The term "quality assurance for AI models" originates from the fields of artificial intelligence, automation, big data and smart data. The aim is to ensure that artificial intelligence (AI) works reliably, correctly and fairly. AI models make decisions or recognise patterns, for example in facial recognition in smartphones or automatic text translations on the internet.
Quality assurance means here: Experts regularly check whether the AI delivers the desired results and does not make any errors. They also check that the AI does not produce any biased (i.e. unfair) results. This is particularly important when AI is used in large companies, banks or in medicine.
An illustrative example: a bank uses an AI model to assess creditworthiness. To ensure that no one is disadvantaged, experts must constantly check whether the AI is assessing everyone fairly. Only if quality assurance is carried out regularly will the AI work reliably. This results in secure and trustworthy applications in which users can rely on fair decisions.















