Evaluation & Quality
Log loss
A prediction score that penalizes wrong answers, especially when the AI gives them a high chance.
Example
A system gets a large penalty for giving almost no chance to the answer that proves correct.
Why people use it
It penalizes predictions that are confidently wrong rather than just counting right and wrong choices.
What you'll hear
“A confident wrong guess gets a bigger penalty.”
What this means for you
Consider practical mistakes as well as the overall score.
Can you control it?
No
No direct control. This describes a wider issue, concept or result rather than something you can simply switch on or off in a tool.
Common questions
- Why do confident mistakes count heavily?
- Giving almost no chance to the result that actually happens produces a particularly large penalty.
- Can two systems make the same choices but have different scores?
- Yes. Their stated chances can differ even when their final category choices match.
- Is a lower score better?
- Generally yes for this measure, when the task and comparison rules are the same.