alborapen

A tendency that favors or harms a person, object, or position. Numerous types of biases can arise in artificial intelligence systems.

For example, in data-driven AI systems, such as those created through machine learning, biases in data collection and training can lead to biases in the AI system. In logical AI systems, such as rule-based ones, biases may arise as a result of the engineer's view of the rules that apply in a particular environment. Biases may also occur due to online training and adaptation through interaction or as a result of personalization in cases where users are presented with recommendations or information tailored to their tastes.

Biases do not necessarily have to be related to human inclinations or the collection of data by people. They may arise, for instance, in the limited contexts in which a system is used, in which case there is no way to generalize it to other contexts.

Biases can be positive or negative, intentional or unintentional. In some cases, they can lead to discriminatory or unfair outcomes.