PAPERS
The LegalAIIA workshop organizers fielded a respectable set of submissions this year.
The accepted papers and their authors and author affiliations are listed below.
Proceedings: http://CEUR-WS.org/Vol-2888/
IDEATION PAPERS
Legal AI Systems in the EU's Proposed AI Regulation
Sebastian Felix Schwemer, Letizia Tomada, Tommaso Pasini+
Centre for iInformation and Innovation Law (CIIL), +Dept. of Computer Science
University of Copenhagen, Denmark
ABSTRACT
In this paper we examine how human-machine interaction in the legal sector is suggested to be regulated in the EU’s recently proposed Artificial Intelligence Act. First, we provide a brief background and overview of the proposal. Then we turn towards the assessment of high-risk AI systems for the legal tasks as well as the obligations for such AI systems in terms of human-machine interaction. We argue that whereas the proposed definition of AI system is broad, the concrete high-risk area of ‘administration of justice and democratic processes’, despite coming with considerable legal uncertainty, is narrow and unlikely to extent into many uses of legal AI and IA systems. Nonetheless, these regulatory developments may be of great relevance for current and future legal AI and IA systems.
APPLICATION PAPERS
Utilizing AI to Improve Efficiency of the Environmental and Land Court in the Kenyan Judiciary: Leveraging AI Capabilities in Land Dispute Cases in the Kenyan Environmental and Land Court System
Leveraging AI Capabilities in Land Dispute Cases in the Kenyan Environmental and Land Court System
Florence Ogonjo, Joseph Theuri Gitonga, Angeline Wairegi, Isaac Rutenberg
Center for Intellectual Property and Information Technology Law (CIPIT)
Strathmore University, Nairobi, Kenya
ABSTRACT
The number of land disputes in Kenya continues to increase with population and economic growth. In 2013, the judiciary established the Environment and Land Court (ELC) to hear disputes relating to environment and land. Unfortunately, the ELC is plagued with the same problems affecting Kenya’s other courts; chief amongst these is an extensive backlog of cases. Past attempts by the judiciary to eliminate this backlog have met with varying degrees of success. In this paper, we argue that augmenting human abilities with AI technology is a viable means of tackling this case backlog. This paper outlines AI tools that may aid legal personnel in the ELC in performing their duties and, ultimately, reducing the number of pending cases.
RESEARCH PAPERS
On the Effectiveness of Porrtable Models versus Human Expertise under Continuous Active Learning
Jeremy Pickens, Thomas C. Gricks III, Esq
OpenText, Waterloo
ABSTRACT
eDiscovery is the process of identifying, preserving, collecting, reviewing, and producing to requesting parties electronically stored information that is potentially relevant to a civil litigation or regulatory inquiry. Of these activities, the review component is by far the most expensive and time consuming [8]. Modern, effective approaches to document review run the gamut from pure human-driven processes such as Boolean keyword search followed by linear review, to predominantly AI-driven approaches using various forms of machine learning. A review process that involves a significant, though not exclusive, supervised machine learning component is\ typically referred to as technology assisted review (TAR).
One of the most efficient approaches to TAR in recent years involves a combined human-machine (IA, or intelligence amplification) approach known as Continuous Active Learning (CAL) [5]. As with any TAR review, a CAL review will benefit in some measure by overcoming the cold start problem: The machine typically cannot begin making predictions until it has been fed some number of training documents, aka seeds. In an early CAL approach, initial sets of training documents were selected via human effort, e.g., manual keyword searching. This approach to selecting seed documents relies on human knowledge and intuition.
Recently in the legal technology sector, another seeding approach that does not rely on human assessment of the review collection but is based on artificial intelligence (AI) methods and derived from documents outside the collection has been gaining momentum. For this technique, which is often referred to as “portable models”, and known in the wider machine learning community as transfer learning, initial seed documents are selected not via human input, but by predictions from a machine learning model trained using documents from prior matters or related datasets. Portable models take a pure AI approach and eschew human knowledge in the cold start seeding process.
Notwithstanding the benefits asserted by the proponents of portable models as a seed-generation technique, we are aware of no formal or even informal studies addressing the overall impact of human-driven seeding. It is an open question whether technology assisted review seeded by portable models offers a clear, sustained advantage over approaches that begin with human input. Therefore, this work constitutes an initial study into the relationship between human vs machine seeding and overall review efficiency.