatGPT is generative AI.
What is Gemini?
Gemini is Google's family of generative AI models and the name of Google's AI assistant. The models also power AI Overviews and AI Mode in Google Search. It is one of the two assistants we measure daily.