Types of AI Tools
While the promise of utilizing AI in e-discovery has been realized in some areas, it has fallen short in others. Lately, technology-assisted review (TAR) tools have achieved far more consistent use than any other AI application. Supporting this assertion is a recent study from the Association of Certified E-Discovery Specialists (ACEDS) and IPro reporting that 81 percent of study participants had regularly used them. While this demonstrates a very high acceptance and implementation rate for TAR technologies, other AI tools, like sentiment analysis, anomaly detection, and behavior analysis, were identified in the same study as having more restricted adoption. These less-utilized AI tools, however, have the potential to significantly improve legal data analysis to drive better case outcomes for those willing to implement them. This is especially true when it comes to evaluating new communication channels and emerging business tools.
Benefits
The most effective argument for TAR adoption has focused on its use as a tool for lowering the cost of legal services for clients. After all, much has been discussed about legal document review being the most expensive part of discovery and, arguably, litigation. While all agree that AI can lower costs, the more important takeaway is its effect on quality, speed, and accuracy. For example, TAR is that it not only reduces overall spend by eliminating nonrelevant data but also increases review speed.
These arguments haven’t gained the same traction with other AI tools that perform more humanlike tasks, such as sentiment analysis—perhaps, in part, because it is harder to quantify the financial impact of these tools. Regardless, one obvious reason for this is that legal practitioners (who actually are human) don’t believe that AI can outperform humans when evaluating context, emotion, and using intuition.
Attorneys should not look at these tools as replacements for human intelligence but rather as tools to augment their abilities. Attorneys should not be afraid that these tools can be a substitute for their decision-making related to context and emotion. Rather, attorneys should see them as a way to provide otherwise-missed insights into complex data.
Examples of Underused e-Discovery AI Tools
Explore some of the more underused e-discovery AI tools and how each can provide important insights in a more efficient and cost-effective way.
Anomaly detection. Anomaly detection is a process used to identify unusual patterns, or outliers, that do not conform to expected behavior. Some outliers may include (1) email messages being forwarded to personal email accounts and (2) communications between parties frequently occurring outside of business hours.
In a typical review project, there are thousands of documents (from various custodians and containing numerous parties) that ultimately will be reviewed by multiple different attorneys. Due to the volume of documents being reviewed and the number of attorneys reviewing them, it is highly unlikely that any of the human reviewers will be able to identify usual patterns for a particular custodian or party, let alone all of the unusual patterns or outliers. Even if human reviewers were able to identify the patterns and outliers, the time it would take to do so, as well as the time it would take to put together a visual timeline or sequence of events, would not fit into the short discovery periods that attorneys have in investigations or in litigation. However, with anomaly detection, these patterns and outliers are not only quickly identified but also visually and dynamically represented on the screen.
Putting this into a real-world scenario, credit card fraud is an unfortunately familiar area to many. Anomaly detection is used to identify fraudulent transactions and can be used to flag those that deviate from normal spend patterns. Think about all the resources, time, and cost it would take for institutions to identify and manage fraudulent charges if handled by humans alone. It would be nearly impossible given the volume of daily transactions. To that point, AI does need human interaction to confirm if the transaction is truly fraudulent.
In the same way that these technologies support the banking industry, aided with human input, legal AI tools support document review with attorney input.
Sentiment analysis. Sentiment analysis, in its most basic use, identifies the tone of a communication and scores a message’s overall sentiment (i.e., if the tone of the message is more positive or more negative). Content conveying sentiment can be especially useful in providing context. It can help attorneys better understand the emotional circumstances leading to the formulation of an idea, the framing of a statement, and the initiation of an action. It can provide insights into what motivated people to act as they did. When used in early case assessment or during an investigation, the sentiment analysis score can be incorporated as an element of searching to prioritize the results through the positive or negative ranks.
Sentiment analysis currently is being used by companies to evaluate product and brand reviews. Today, people do not hesitate to write reviews on social media when they like or dislike a product. Enabling sentiment analysis, AI helps the companies categorize the various feedback messages into positive and negative ranks, which in turn can help assess product or market changes.
Combining AI tools for legal use. Even in the legal space, finding negative and positive sentiments adds greater context to the patterns found within the data that help illustrate and uncover narratives. This is why it makes sense to use anomaly detection and sentiment analysis together. When used carefully, these combined features offer powerful techniques to identify unusual behavior, potentially inappropriate conduct, and emotional exchanges, which are often tied to the underlying legal issues in internal investigations and litigation. Not only does this AI assist in the identification, but it also does so in a small fraction of the time it would take human reviewers to do the same.
Conclusion
The argument to be made for adopting e-discovery AI tools that interpret the emotion and context behind the data should be reframed from one that focuses on cost savings alone to one that centers on how these tools increase the quality and accuracy of the legal services that attorneys provide. To do that, attorneys need to not only accept but also embrace the notion that this technology can augment and enhance human decision-making abilities. No suggestion needs to be made that computer algorithms should be replacing the analysis performed—but, rather, they can support and improve it. In the end, quality and accuracy should be the central part of the argument for the adoption and implementation of these AI tools in e-discovery.
