Use Cases for AI in the Legal Function

  • September 13, 2024

Nearly 75% of in-house legal department leaders believe that generative artificial intelligence use will lead to more work being completed in-house, thereby supporting increased efficiency and reduced costs. While actual implementation of the technology has not yet moved past planning and proof of concept phases within most legal departments, there is a healthy interest in investing in generative AI to support efficiencies across legal functions.

As legal departments work through proofs of concept and plan their technology investments, leaders must remember the importance of responsible generative AI use supported by experts. Additionally, in the current iteration, generative AI is a complement, not a replacement, for human input and intellect. Yes, the technology has great potential, however, there’s currently more opportunity in the middle ground, where certain generative AI models can be applied to tasks that will supplement human efficiency. Acting as the human in the loop at all times for quality control and oversight, and in partnership with technologists who understand the pitfalls and limitations, general counsel and their teams can begin to experiment with generative AI for the following use cases:

  • Document summarization. In an e-discovery context, document summarization is where responsible use of generative AI currently shows the greatest potential in the legal function. In the 2024 edition of The General Counsel Report, 92% of general counsel reported they were at least somewhat comfortable using AI in e-discovery, and over half were either very comfortable or extremely comfortable using AI in e-discovery. In testing within FTI Technology, e-discovery experts have found that large language models may support faster and more efficient early case assessment when used to summarize the contents of key documents, provide an overview of a case, or build reports of themes and patterns within document sets, to support development of case strategy and subsequent steps in the discovery process.

    Summarization also offers the potential to help counsel create chronologies, indexes, privilege logs, or summary reports for hot documents and witness statements, tasks which are time-intensive when handled without effective technology. An important note to remember though, is that success largely depends on what models are used, how they are applied and the level of oversight given to their inputs and outputs.
  • Personally identifiable information identification. Generative AI may also support investigative work in the wake of a data breach or other data-privacy related matter. Similar to the way some models can be used for document summarization, generative AI tools may also be applied with human oversight to interpret a document or set of documents and identify and extract any information the model has been trained to recognize as personally identifiable or sensitive. This can help legal teams determine the scope and scale of information that may have been exposed in a data breach or to remove sensitive information from documents before it is produced to courts, regulators or opposing parties during the course of litigation or an investigation.
  • Data governance. Analytics and data visualizations are already available to help review large data sets and algorithms for instances of bias. With human guidance, AI can enhance quality control over datasets being used to train AI models, such as by analyzing databases and identifying missing values, errors or duplicative entries to ensure completeness and accuracy. This functionality can include data cleansing, normalization and standardization, allowing generative AI to automatically correct formatting and remove erroneous characters.
  • Compliance. In a compliance context, AI can be trained to read and analyze complex and lengthy third-party contracts and insurance policies to identify inconsistencies, pricing anomalies and exposure to legal risks as part of third-party risk management efforts. AI-driven regulation monitoring can support alignment of internal company policies with applicable regulatory and industry standards. By scanning and mapping large volumes of regulations to specific internal controls and procedures, AI can highlight areas of non-compliance and potential impacts, and suggest necessary adjustments for legal teams to evaluate and act upon. Similarly, compliance and legal teams can use generative AI to summarize complex and multijurisdictional regulations in support of gap assessments and compliance analysis across different requirements and standards. Again, these uses are intended to be in tandem with human oversight and success will depend on applying the right tools in the right way.
  • Contract solutions. Efficient creation and analysis of contracts is critical to meeting regulatory requirements, recognizing business value and reducing legal risk, yet many steps across the contract lifecycle can be time and resource intensive. Many legal teams are considering how generative AI can help with streamlining their contracting processes. In addition to supporting contract creation, AI can also help to summarize and extract key information from contracts to support human review when contracts must be analyzed during corporate transactions, restructuring, litigation or other legal matters.

Organizations looking to leverage AI in their legal department workflows should first consider the spectrum of predictive analytics and machine learning tools that have already been widely tested, proven and accepted for these use cases. Many legal departments have already been using various forms of AI for many years, and there are tremendous potential time and cost savings available with analytics that have already been tested and proven in the real world. These should be integrated into standard workflows before generative AI is adopted.

As the technology’s capabilities mature and become more accessible, there will be a range of opportunities for legal departments to evaluate and deploy new tools and to build upon analytics and machine learning workflows already available within their current processes. Close attention to responsible use, defensibility, accuracy and consistency will be paramount, as will evaluation of benchmarks that measure whether AI implementations are meeting cost and performance expectations.

Related Insights

Published

September 13, 2024

temp Key Contacts

Wendy King

Senior Managing Director

Sign up to get access to FTI Consulting Insights

Subscribe