AI is reshaping litigation, but its value should not be measured by how many lawyers it removes from a review team. A better measure is how much more lawyers can understand, and how much sooner they can understand it.
The most important document in a lawsuit rarely comes bearing a red flag. It may be a three-line email buried in a chain, an otherwise unremarkable attachment, or a text message whose significance appears only after another witness tells a different story. Finding it has traditionally depended on a lawyer working through thousands of documents one at a time, gradually learning the names, dates, and relationships of the case until they have enough context to recognize the importance of any one document. Artificial intelligence is beginning to change that process. It can help lawyers locate related communications, follow events across a collection, and recognize connections that might not become clear until weeks into the review. What it cannot do is decide what the evidence means or how it should be used. Those decisions still belong to the lawyers doing the doc review.
Some commentators(1) envision junior lawyers moving away from reviewing large document sets altogether and focusing instead on extracting strategic insights from material organized by AI. But responsiveness, privilege, confidentiality, and issue coding are only part of the work. Doc review is where lawyers learn what happened and begin developing the narrative that can win the case. The documents show who was involved, what the parties knew, and whether positions taken after a dispute arose match the contemporaneous record. They identify potential credibility problems, reveal contradictions, and expose gaps requiring further investigation.
The weakness in traditional review is that the lawyer encounters the evidence in fragments. An email reviewed on Monday may appear routine based on what the associate knows at that point. Two weeks and several thousand documents later, a spreadsheet, contract, presentation, or text message may completely change its significance. The associate did not necessarily overlook anything. The information needed to understand the email had not yet appeared. Lawyers compensate with searches, notes, chronologies, and memory, but a large collection can span thousands of conversations and years of events. By the time the significance of the earlier email becomes clear, it may be buried among thousands of documents that received the same general coding decision.
AI can bring the discovery and analysis of those documents closer together by summarizing material for faster review, keeping track of names and dates, translating foreign-language communications in real time, connecting related discussions, and organizing the evidence around the issues in the case. An email referring to a meeting can be connected to other communications about the same meeting, the people who attended it, and the decisions that followed. Conflicting descriptions of the same event can be brought together while the review is still underway. Instead of waiting until the end of the process for the larger picture to emerge, the litigation team can develop that picture as it reviews the documents.
Junior associates will still need to read the documents themselves, evaluate their context, and decide how they fit into the case. But they should be able to move more quickly from asking whether a document is responsive to asking what it proves. Does it contradict a pleading or anticipated testimony? Does it identify a witness who has not been interviewed? Where does it belong in a deposition outline or motion, and what argument does it support? If the events described in it occurred, what other documents should exist? Those questions turn review into fact development. AI may help surface potentially important evidence, but the associate must determine whether a connection is real and a document is useful. Used well, the technology does not remove the lawyer from the factual record. It helps the lawyer understand more of that record, sooner.
Like any analytical tool, AI has limitations. It may not always capture the full context of a communication, distinguish between people with similar names, interpret ambiguous language as a lawyer would, or reflect the degree of uncertainty behind an inference. In litigation, where a single date or phrase can alter the meaning of a document, a plausible summary is not enough. The source document remains the evidence, and the attorney remains responsible for verifying what it says. The American Bar Association’s Formal Opinion 512(2) explains how existing duties of competence, confidentiality, communication, supervision, candor, and reasonable fees apply to lawyers’ use of generative AI. Lawyers must understand enough about the technology to evaluate its limitations, determine what happens to client information before uploading it, supervise its use, check its output, and charge clients fairly for the work performed.
The ethical question is not simply whether lawyers may use AI. It is whether they understand the technology well enough to use it responsibly and effectively. That requires an active process in which the attorney directs the AI, evaluates its work, and verifies its conclusions. A lawyer who uploads confidential material without understanding where it goes has not solved a discovery problem. A lawyer who accepts an AI-generated analysis without checking the underlying documents has not exercised professional judgment. Competent use requires a workflow in which the technology assists the lawyer without obscuring who remains responsible for the result.
AI is reshaping litigation, but its value should not be measured by how many lawyers it removes from a review team. A better measure is how much more lawyers can understand, and how much sooner they can understand it. Speed matters because discovery is expensive and time is money, but the larger opportunity is clarity: finding the important evidence sooner, seeing how the documents fit together, and using that understanding before depositions are taken, motions are filed, and strategy becomes fixed. Junior associates will continue to review documents. The difference is that they may no longer have to wait until the end of the review to see the case hiding inside the document production.
Sources:
1) https://news.bloomberglaw.com/legal-exchange-insights-and-commentary/with-ai-junior-lawyers-will-excavate-insights-not-review-docs
2) https://www.americanbar.org/news/abanews/aba-news-archives/2024/07/aba-issues-first-ethics-guidance-ai-tools/
Join the conversation!