Cognitive biases are distortions in the way we process information in the course of our thinking.
The term bias refers to a systematic deviation of logical, rational thinking from reality.
Our brains receive an enormous amount of information every day. To process this information quickly, it uses shortcuts. Cognitive biases are therefore errors of judgment caused by our quick thinking and unconscious prejudices. This can lead us to make decisions that are irrational and unfair to others.
In artificial intelligence, algorithmic bias is a real issue. Most often, it involves training machines with subjective data.
If we’re not careful, algorithmic biases can maintain or aggravate inequalities and lead to discrimination.
Example: if an algorithm is trained on historical data that shows racist or sexist prejudices, it can reproduce these prejudices in its decisions. This can have serious consequences on important aspects of life, such as hiring, bank loans, access to education, housing, etc…
Confirmation bias
It’s the unconscious tendency to seek out information that confirms our beliefs and ignore information that contradicts them.
Selection bias
When gathering data for training, a selection is made which can never fully represent the target audience.
Historical bias
It’s the reproduction of real-world inequalities.
