Generative AI is artificial intelligence that creates something new, such as text, pictures, music or computer code, when you ask it to. Older AI mostly sorted or predicted things: is this email spam, will this customer cancel? Generative AI produces content instead. You describe what you want in plain words, and it writes the letter, draws the image or summarises the report. ChatGPT, Claude and Gemini are well-known examples.
An everyday analogy
Imagine two kitchen helpers. The first one can look at a plate and tell you "that is lasagne". That is traditional AI: it recognises and labels.
The second one has tasted thousands of dishes and, when you say "something warm with spinach, no meat", cooks a new dish that fits. It has never made exactly that dish before, but it knows how dishes usually go together. That is generative AI.
What generative AI can create
- Text: emails, summaries, reports, translations, meeting notes.
- Images: illustrations, product mock-ups, diagrams from a description.
- Audio: natural-sounding speech from text, and transcription of speech into text.
- Code: small programs, spreadsheet formulas, fixes for bugs.
- Video: short clips from a text description, though this is still the least mature area.
Text tools are built on large language models. We explain those in what is an LLM. Tools that handle several of these at once are called multimodal: see what is multimodal AI.
How it works, simply
Generative AI models learn from very large amounts of examples. A text model learns from books, websites and other writing. An image model learns from pictures paired with descriptions.
From those examples, the model learns patterns: which words tend to follow which, what a "sunset over the sea" usually looks like. When you give it an instruction (a prompt), it uses those patterns to build something new, piece by piece. For a text model, that means one small piece of text (a token) at a time.
The output is new, but it is shaped entirely by what the model saw in training and what you put in your prompt. That is why a clear prompt makes such a difference: see what is a prompt.
Examples at work (fictional)
In a GP practice. Dr "Anna Keller" records a consultation with the patient's consent. A generative AI tool turns the conversation into a structured note: complaint, findings, plan. She reads it, corrects one dosage, and saves it. Ten minutes of typing becomes one minute of checking.
In a law firm. A trainee at "Brook Legal" asks an AI to turn a long email chain into a one-page chronology for a partner. The AI drafts it; the trainee checks dates against the originals.
In an office. An office manager at a small engineering company asks for a polite reminder to a client about an overdue invoice, in German and English. Done in seconds.
In each case the AI produces the first draft and a person stays responsible for the final version.
The limits
- It can be wrong. Generative AI writes what is likely, not what is verified. Made-up facts are called hallucinations.
- It reflects its training data. If the examples were biased or out of date, the output can be too.
- It does not know your business unless you give it your documents or context.
- Copyright questions around training data and outputs are still being argued in courts.
What this means for confidential data
Practical tip: use generative AI freely for general tasks, but before you upload patient records, client files or internal figures, check whether the tool's provider may use your input for training or keep it.
Why it matters: data protection law like the GDPR protects the identifiable part of your data. If a provider is allowed to reuse your input, it can remove the names, and the rest (the diagnosis, the contract terms, the figures) is no longer protected. It can then legally be pooled and sold. Follow one record on the data trail.
Where Wysor fits
Wysor is a private EU workspace for generative AI, with GPT, Claude, Gemini and more in one place. We act only as your processor, take no licence to reuse your content, and never train on it or sell it. Every connected model provider runs with zero data retention, and these commitments are in the contract.
FAQ
What is generative AI in simple words? It is AI that makes new things, like text, images or audio, from a short instruction you give it in plain language.
What is the difference between AI and generative AI? "AI" is the broad field. Generative AI is the part of it that creates content, while other AI recognises, sorts or predicts things.
Is ChatGPT generative AI? Yes. ChatGPT generates text (and, depending on the plan, images and speech) in response to your prompts.
Can I use generative AI with patient or client data? That depends on the tool's terms: whether it trains on your data, how long it keeps it, and whether it acts only as your processor. Our guide is ChatGPT safe? walks through the checks.
Further reading
Questions about using AI with sensitive data? Reach us at [email protected] or through the in-app contact form.