Training a model versus using it
An important distinction: a model is first built, then put to use.
What is it?
A model is first trained on large amounts of data (pre-training) and may then be adapted for particular tasks. Using it is called inference: the model computes an answer, and its parameters usually do not change in the process.
An everyday example
When you tell a chatbot your name and it uses it in the next reply, the name is stored in the conversation or a separate memory feature, not in the trained model.
Why does it matter?
The distinction explains why a model does not know the latest events, how an application can still look up fresh information, and what happens to what you type in.
A common misconception
That a model learns from every conversation as it goes. Whether your data are ever used for training depends on the provider and your settings.
In more detail
What is pre-training?
The model learns general patterns from a huge number of examples. For language models, a common task is predicting the next token.
What is fine-tuning?
A model can be trained further for a particular task, language or style of answering.
Does it learn from every answer?
Usually not, in the sense of permanently changing its parameters. But the app may store data, use a memory feature or allow later training.
What is inference?
The process in which an already-trained model produces output from the current input.
Remember: “I’ll remember that” in a chat does not necessarily mean the model’s parameters have changed.
Reliable sources
Last reviewed 1 November 2026. An explainer, not professional, legal or medical advice. For rules and figures that change, check the official version.