Digitalization has long since found its way into legal work and algorithms are also changing the legal profession. Technology is helping to increase efficiency and quality and save costs. These developments are summarized under the term LegalTech. The spectrum is wide: it can involve single work processes, but also entire legal services.
LegalTech is associated above all with a specific approach. Legal advice is no longer regarded solely as an individual service, but also as a (partially) automatable and thus scalable product. LegalTech facilitates the daily work of lawyers and ensures that legal profession continues to develop. But the form and extent to which algorithms will capture the core area of legal activity in the future is quite controversial.
On the part of the legislator, there are initial steps to shape the topic. In my article to the Consumer Protection Act, I examined in what form legal services are also affected by standardization and automation, especially with respect to consumers. In this article, the question is whether and in what form software can be used in the core competence of legal proceedings – the subsumption. It is about the application of a legal norm to facts of life, i.e., the concrete case.
Subsumption: two different approaches
There are two different approaches to the question of what is meant by subsumption: Conceptual jurisprudence and evaluative jurisprudence. In conceptual jurisprudence, all legal concepts are in a strictly logical order. Standards at a particular level can be derived from those at a higher level, creating a hierarchical system from general to concrete norms. Such a system is based on the idea that all legal concepts are predetermined in an uninfluenceable way. In such a strictly logical legal system, the correct result can be found for each set of facts by subsumption. This is actually an ideal prerequisite for algorithm-based subsumption, were it not for a crucial catch: Such a conceptual system is subject to be closed and complete, and that every single fact can be legally classified by means of a logical derivation. Thus, there would have to be exactly one correct result for each circumstance.
The other method is called evaluative jurisprudence. This too is based on the notion of a system of logical deductions. But unlike the conceptual jurisprudence, the evaluative jurisprudence does not operate in a gapless system. It accepts that legal concepts are fuzzy in the margins. This means that they can be interpreted not only logically, but also by reference to prevailing norms. This fuzziness can thus be closed by normative, i.e. evaluative subsumptions. Thus, different results can be derived from the same law. Such a view assumes that legal concepts are dynamic and thus subsumption depends on the point of view. Therefore, the application of subsumption rules can lead to different results.
Dealing with haziness required
In order to deal with these uncertainties, a deep understanding of language is required - especially meaning of words, the respective context, and the underlying value system. Humans are able to do this, but machines have only been able to do so to a limited extent. The existing possibilities for using artificial intelligence (AI) to determine the meaning of legal terms are (still) limited. Moreover, the steps on the basis of which an AI arrives at findings are not always transparent (black box problem).
However, if the meaning of legal terms cannot be unambiguously deduced, a computer cannot perform legal checks in such peripheral areas. A closed and completely logical system is the prerequisite for computers to be able to do the work of humans. As soon as extraordinary circumstances come into play, the computer is overwhelmed. To put it in the context of legal subsumption: Algorithmic systems can only perform the causal, but not the normative subsumption steps.
Areas of application of LegalTech
LegalTech can thus be helpful for narrowly defined standard processes. In addition, technology can contribute to the actual adjudication process for many preliminary services. These include:
• Applications that support the work of lawyers, e.g. document management software, law firm organization, online databases.
• Technologies that (partially) automate the work of lawyers in specific areas, e.g. document analysis tools or contract generators.
• Platforms that connect lawyers with each other and with clients, such as the Common Legal Platform CLP based on the European cloud initiative Gaia X.
• Online legal services and business models that provide better access to justice, especially for low amounts in dispute, e.g., collection services and air passenger rights, as well as penalty notices. In such areas, legal services can be almost completely automated. Machines can handle cases that occur hundreds of times in a similar way largely on their own. The software takes over the entire process, from recording the data to applying for compensation. Only when the case is heading to court the lawyers do come into play again.
LegalTech changes the way legal work is done
So far, there have only been a few decisions on LegalTech applications in case law. But this is about to change. Due to the often unclear legal situation, courts will increasingly deal with their permissibility in the upcoming years. The code of professional conduct and the associated legislation are also addressing the requirements and possibilities of LegalTech.
It also reveals what LegalTech cannot do: Dealing with vagueness, interpreting facts within a legal system in the context of social norms and values, which are always subject to change. In the future, algorithms will do a lot of the preliminary work for us, but when it comes to actual legal advice, lawyers with their professional specializations, networks and experience will remain the relevant point of contact.