Bachelor's Thesis Jan Rothweiler

Evaluating Legal AI Use Cases based on Experience Reports from the DACH Region

The creation of the Legal AI Use Case Radar at the end of 2023 and the publication of its complementary report in July 2024 marked the first iteration of the project. Now, in the second iteration I am able to contribute with my thesis. In a process which consists of literature research, interviews, AI use case identification and further online surveys, my task is to find a pool of legal professionals and hold semi-structured interviews with them. The results will be processed and categorized into distinct AI Use Cases. My thesis will focus on tracking shifts in the significance of specific use cases and NLP technology categories over the past year.

My thesis is based on the interview part of the project. In our endeavor to create an objective standard of distinct AI use cases in the legal landscape, we must consult the industry. For this, we will contact a wide range of professionals, being firms that develop software products for legal use, consultants in the field, and legal practitioners who use the aforementioned products.

My methodology consists of several steps. The first step of my thesis will be developing a strong understanding of the AI Use Cases and the NLP technology categories. I will read into related work in the field and past theses from this project. Through this, I am prepared when holding the interviews. 

For the second step I will need to build contact with a large amount of potential interviewees. I must do extensive research about legal professionals in the DACH region and seek in-person interactions. All interviews are set to be held online, however I would be interested in potential in-person interviews, if possible.

The interview will be set for about 45 minutes to 1 hour. The style of the interviews will be semi-structured. I prepare a set of questions, which I also send to the interviewee in advance. While I will stick to these questions during the interview, I’ll allow room for digression into related topics. To achieve my goal of deriving the AI use cases, I will base my questions around efficiency on daily activities and decision-making, ethics and compliance and challenges connected to the usage of AI. These areas are directly linked to our evaluation metrics for the Use Case Radar. 

The last step of my methodology is the evaluation of the interviews. First I must transcribe the recorded interviews and identify each individual use case. Then, I will assign each to an overarching AI use case identified in the first iteration of the project. From these results our team is able to derive an overview of current trends of AI in the legal industry. In combination with academic literature and online surveys the updated Use Case Radar will be developed. 

One foreseeable outcome I anticipate is the interviews will likely vary depending on the role of the interviewee within the legal ecosystem. For example, it is probable that a provider will speak highly of their product and list a range of use cases it offers. It is my responsibility to inquire further and break down fundamental use cases.

Research Questions: 

- How has the relevance of specific AI use cases in the legal landscape changed over the past year?

- How does AI impact decision-making processes within legal practices, and are there significant improvements or concerns? 

- What are the predominant AI use cases in the legal domain?