Model Training & Adaptation
Active learning
AI choosing which examples would be most useful for someone to label or explain next.
Example
An AI system asks reviewers to label cases about which it is uncertain.
Why people use it
It can reduce labeling work by asking people to check the most useful examples.
What you'll hear
“Which cases would teach it the most?”
What this means for you
Check how selected cases are labeled and whether they represent the wider task.
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
- Is active learning the same as an AI learning from every chat?
- No. It is a specific data-selection method within a training process.
- Does it always choose the hardest cases?
- Not necessarily. Different methods judge usefulness in different ways.
- Can its choices miss part of the real-world task?
- Yes. Focusing too narrowly can leave some types of cases underrepresented.