Public discourse concerning the potential of the use of artificial intelligence in society describes the legal profession often as automateable. The performance of lawyers usually consists of the mastering of a particular form of language - legal prose. In order for lawyers to effectively be replaced by algorithms, the peculiarities of legal prose would need to be interpretable. From a legal perspective, this subproject deals with questions related to the automation of legal argumentation.
Hereby it is decisive to take into consideration that legal prose contains particular semantics, dynamic syntax and rule-exemption based contexts which would not be recognisable from literal wording. Advances in the reach of machine learning, in particular artificial neural networks, facilitate the discussion about the development of machines in the context of recognition of semantics, which would imply the mastering of legal prose.
The mastering of legal prose by machines is the subject of intensive work in particular in the field of Legal Tech. The subproject “Automation of Legal Argumentation” deals with both technical possibilities for implementation of such automation as well as with any associated technical shortcomings. In doing so, the perspectives of the disciplines of computer science, philosophy and both and law have to be combined, taking also the social consequences connected with automation into consideration. Which legal decisions are in practice automateable? How does the automation of legal procedures impede on the legitimating functions of law enforcement through courts? Which legal areas should be refrained from automation from the outset, in particular with regards to the perpetuation of social prejudices and discrimination?