Artificial intelligence as the end of criminal law? On the algorithmic transformation of society
Does Artificial Intelligence (AI) imply the end of criminal law and justice as we know it? This article submits that AI is a transformative technology that seemingly assumes and optimizes the rationalities of criminal law (the effective prevention of crime; the objective, neutral and coherent application of the law etc.), namely by replacing the counterfactual guarantees of the law with the factual guarantees of technology. As a consequence, AI must not be trivialized by criminal law theory. Likewise, it is not enough to subversively criticize the current weaknesses of AI (e.g. vis-à-vis the “bias in, bias out” problem). Rather, criminal law theory should draw on the highflying promises of AI to reflect upon the foundational premises of criminal law. For a criminal law that is mostly a governance tool in the administrative and/or welfare state, AI applications promise the culmination of the law’s very objectives (like the effective inhibition and prevention of crime, e.g. by means of predictive policing; or the political determination of fuzzy sentencing rationales in sentencing algorithms that ensure equal sentences for comparable crimes). For a criminal law, however, that protects liberal freedoms and rests on inter-personal trust, AI may well lead to the passing of the law’s very ideals (e.g. of the presumption of innocence, which can no longer be upheld once everyone, ordinary citizens and judges alike, is deemed a possible risk). The question about “AI as the end of criminal law?” thus eventually raises the two-pronged question “Which criminal law for which society?”. Indeed, what is the status of freedom (esp. in a surveillance society needed to power Big Data driven algorithms), trust (esp. under the zero trust paradigm that underlies many risk assessment algorithms) and future (esp. when algorithms make predictions based on past data) once AI enters into the administration of criminal justice? These are the questions, or so I respectfully submit, that criminal law theory needs to address today in order to come up with a criminal law that is both (for pragmatic reasons) open to technology as well as (for humane reasons) sensible. In all of this, we must take to heart Joachim Hruschka’s great legacy and remain intellectually honest.
