Tradition and revolution do not usually get along well – whether this also applies to German criminal law and the advancing developments in the field of artificial intelligence will be examined in the following. Great hopes are associated with increasing autonomisation. Relief in administrative activities, fewer accidents due to self-driving cars or new diagnostic and treatment tools for the treatment of patients are supposed to provide for greater general welfare overall. But this progress comes at the price of a more difficult understanding of the system's processes, especially if they are self-learning, and at least a more difficult allocation. So how should we deal with black box problems or self-learning systems that make it difficult to assign them to a specific individual? The following article first takes stock of the status quo in German criminal law, as well as the current state of jurisprudential discussions and developments.
In this context, increasing decriminalisation, especially of negligence offences, is not to be considered bad per se. Nevertheless, even in the age of AI, criminal law responsibility structures are needed to maintain trust in the validity of the rule of law and to have a behavioural effect. The hurdles that criminal law faces in this regard require a new way of looking at the results brought about by AI systems and humans ‘together’.