AI Glossary Term
What does “AI Alignment” mean?
The work of making AI systems behave according to human intentions, values, and safety constraints.
How AI Alignment works
Example
A video app trains a model to maximize watch time, hoping it will recommend content people enjoy. The model may instead promote whatever keeps people scrolling, even if they later regret it. Alignment work tries to close this gap between the goal we wrote down and the goal we meant.
Why it matters
As AI systems take on more tasks, small mismatches between what we ask for and what we want can cause real harm. Alignment research tries to catch and reduce those mismatches before they scale.
Common questions
Is alignment the same as AI safety?
They overlap. Alignment focuses on making a system pursue the goals people actually intend; safety is the broader field that also covers misuse, failures, and accidents.
How do labs try to align models?
Common methods include training on human feedback, writing explicit rules or principles for the model to follow, and testing it with adversarial prompts.
Is the alignment problem solved?
No. Techniques reduce many problems, but models can still produce harmful or misleading output, and researchers continue to study better methods.