SECAI Core · phase 6 of 15 · noun

Fairness

An AI principle that means outcomes should not disadvantage individuals based on protected or irrelevant characteristics

The Explain card

Plain English
Fairness is the principle that AI outcomes should not disadvantage people because of protected or irrelevant characteristics such as race, gender, age or disability.
Example
A hiring screener trained on past successful candidates learns to favour a particular university and penalise employment gaps. It never sees gender directly, yet the proxies it relies on produce a skewed shortlist.
Why it matters
Unfair models cause real harm, attract legal action, and indicate the system is keying on the wrong signals, which is also a security weakness. Defenders test outcomes across groups and watch for proxy variables.
Hook
Fairness: the model should judge the case, not the characteristics that should not count.

Word knowledge

How the word is built, where it came from, and what it sits beside in memory.

In a sentence

Before the hiring screener ships, the evaluation run reports fairness by comparing false-negative rates across gender and age on the same resume pool.

How it is built
  • fair root beautiful; later even-handed
  • -ness suffix state or quality
Where it came from
Origin
Old English fægernes, from fæger, beautiful or pleasing, plus the native suffix -nes; kin to Old Norse fagr and Old High German fagar
Entered the language
c. 1000
What changed
Old English used the noun for beauty. By the mid-1300s the adjective fair also meant even-handed, and the noun took that sense in the late 1300s. Evaluation write-ups of the 2010s kept the even-handed sense when they compared error rates across groups.
How it is spelled
Pattern
Adjectives add -ness to name the quality, with no extra letter after r: fairness, nearness, clearness
Pattern
The vowel of fair is spelled air, as in hair, pair, and chair
Breaks the pattern
fair and fare sound alike but come from different Old English roots, fæger and faran
Breaks the pattern
Old English wrote æ in fæger; modern English writes ai

Spelled like

  • nearness
  • clearness
  • kindness
  • hair
  • pair
  • chair
Broken into chunks
  • fair
    • unfair
    • fairly
    • fairer
  • ness
    • kindness
    • goodness
    • darkness
    • awareness
What it sits beside

Same subject

Same shape

  • kindness
  • goodness
  • awareness
  • unfairness

Responsible AI principles

Adjectives plus -ness

  • kindness
  • goodness
  • darkness

air family

  • hair
  • pair
  • chair

Where it sits in the deck

Phase 6: Model Quality, Ethics, and Responsible AI Principles

Before deploying a model, understand the quality and ethical dimensions — bias, fairness, explainability, and the human oversight needed to govern outputs responsibly.

Neighbours in this phase