As we enter 2024, many lawyers and much of society in general are thinking about the future of artificial intelligence (AI). Few of us had heard the phrase “generative pre-trained transformer” until the last year or so with the introduction of generative AI tools like ChatGPT. While generative AI tools should be researched, studied, and ultimately embraced if proven for their ability to assist lawyers with legal tasks, they first must be properly and accurately trained, tested, and verified.

In e-discovery, AI has been used for decades to assist with the review of electronically stored information (ESI). Clustering, concept and categorization tools, fuzzy searches, and technology-assisted review (TAR)—both 1.0 and 2.0 continuous active learning—have been used for years with great success. Search terms, however, remain the preferred method for initial culling of large collections prior to TAR or any large-scale human review.