Programme

Abstract: This talk will argue that we repeatedly ask the wrong questions about legal and policy responses to new technologies such as artificial intelligence. For example, when the Australian government proposes AI Ethics Principles including “Throughout their lifecycle, AI systems should benefit individuals, society and the environment,” it is difficult to disagree. But it is also difficult to understand why such a requirement might only be applied to “AI systems” rather than, for example, “government agencies” or “Australian corporations”. In other words, there is no ethical reason why an entity ought only to be good, fair, reliable, safe, transparent and accountable when it deploys particular technologies. A similar point applies to the creation of legal requirements that only apply where “artificial intelligence”, a contested term at best, is deployed. Indeed, the problem there is worse as is evident from the history of now-obsolete or poorly targeted laws that operated within a narrow technologically-defined scope (from digital tapes to nanomaterials). None of this is to say that we do not need to think about the legal and policy implications of artificial intelligence, but we need to differentiate among: (1) legal requirements, ethical principles and technical standards that ought to apply only to “artificial intelligence” (according to some well-scoped definition), (2) legal rules that require reform in order to apply appropriately in contexts involving artificial intelligence (for example, enhancing data protection or reformulating discrimination law), (3) technical standards that specify how to meet legal requirements in the context of artificial intelligence (for example, how to construct ‘reasons for decision’ that meet the requirements of administrative law; how to evaluate systems for compliance with discrimination law), and (4) applied ethics addressing new scenarios involving artificial intelligence within broader ethical frameworks. In other words, we need to ask not “how do we regulate artificial intelligence” but rather “what needs to be done to ensure that legal, regulatory and policy frameworks are well-adapted to a world that includes artificial intelligence technologies as they continue to evolve”.

Abstract : In this talk we examine the concept of Digital Legislation, an approach where legislations are represented in a format that is at the same time understandable by human and processable by machine. We discuss the benefit of this approach and we present the theoretical underpinning for it based on Computational Law. The approach has been successfully validate by several experiments.

Abstract: In this presentation I will address the explainability of web search engines, and legal search engines in particular. The purpose of explainable methods is to give the user an understanding of why the retrieved documents are considered relevant to their information need. I will present recent work in which we added explainable elements to an existing neural ranking model. In a user study we showed that a visualization of these explainable elements on the search engine result page improved the assessability of the search results: users judge our proposed interface significantly more explainable and easier to assess than a regular search engine result page. This indicates that the explainability of the search engine result page leads to a better user experience. Our current research follows up on these results in the direction of professional search contexts, with three projects on legal information retrieval. Explainable search is particularly relevant for the legal domain, because these users are critical towards search results and have the need to be in control. I will show our work on relevance factors in legal search, and how these relevance factors can be taken into account in an explainable search engine.

Abstract: The application of artificial intelligence technologies, particularly iterative active learning, has revolutionized document review in civil discovery and other legal applications. Users of these technology-assisted review (TAR) approaches are faced with the dual need to statistically demonstrate a high level of recall (proportion of responsive documents found) and to minimize costs. We present a new analytical model and visualization technique for total cost in one-phase and two-phase TAR workflows. We show how task characteristics affect optimal choice of workflow structure, active learning method, sample size, and stopping rule. We end by presenting several open problems in statistical quality control for TAR.

Abstract : Question Answering has been a challenging problem to solve in the legal domain due to the complex nature of the legal text. To effectively solve the problem of question answering in legal domain, the answers presented need to be relevant to the question asked, but also present diverse answers so that it can guide the legal researchers with the ill-formed queries. While some approaches have been proposed and implemented to solve this problem like factoid-based question answering and BM25 based retrieval systems, these approaches have suffered in the areas of either relevance or novelty or both.
In this presentation, I will be presenting the approaches implemented at LexisNexis for answering the factoid questions for a limited number of legal question types like statute of limitations, doctrines, etc. Furthermore, I will be discussing a high-level overview of our Neural Information Retrieval based Open Legal domain Question Answering which expands on the coverage of various legal domain Question types which not only address factoid questions but also answers complex procedural questions.

Abstract : Advances in Legal AI have been hindered by the dearth of human-annotated legal data sets. While labelled documents represent an essential input for the development and optimization of supervised machine learning models, producing high-quality legal annotations for supervised tasks presents important challenges. Annotating legal documents is a tedious and often expensive process, requiring annotators to possess a degree of domain expertise beyond what can be expected from crowdworkers. In my talk, I sketch out how an ambitious open-source project might be able to tackle this data bottleneck by embedding legal education in AI legal research.

Abstract: Machine learning models are often assumed to be sufficiently abstract (and, therefore, anonymous) to avoid the applicability of data protection law. However, personal data can in fact be derived from these models in many cases. The talk will present the consequences from the perspective of European data protection law. In case the General Data Protection Regulation (GDPR) applies, the data subjects' rights, as well as the controllers' responsibilities, have to be taken into account in system design. Moreover, automated individual decision-making is restricted, seemingly limiting potentially useful AI applications. We will discuss whether these limitations are a sensible attempt at protecting the privacy of data subjects, or an unnecessary barrier to innovation.

Abstract: This talk will begin by examining the roots of Artificial Intelligence and Law – including applications involving NLP, data mining, machine learning, and more broadly, data analytics – noting that it has been around for much longer than the recent buzz would suggest. We will explore the field of AI and Law in terms of its development and expansion starting in the 1980s and study how seminal research was conducted and reported on in conference proceedings such as ICAIL and publications such as the AI and Law journal. After having established the foundations of today’s field of AI and Law, we will look to the future and examine some of the use cases and AI-based applications that have been created to address them. These include next-generation tools for legal professionals that can augment their skill sets by providing analytical abilities to help in the crafting of legal strategies. Lastly, we will investigate the contributions that more recent neural-based capabilities (deep learning) are making in the field, as we look still further into what the future may hold for the intersection of AI and Law.

Abstract: The field of Artificial Intelligence and Law studies how legal argumentation can be formalized in order, eventually, to be able to develop systems that assist lawyers in the tasks of researching, drafting and evaluating arguments in a professional setting. To further this goal, researchers have been developing systems, which, to a limited extent, autonomously engage in legal reasoning, and argumentation on closed domains. However, populating such systems with formalized domain knowledge is the main bottleneck preventing them from making real contributions to legal practice. Given the recent advances in natural language processing, the field has begun to apply more sophisticated methods to legal document analysis and to tackle more complex tasks. Meanwhile, the LegalTech community is thriving and companies/startups have also been trying to tap into the legal industry's need to make large-scale document analysis tasks more efficient, and to use predictive analytics for better decision making. This talk will present an overview of the history and state of the art in academic AI&Law, as well as selected examples of developments in the private sector. Aspects in focus are rule- and case-based reasoning, legal text analytics, and the use of predictive models.

Abstract: Traditionally, the field of AI and Law has focused on representing legal knowledge in ways that computers can use to perform legal reasoning, or something like it, with legally intelligible results. Today, the research paradigm in AI and Law has largely shifted to applying new machine learning and natural language processing techniques to legal texts. Although for some time ML models have been predicting outcomes of cases directly from their texts, they cannot yet explain their predictions or support them with arguments. This talk surveys recent research efforts that tease elements of legal meaning, including legal concepts and argument structures, from legal texts. These methods can improve legal information retrieval and may eventually enable ML models to explain and justify their results.

Abstract: Post 9/11 counter-terrorism decision-making processes are characterized by a mounting reliance on predictive technologies, including drone and satellite imaging and artificial intelligence (AI) targeting systems. This presentation sheds light on the effects of these predictive technologies, and the opaque epistemologies they entail, on counter-terrorism decision-making processes. Predictive technologies add overwhelming amounts of time-sensitive and relevant data, collected and analysed to improve responses to the threats of terrorism. At the same time, these methods may place additional burdens on decision-makers, by legitimising pre-emptive fact-finding processes, and creating a persuasive virtual reality that is difficult, if not impossible, to refute. The presentation focuses on three core problems: first, predictive and visualization technologies may malfunction, and have technical limitations, including insufficient or corrupt data inputs, blind spots, and time and space constraints. Second, the outputs of these technologies may reduce situational awareness, as decision-makers tend to place an inappropriately high level of trust in these outputs. Third, reliance on AI technologies creates an accountability gap, as technology-induced errors remain unaccounted for. Based on an interdisciplinary scholarship in law, science, and technology, as well as empirical observations from investigations into battlefield operations, this presentation develops several recommendations to better incorporate predictive technologies into counter-terrorism decision-making processes.

Abstract: When we discuss technology in Law, we often tend to discuss how wonderful the current state of data-driven Artificial Intelligence is and how it will be disrupting the legal industry for good. On the other hand, we can not argue the fact that the adoption of such technologies is low among legal industry and laymen alike. When we discuss the future of AI technology, we need to keep both scenarios in mind and find constructive ways to improve the adoption while handling the complications involved with emerging technologies.
In this session, I will discuss three major topics that may help AI products to reach millions of users: Firstly, an overview of the strength and weaknesses of the present state of AI in the legal sector. Secondly, how can we overcome the current limitations with the help of Design and Behavioural Psychology. Lastly, how can we build a sustainable ecosystem for AI products. Attending this session will give you strong grounded concepts of what AI product in the legal industry should look like that will improve the adoption among a diverse set of users.

Abstract: In this session, Professor Maura R. Grossman—a well-known pioneer in the use of machine learning in legal and other applications—will provide an overview of how technology-assisted review (TAR) has been used in electronic discovery to find relevant evidence in civil litigation and regulatory investigations, and how the same machine-learning technology has been applied to the curation of government records and systematic review in evidence-based medicine, including for research related to COVID-19.

Abstract: Arbitration has emerged as the preferred mode for the resolution of commercial disputes. Disputes of all genres ranging from relatively simple disputes about erroneous invoices to complex service contracts for the supply of services in oil drilling rigs are dealt with by arbitrators. Considering the huge sums of money linked to the disputes and to make the arbitration process more efficient, artificial intelligence-based tools have been suggested as a potential alternative.
The question at this point however remains as to where does artificial intelligence comes into play in arbitration. In this talk, I will discuss the different use case scenarios of artificial intelligence-based tools in arbitration and also look into a few examples where they have been sought to be utilized. This will be followed by a look into the changes in arbitration rules which have been brought into force to facilitate AI-based tools. Finally, an attempt will be made to understand the future for AI in arbitration with an effort to understand, if there will be a need or acceptability of an AI arbitrator.

Abstract: A client who runs into legal problems about debt or housing might turn to a legal aid advisor on how to resolve them. The client must provide the advisor with detailed personal data and preferences; the advisor must reason over the data and preferences; then a solution must be selected. A range of social service organisations and debtors may need to be considered. The problem is fairly complex and requires expert advice. In the UK, charities often provide legal advice, though with insufficient funding or staffing. This gives rise to a gulf between the need for advice and the resources to provide it.
The talk presents some initial work on the COVID-19 Debt Advice Project to address legal aid with an online consultation tool.