What is data for AI?
Data is the examples or measurements a model learns from or draws its answers from.
What is it?
Data are the examples or measurements a model learns from or draws its answers from: text, images, sound, numbers or records of events. The quality and make-up of the data strongly affect what a model can do and where it goes wrong.
An everyday example
A speech recogniser trained mainly on one dialect may understand other dialects less well.
Why does it matter?
Knowing what the data looked like makes it easier to predict a system’s mistakes and to judge whether it suits your task, language and setting.
A common misconception
That more data always means better data. A large but biased or outdated collection can produce a worse model than a smaller, carefully chosen one.
In more detail
Is all data text?
No. It can be words, images, sound, numbers, labels or sensor readings.
What makes data good?
It is relevant to the task, accurate, well described, varied enough and lawfully obtained.
Why does it matter who is in the data?
Because missing groups, languages or situations can lead to worse results for certain people.
Can a model use private data?
That depends on how and where the data is collected, stored, processed and shared. Do not enter sensitive data without a clear legal basis and protection.
Remember: More data does not necessarily mean better data.
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.