101
Start free
Safety and privacy

Alignment

Level 2

How well an AI system's behaviour matches what people actually intend.

Term 5 of 6 in Safety and privacy

In plain language

Alignment covers the gap between the instruction given and the outcome wanted. A model can follow the letter of a request and still do something unhelpful or harmful.

Think of it like this

A wish granted too literally.

Why it matters

Alignment is why "just tell the AI what you want" is harder than it sounds, at the scale of a product or a society.

Next terms in Safety and privacy