Peer to Peer Magazine

P E E R T O P E E R : I L T A ' S Q U A R T E R L Y M A G A Z I N E | F A L L 2 0 1 9 19 legal domain especially, much of that data is extremely sensitive. AI can also depend on using customer data to learn and to create "virtuous" feedback loops. Legal organizations that want to leverage the power of AI cannot simply trust third- party vendors to meet their contractual and regulatory compliance obligations. Data that moves beyond the customer's firewall must be properly anonymized and cleaned, and that work takes real expertise. Technolo providers with decades of experience and authority handling customer data will have a big advantage in the marketplace because the protocols that protect private data are quite complex and constantly evolving. Data integrity and algorithmic transparency is another area where we need to be vigilant. As products become more sophisticated, users understand less about how those products work, which can pose a serious problem. For example, if you apply analytics to data that contains lots of errors or "noise," has not been properly cleaned and structured so it can be searched effectively, or is not comprehensive in spite of claims to the contrary, you will get very unreliable results. If you make a high-stakes legal decision based on those results, you increase the risk and exposure for your clients. Normalizing and cleaning "dirty" data is time-consuming and resource-intensive, but it is essential. Legal organizations who want to adopt AI tools need to know more about how different vendors approach this task. The industry also needs more side-by- side, "apples to apples" testing of analytics solutions where we can closely compare results to identical research questions. We need transparency from vendors about the underlying data they rely on, and about the data they use for algorithm training. The legal community should insist on public benchmarking of AI tools – not just for the sake of customers, but also to advance the technolo. As the use of machine learning and natural language understanding becomes more widespread in the legal world, we need to advocate for regular testing of solutions in the marketplace using benchmarks like the General Language Understanding also use machines to get "more like this" recommendations and actually receive extremely relevant answers – not just links to documents – to their research questions. Other areas in the legal space where AI is currently gaining traction include back office applications that help firms with timekeeping, billing and document classification; project management systems that can identify efficiencies based on variables like workflow, the current availability of staff and billing rates; and legislative analysis, which can track and predict the progress of pending legislation and suggest next steps. At this point, patent law has probably generated more AI-related innovation than any other practice area, mostly because of the sheer volume of data involved in patent litigation and the highly specialized knowledge required in patent practice —which, by the way, "smart" AI technologies are able to very effectively help. Contract management is another area in which AI is already saving corporations millions by enabling workflow efficiencies and streamlining procurement and the renegotiation of contracts – tedious processes that can get very expensive in a hurry when performed by billable (human) attorneys. We've already touched on the value of legal analytics for litigation, which is providing legal teams with data-based insights that inform key decisions and determinations, including: which claims to file; the value of a case; the likely outcome if litigation is pursued; the likelihood of success of specific motions and arguments before a specific judge; the strengths and weaknesses of specific expert witnesses; and much more. We've only begun to tap the potential of use cases for analytics. The challenges and responsibilities of working with data With powerful technologies come serious responsibilities. To be effective, AI requires large amounts of the right data. In the Evaluation (GLUE). Vendors that are making long-term investments in AI should have a comprehensive set of internal benchmarking and testing procedures in place, including protocols for human raters, who are indispensable for performing more nuanced assessments such as relevance in search results or question answering. Future opportunities There are many areas in which AI is poised to break through in the legal domain. High quality legal analytics is now a proven technolo that saves money and gives practitioners a significant competitive advantage over those without it. Natural language searching and querying will continue to improve in the near future. Intelligent interactive agents that can answer your natural language questions and even accomplish some of your basic tasks is one of the most exciting developments on the horizon. These will not resemble the rules-based chatbots we've all encountered on the web, but will be truly intelligent agents that can understand legal language, personalize your experience based on your behavior over time, decipher your intention, and provide useful professional guidance and recommendations to make you more productive and effective in your work. The result will be a more human interaction between the user and the service. Whatever the application, AI technologies in the legal domain should always be use-case driven. We should never apply these tools "just because we can." We apply these technologies because they are well-suited to a specific problem. Properly conceived and deployed, AI applications will bring enormous benefits to lawyers and their clients, ranging from cost-efficiency and error reduction to more intuitive workflows and better legal outcomes. Ultimately, these applications will make the law more accessible and understandable by all, and help us create a more just world. ILTA

• Cover