While the use of artificial intelligence for drafting legal pleadings is a relatively new phenomenon, the deployment of artificial intelligence in pretrial discovery of electronically stored information is not. In fact, the deployment of AI-powered technology-assisted review (TAR) tools to sift through electronically stored information is arguably the legal profession’s first use of artificial intelligence in litigation. TAR tools continue to be indispensable and widely deployed today.
Federal magistrate judge Andrew Peck’s influential 2012 opinion in Da Silva Moore v. Publicis Groupe, 287 F.R.D. 182 (S.D.N.Y. 2012), is frequently cited as early judicial acceptance of the usefulness of then-emerging technology tools for pretrial discovery. The technology under discussion in Da Silva Moore used a combination of concept-based searching (as opposed to keyword searching), statistical learning algorithms (i.e., the computer program “learns” what to look for as it encounters and categorizes documents) to finding and coding relevant documents in ESI submissions. Concept-based searching and statistical learning technologies are common components of today’s generative artificial intelligence tools.
Judge Peck’s embrace of technology-assisted review of ESI was based on his view that TAR tools were not only more efficient and cost-effective for sorting through large ESI submissions, they were also more accurate than manual reviews conducted by lawyers. A study published in the Richmond Journal of Law and Technology punctured what its authors called the “myth” that human review of electronically stored information was superior to computer-assisted tools available in 2011. “Technology-assisted review can (and does) yield more accurate results than exhaustive manual review, with much lower effort,” they concluded. Maura R. Grossman & Gordon V. Cormack, Technology-Assisted Review in E-Discovery Can Be More Effective and More Efficient Than Exhaustive Manual Review, 17 Rich. J.L. & Tech 11 (2011).
In addition to improved accuracy and cost-effectiveness, the use of technology for pretrial discovery suggests another similarity with litigators’ use of generative artificial intelligence tools – namely, their potential to empower small law firms to compete with larger law firms. TAR tools lower the cost of managing ESI in pretrial discovery, making it easier for small firms to effectively manage the large volumes of electronic data typically encountered in high-value, high-stakes litigation.
Preferences for human review over technology might seem naive to today’s litigators, but the effectiveness of TAR tools was hotly debated in 2011.
In addition to improved accuracy and cost-effectiveness, the use of technology for pretrial discovery suggests another similarity with litigators’ use of generative artificial intelligence tools – namely, their potential to empower small law firms to compete with larger law firms. TAR tools lower the cost of managing ESI in pretrial discovery, making it easier for small firms to effectively manage the large volumes of electronic data typically encountered in high-value, high-stakes litigation.
AI-Powered TAR Tools: Better and Faster Than Lawyers
So … how does artificial intelligence power modern technology-assisted review tools? For managing e-discovery obligations, artificial intelligence can:
- Prioritize and tag documents as responsive or non-responsive, learning as it proceeds through the dataset
- Categorize documents as either “privileged,” “responsive,” or “confidential” without the need for human intervention
- Identify and group related documents, and remove duplicate submissions
- Perform concept-based searches, finding documents that might not have surfaced with a traditional keyword-matching approach
- Extracts names, dates, organizations, and other key entities from large datasets
- Analyze tone, sentiment, and communication patterns in ESI submissions
- Visualize relationships, spot trends, and make connections within large datasets
- Detect and redact sensitive personal information and sensitive health information
- Translate documents and convert scanned documents into searchable text
- Summarize and extract patterns from large datasets in a manner that aids litigation strategy
Though definitely not an exhaustive list (or an endorsement), representative electronic discovery vendors claiming AI-powered features include offerings from Relativity (analytics, active learning, technology-assisted review), Everlaw (predictive coding, concept-based search, visualization tools), Logikcull (automated data processing, culling, document review), Reveal (sentiment analysis, communication pattern detection), and ZyLAB One (entity extraction, automated document classification).
Little wonder the legal profession has embraced generative artificial intelligence with so much enthusiasm. Yes, there have been artificial intelligence missteps in court, and a substantial amount of professional ethics guidance from the American Bar Association and state bar groups during recent months. But, as the experience with TAR tools for electronic discovery has shown, technology and litigation have gone together for a very long time already.
This article is the second in a series on the uses of artificial intelligence in litigation. Last week: The Promise and Perils of Using Artificial Intelligence for Legal Research.
