Causality: what causes what?
Most patterns in data do not automatically tell us what a change will cause.
What is it?
Causal AI tries to work out what causes what, not just what appears together. It answers questions such as “what would happen if we changed this?” For that it needs more than data: it also needs assumptions about how the world works.
An everyday example
People who carry umbrellas catch colds more often. The umbrella is not the cause; both follow from bad weather.
Why does it matter?
In health, policy and the economy we care about the consequences of decisions. A model that only detects associations can lead to the wrong measures.
A common misconception
That a link in the data means a cause. A correlation can be chance or the result of a third factor.
In more detail
What is correlation?
When two measurements change together. That alone does not tell us whether one causes the other.
What is causal analysis?
An attempt to understand what would happen if we deliberately changed something.
Why does it matter for AI?
Because decisions in medicine, policy or business often need more than a prediction.
Can AI work out causes by itself?
Under certain conditions it can help, using models, experiments and assumptions. Without suitable data and a good study design there is no magic shortcut.
Remember: A prediction is not the same as an explanation of causes.
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.