Applying AI in Privilege Review
- 7 min read
The use of artificial intelligence (AI) continues to expand in the legal industry. From legal research search engines to technology assisted review (TAR) to machine learning for contracts review, this technology is rapidly evolving. As the technology changes, lawyers discover new ways they can integrate it into their practice or business operations. The next area that AI is revolutionizing is privilege review, which lawyers are required to complete during litigation or any other investigation which involves confidential material. Privilege is the heart of the attorney-client relationship and must always be maintained, even after resolving a matter. Because of the delicacy involved in privilege, lawyers have refrained from using advanced technology and often depend on identifying privileged documents by running targeted searches then deploying experienced manual reviewers. However, this method can often miss privileged documents due to the inherent risks of a manual review, create excess time and money spent on privileged tasks, and generate inconsistent results. AI continues to advance and prove to be superior to manual review. As these AI-enabled options become more normalized, lawyers need to learn why and how they should add AI to their regular set of tools used for privilege review.
Why AI for Privilege Review?
Recent history has demonstrated that the benefits of using AI in legal practice greatly outweigh any potential risks, such as the use of TAR, which is a court approved use of AI that saves significant time and expense. The same will ring true when looking back at the evolution of AI in privilege review. While keeping privileged information confidential is crucial, this is also a difficult and multi-faceted process. What qualifies as privilege and requires redacting before disclosure is often not straightforward. This is where exploring the benefits AI offers becomes vital. The goal of this technology is to decrease the stack of documents that manual reviewers ultimately need to examine. Letting AI technology prioritize highly suspect documents and weed out unprivileged material is the most efficient way to obtain this goal.
A major reason lawyers should use AI for privilege review is because this technology has superior detection abilities when analyzing important conversations. It can locate privileged data that manual reviewers miss, which happens frequently with items like email chains in which it is not immediately evident that there is relevant data embedded. One other common example is when legal assistants, paralegals, or law clerks correspond on behalf of an attorney. AI programs have capabilities to analyze language and company dynamics in order to determine a person’s actual role and reason for communicating, which can help locate missed privileged data. This feature also is beneficial for finding name variations or missing references that the person is a lawyer.
AI can also eliminate much of the unnecessary data that keyword searches generate and catch documents that would have slipped through the cracks. While contextual language can yield a large amount of unprivileged data from a keyword search, reviewers still need to pull out the privileged communication. Some AI programs can now identify semantics, which helps solve this dilemma by reducing how much privileged data reviewers wrongfully categorize. The technology can identify and pull only the documents using a certain word or phrase when the context infers privilege. It becomes a tremendous help when clearing out the confidentiality disclaimers generally found at the bottom of lawyer emails. Many conversations containing such disclaimers do not actually contain any privileged communication, thus it can significantly reduce review time and costs, while also consistently generating the desired results to include in an automated privilege log. As a result, lawyers will end up spending the bulk of their time on reviewing a more targeted data set and making any necessary tweaks to the privilege log.
Different Approaches to AI-Based Privilege Review
After learning about the benefits, the question remains: how should lawyers add this tool to their practice? As with any AI technology, the initial time investment is key. Lawyers looking to use AI for privilege will need to stabilize a prebuilt model in order to make it more effective. The process consists of manually reviewing a large data set and training the program on what is relevant. Then, the program will know what data to target and what to eliminate. Additionally, periodic training will be necessary to account for new privileged material, which could include things like language specific to a case or new lawyers.
Lawyers also need to consider the purpose for deploying this technology. The first way AI can help with privilege review is by improving quality control. It is a larger misstep to accidentally disclose privileged data than it is to label nonprivileged data incorrectly. The act of breaking privilege cannot be undone. Another approach is using AI to reduce the risk surrounding privilege review, which goes hand in hand with quality control. AI historically has reduced the risk with any review project, as human error is often the cause of most review problems.
Lawyers can also use AI to significantly control the costs that accompany privilege review. Even though the initial investment may be steep, over time, the costs are cut by eliminating extensive, routine manual review. Also, since the digital world keeps growing, there will be more and more data to review. Being able to use technology to sift through electronic conversations will greatly reduce time and costs in the future. Lastly, lawyers can integrate AI into their privilege review practice to increase efficiency. This software can consistently detect data manual reviewers would miss, get the job done faster, allow lawyers to focus on high value privilege review functions, and is defensible.
While these examples are all sound ways to use AI for creating privilege logs, in order to get the best results, it is important to still incorporate human elements in this process. Conducting simple domain and attorney name searches in conjunction with using AI software is an excellent way to achieve an efficient product. Lawyers have been using this hybrid approach to AI for years, like when users train TAR software in order to get the most responsive data during discovery. Without input from skilled legal professionals, this technology will not be able to perform or provide the most efficient results. With more lawyers turning to technology to create efficiencies, expectations are this area of AI will continue to develop.
To learn more about AI, please join us for a complimentary webcast: Kick off 2021 with an AI-centric mindset within your litigation and investigation teams