The rise of artificial intelligence (AI) in legal technology began during the document review phase of electronic discovery, which accounts, on average, for over 70 percent of the costs associated with pre-trial discovery. Technologies like predictive coding, in which algorithms learned from “seed sets” of documents coded by human reviewers, provided fast, accurate and less expensive means for legal teams to work through a vast body of potentially relevant electronically stored information (ESI). In the past decade, AI has begun addressing tasks like contract review and management and case law review and has begun predicting case outcomes.
Although AI has been used in the review phase of e-discovery for approximately a decade, its current integration into document review has become simpler and more elegant. In the past, document review technology required seed sets and users who could define the parameters for relevance. Today, deep learning algorithms, which simulate the human brain by combining several layers of neural networks, can operate in the background, observing as human attorneys review documents, learning the criteria that make a document relevant to a particular matter.
AI can apply data mining techniques to vast bodies of data, not just ESI but also custodian identities and relationships, before any data has been collected. By examining relationships between concepts and existing custodians, AI can suggest new keywords and search terms to find relevant ESI.