Anti-bias training for AI is primarily found in the fields of Artificial Intelligence, HR work and teams, and Digital Society. The term describes training and measures that attempt to reduce prejudice and discrimination in artificial intelligences such as chatbots or automated application systems.
Many AI systems learn from data created by humans. This data often contains unintended biases, for example, relating to gender, age, or origin. Without anti-bias training, AIs could make wrong decisions and, for example, preferentially invite men to job interviews, even though women are equally qualified.
During anti-bias training for AI, algorithms are trained to recognise where biases lie within their data and learn to act more neutrally. This ensures that decisions become fairer and no one is disadvantaged.
An illustrative example: If a company uses an automated system for personnel selection, anti-bias training helps the AI to really only select applicants based on qualifications and not on name, gender or other personal characteristics. This results in fairer decision-making processes that benefit everyone involved.
Deutsch 
English (UK)