Präventive Verhaltensvorhersage durch „künstliche Intelligenz“ Möglichkeiten, Maßstäbe, Misstrauen

Preventive Behavioural Prediction via „Artificial Intelligence” Functionality, Framework, Fear The use of machine learning for predictive data analysis by German security authorities is highly disputed. Although there are currently no verified national use-cases, fundamental rights concerns surrounding “predictive policing” are resulting in a case-law-based framework for its application. Setting aside the courts’ assessments of the technology as “particularly intrusive”, this article focuses on their problematic blanket approach. So far, the courts seem to be primarily driven by distrust and fear rather than a substantive understanding of machine learning’s functions and limitations. This further exacerbates the already dysfunctional communication with security legislation. The results undermine the authority of both, the judiciary and the legislature, and get in the way of a rational legal evaluation of the technology.