By Kelly Choo, Deaundre Espejo and Duranka Jayasinghe
Our legal landscape is rapidly evolving, and one of the most significant disruptors of the past few years is the use of predictive analytics.
Predictive analytics refers to the forecasting of outcomes through analysing quantitative data, which allows lawyers to engage in more efficient and strategic decision-making. This includes predictive judicial analytics, which involves the analysis of judicial behaviour by looking at decision-making patterns of judicial officers and predicting the likely outcome.
While this is something that lawyers have been doing anecdotally for years (e.g. ‘Judge X is generally pro-defendant in these types of cases’), the rise of Artificial Intelligence (AI) allows for a more data-driven analysis.
According to Jeff Arvidson, Director of Product Development at Thomson Reuters, judicial analytics allows lawyers to conduct more focused and targeted research based on wider sets of information.
There are several ways in which such technology can enhance legal practice. For example, if a client is seeking to pursue a particular motion strategy, AI can calculate how often a particular judge rules in favour of that type of motion, and how long it might take to obtain that ruling.
Another example is the ability to look at a judge’s experience either in an area of law or a specific industry, based on how frequently they have handled a type of case. As a result, lawyers would be better able to tailor their arguments to suit the type of legal thinking the judge has been familiar with.
“[S]uch granular data has clear implications for budget determinations and trial strategy” Arvidson writes. Indeed, the ability to quickly identify patterns in large datasets has the potential to help lawyers improve client service, refine legal research, and optimise law firm operations.
Today, platforms which currently offer this service include LexisNexis, Premonition Analytics and Bloomberg Litigation Analytics.
But despite its considerable benefits, judicial analytics poses several ethical problems for the legal profession and the law more broadly.
Concerns of Adopting Predictive Judicial Analytics
Predictive technology has been criticised as exemplifying a “shift from reason to statistics.” The nature of such technology is reliant upon quantifying large sets of data and utilising algorithms to identify correlations, which inevitably presents a number of concerns.
First, predictive analytics disregards the explicit reasoning and causal inferences that are central to judicial opinions. As conclusions are derived from unintelligible algorithms that determine correlations between various data points, a substantive understanding of the law, or the merits of any given case, is arguably rendered moot.
In fact, according to Eliot Siegel, Professor at the University of Maryland, the objective of predictive analytics is “more to predict than it is to understand the word.” This indicates not only the lack of consideration for legal argumentation, but the absence of the contextual and social frameworks through which the law is developed.
Consequently, predictive judicial analytics illustrates a transition towards a codified system that is controlled and understood exclusively by “technically sophisticated individuals.” It is thus unable to provide an explanation for the conclusions it reaches in predicting judicial outcomes to lay individuals.
This is particularly the case if forms of artificial intelligence that possess machine learning systems are applied in judicial prescriptive analytics, as they would come up with reasoning and ‘evolve’ beyond the comprehension of their creators. In such cases, fundamental principles such as the rule of law and open justice – which allow for judicial systems to be open and transparent, and judges to be held accountable through providing reasoning for their judgements – would be contravened.
Secondly, not only are these data points removing legal reasoning, but accurate predictions would result in the modelling of unconscious biases which currently exist in judicial decision-making.
A study conducted by Daniel Chen, professor at Toulouse School of Economics, looked at asylum decisions in the US since 1981. It found that the time of day significantly influenced decisions, as well as other factors such as the weather, how masculine they perceived the applicant to be, and the applicant’s family size.
A highly effective analytical tool would necessarily include these factors. However, such biases are not only difficult to account for, but have no place in decision-making.
Additionally, there can also be issues with the datasets used in judicial analytics. Predictive tools will need to be trained on thousands of historic judgements before they can spot trends and patterns, which will reproduce existing imbalances.
For example, the Australian Law Reform Commission has found that Indigenous peoples are less likely to be granted bail than non-Indigenous persons. Utilising this data would therefore normalise the uneven application of the law by accepting imbalanced predictive solutions which are skewed against Indigenous clients.
In addition to furthering existing biases, predictive analytics may become another mechanism through which inequities in access to justice continue to proliferate. The use of highly technical, and most certainly expensive, technology will be exclusively for the “most capable litigants” whilst continuing to disadvantage others.
Moving Forward, The Role of Lawyers
Finally, the use of predictive judicial analytics may shift the role of legal practitioners from traditional ‘advocates’ to mere ‘statistical advisors’. Until today, successful lawyers have been responsible for performing high-volume, routine legal tasks. This forms the groundwork necessary to appear before a judge.
These roles are now at risk of automation, with new roles for lawyers being suggested in the context of growing interconnectivity.
Particularly, the best lawyers will be expected by their clients to utilise technology to reduce legal costs on automatable tasks, maintain effective relationships between lawyer and client, and really understand how the technology they use works. An example of this is how the Hewlett Packard Enterprise legal team uses approximately 30 bespoke legal applications to support a multitude of work, including litigation, mergers and acquisitions, digital signatures, ebilling and contract negotiation.
Similarly, predictive judicial analytics could be used to ensure the best chances of obtaining a favourable result for any case. This becomes prescriptive analytics - the act of suggesting specific direction based on probable outcomes, which is dangerous because it legitimises the use of prediction technology as a tool necessary to provide good legal advice.
Consequently, the mere “doing” of legal work becomes less important for lawyers - with a thorough understanding of the implications of all that surrounds this becoming more important to enable lawyers to advise on the risks and returns of pursuing any legal course of action. This is the view of Professor Richard Susskind, who is convinced of the inevitable changes to the legal profession and the role of lawyers - predicting “the end of lawyers” especially due to the unaffordability of legal services.
However, Satyajit Das emphasizes the innately personal nature of legal services, which requires a bespoke and complex response - much like with health professionals, requiring various important interpersonal skills such as empathy in each unique dispute. This is reflected in the five skills that LexisNexis has identified as important for future lawyers; The ability to “think like a business person”, “acquire soft skills, emotional intelligence and technology skills”, “communicate your knowledge, ideas and value”, “develop a personal brand and profile and form strong relationships with clients and employers”.
Additionally, new roles in the legal space have emerged in the United States due to the centralisation of responsibility for legal operations, demonstrating a “paradigm shift” in corporate legal services. Some of these roles are “Legal Operations and Budget Manager”, “Director of Legal Administration” and “Legal Operations Analyst”. Relying solely on technology such as predictive judicial analytics then, it seems, is insufficient to fulfil the role of a future lawyer.
Thus, it is important for law schools to extensively cover content on technology and diverse legal roles going forward, equipping future lawyers with the knowledge and skills necessary to remain relevant to their clients. This will ensure the optimal use of predictive judicial analytics in a controlled capacity - possibly in ways that may benefit broader society, rather than the commercial value provided exclusively to firms and clients. For instance, if predictive judicial analytics is used in the self-analysis of judges, such that they can correct their own biases in judicial decision-making, this would ensure the law is applied more uniformly across cases.
Furthermore, an ethical guideline could be established in Australia, to aid in this endeavour. Pamela Stewart and Anita Stuhmcke, law professors at the University of Technology Sydney, suggest that there should be regulation on how such technology is used, including preventing it from providing “predictive or opinion-based inferences”, and ensuring that inferences are verified by the user. This would enhance the robustness of and confidence in our judiciary.
Insofar as we avoid the ethical problems that disregard explicit reasoning and causal inferences, and the perpetuation of unconscious biases which currently exist in judicial decision-making, the following will hold true: predictive judicial analytics has the potential to deeply invigorate the fundamental strengths of our justice system. So long as it is not swayed by corporate instincts and developed carefully for select purposes within the judiciary, it can be used for good.