Model Training & Adaptation
Causal AI
AI methods that try to work with cause and effect, rather than just matching patterns.
Example
An AI system estimates how changing a promotion might affect sales.
Why people use it
It helps explore what might change when someone takes an action.
What you'll hear
“Would changing the price actually change sales?”
What this means for you
Ask which causal assumptions were tested and which remain uncertain.
Can you control it?
Developer-only
The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.
Common questions
- Does calling a system causal prove its conclusions?
- No. Its assumptions and supporting evidence still need examination.
- Does it need more than ordinary prediction?
- Usually. It needs assumptions or evidence about how causes and effects are connected.
- Can it study an action we haven't taken yet?
- It can estimate possible effects, but the estimate depends on assumptions that may not hold.