While AI in law is no longer in its infancy, its acceptance and application have yet to be seamlessly integrated across a firm or enterprise, and it is not yet a pervasive component of legal operations. Its use is localized as discrete applications or services, each with a unique use case, behavior and structured results. This isolation will continue until frameworks are implemented to coordinate and manage information at an infrastructure level with integration between AI tools and current information systems to access case decisions, code, and other relevant internal and external data sets. The implementation of these discrete AI tools and services will require specialized skill sets and expertise to realize and exploit AI’s growing capabilities in the practice of law.
Legal AI: Follow the Money
To understand AI’s ascendancy, consider the growth of computing power over the past 20 years and the investment behind that. (Think of your beloved WordPerfect 5.2 machine and the small shiny slab of concealed AI you carry around now.)
For example, the explosive growth in the legal tech start-up market. Forbes reports that in 2016, there was $224 million invested and at the close of 2018, there was $1.6 billion of investment capital put forward. This investment will push deeply into both the operational and retail markets for legal services. Combine this investment with the vast amount of data generated by the profession, the public and the courts that serves as training material for AI researchers and developers, and this growth can be better appreciated.
A Thomson Reuters survey of corporate counsel concludes that AI tools will be in routine use within 10 years. A separate but complementary analysis by Deloitte predicts that within that same 10-year period, 100,000 legal sector jobs will be automated, with AI-based tools and services providing the foundation for the changes in the market.
Looking at AI through a use-case lens leaves an important open question: Can AI perform as an attorney? AI systems can apply strict, rules-based logic, learn convoluted workflows, weigh probabilities based upon a historical data set, and recognize patterns and extrapolate outcomes. These are skills that attorneys use each day but are becoming an expected set of features in AI services. The key differences between human analytic thought and constructed analytic capabilities exist at the edges—where things get fuzzy, and abstract, intuitive or creative reasoning comes forward. So, while an AI cannot replace an attorney, it can supplement areas where an attorney may not be strongest. A shining example of this are the research and tools that have been emerging in the areas of legal analytics.
“Legal analytics is the implementation of established data analysis methodologies, using supporting tools, to common data sets within the field of law to improve efficiency, gain insight and realize greater utility from the available data.” —Google Assistant
Application of AI Through Discovery and Investigations
AI e-discovery tools began as a “simple” search tool used to sift structured data to locate specific terms or keywords in a given data set and identify potential duplicates or relationships. It evolved into technology-assisted review (TAR) and predictive coding. TAR 1.0 used human guidance and static data models during a training stage that prepared it to support reviews. TAR 2.0 implemented continuous active learning to monitor a review from the outset to refine its understanding of the data, the demonstrated intent of a reviewer’s decisions, and from that provide insights about the relevancy of a document. Those insights are then returned to the monitoring process as ongoing refinement.
The opposite side of this pattern matching and relevance evaluation is the ability to recognize broken patterns such as unexpected shifts. Gaps in sequence or variations in frequency, changes in word use behavior, and other indicators of omission or deceit can be teased out and appraised. Finding these anomalous breaks or changes in tone that can shift the definition of what information may be relevant in an investigation. AI-based tools are being created to recognize these more nuanced classifications of information, recognizing inferences that would be missed in a human review.
AI tools take these patterns, whole and broken, word use and context, and inclusions and omissions to isolate inferences for additional investigation. Companies such as Keen Corp. have developed technology capable of assessing the overall morale or intent of an organization by the analysis of emails, while researchers at Florida State University have built a framework capable of detecting deception in text messages with an 85 percent accuracy rate. The ability to conduct a search, logically aggregate and present data, provide new types of context, and then flag something as tenuous as deceptions, intent or mood defines the progression of AI analytics in investigations, discovery and trial prep.
These capabilities are being tested against the steadily rising tide of new data sources. People have adapted to information created by the formal use of tools such as desktop or mobile applications that generate content in highly structured formats, making their assimilation and assessment a straightforward task. The courts and legal professionals are still learning to manage more dynamic data streams created by casual data creation tools such as social networks and texting, as well as automated systems such as GPS tagging and internet of things-based devices. The convergence of these classes of data creates a chaos of information from which litigation professionals are expected to collect, tame and derive relevant analysis.
AI tools can quickly sort and organize this chaos and then provide detailed analytics to better understand it. Analytics are often presented to legal professionals as visualizations that create intuitive understanding of the relationships between abstract data sets. This allows an experienced attorney or legal support professional to track threads of relevant information across the identified and collected sources. The thread may start as a document that is then referenced in an email which may generate texts that may need to be analyzed in the context of a video deposition. At each stage, meanings and intent may change, the language will shift to accommodate a given culture, and context may evolve. This is a prime example of the reach and analysis that AI can bring to a legal matter.
Litigation Forecasting: Looking Outward and Inward
A growing but controversial area of AI-driven analytics is litigation forecasting. With publicly accessible case law, supporting documents and other tangent materials being published online, AI tools have access to an incredibly rich set of data to analyze and use as a foundation for predicting potential outcomes in a given type of litigation. With these predictions also comes the model of how these predictions were made and, perhaps, a path to re-creating successful outcomes on similar issues going forward.
In the past, an attorney might have looked at a judge’s reported decisions. AI goes far beyond this typical framework, conducting a detailed review of a jurist’s decision history, how they ruled on types of motions, their word use and the tone of their writing to develop a profile of the jurist, within the lens of a specific type of case. This analysis could be supplemented by similar drill downs on opposing counsel and then extended further outward to the evaluation of expert witnesses and the content they have produced. Given these threads, and a growing skill at more nuanced evaluations, AI-driven research is then potentially capable of revealing unconscious bias, falling back on rote decisions, or other tendencies that can be assessed and then applied to litigation strategy.
Courts are challenged by this research and the apparent “moneyballing” of judicial viewpoints. There are those who say this approach can ease the congestion in the courts by objectively determining the merits of going to trial, possibly encouraging settlements. Others say litigation forecasting provides attorneys the ability to develop a legal strategy tailored to match a particular jurist, resulting in gamification of the judicial system. Both points of view are based on a basic belief that people who have a grasp of a potential outcome, and the steps needed to influence elements to create that outcome, will take those steps. An extreme response to this new reality is France’s implementation of a law banning the publication of judicial analytics.
Litigation analytics is also useful in the evaluation of a firm in comparison with its competition. Applying these tools to assess the performance of a firm and then to other firms practicing in the same areas can provide an objective look at the firm and provide guidance in developing new approaches for client outreach and engagement. A further extension of this type of business intelligence is the evaluation of internal operations. Understanding resources, time lines and staffing applied to a matter and being able to use tools to compare them objectively to similar matters within a firm can assist in the refinement of the practice, lead to stronger controls on pricing and ensure consistent legal project management processes.
Legal operations will continue to be supplemented by various types of AI services going forward. The continued growth of AI’s flexibility in the acquisition and analysis of data, driven at a speed that cannot be matched at the human level, positions AI applications as a perfect adjunct to legal professionals. The key word here is supplemented, as the ultimate success of AI applications will be driven by the attorney. As people learn to evaluate the work done by AI and provide their feedback to the system, the overall consistency of the work performed can improve. These tools require innovative thinking by firms and attorneys about their relationships to their own processes, their clients’ data and to the growing universe of data sources being created each year. For these practitioners, AI-driven legal analytics will be an invaluable addition for clients and for the improvement of their practice.