Can AI be fair to everyone?

Perhaps you already know that AI can be biased or unfair to its users. But what do we actually mean when we say that an AI is ‘unfair’?

There is more than meets the eye. Since there is no universal definition of fairness, this leads to many different interpretations of what AI fairness means in practice. In other words, what is fair according to one definition may not be fair based on another. What happens if we only build an AI that addresses one type of fairness but not another?

Using examples from a type of AI application called ‘recommender systems’, we will explore:

- Various concepts of fairness in AI and their importance
- The many ways AI fairness is quantified (which can sometimes be unfair, too!)
- And why some notions of fairness conflict and what can be done about it.

(Foto: Shutterstock)