Bias
Level 1
A model systematically favouring or disadvantaging certain people or outcomes.
Term 1 of 6 in Safety and privacy
In plain language
Bias in AI is rarely deliberate. It comes from data that reflects an unequal world, so the model reproduces that pattern and applies it at scale and at speed.
Think of it like this
A mirror that flatters some faces and distorts others, but is always used as though it were accurate.
Why it matters
Wherever AI touches hiring, lending, health or policing, bias is the first thing to test for rather than the last.
Next terms in Safety and privacy
- Personally identifiable informationAny information that can be traced back to a specific person.
- Data retentionHow long a service keeps what you send it, and what it does with it.
- Prompt injectionHiding instructions inside content so a model obeys the attacker instead of you.
- AlignmentHow well an AI system's behaviour matches what people actually intend.