[EDRM Editor’s Note: The opinions and positions are those of Craig Ball. This article is republished with permission and was first published on August 1, 2024.]

Yesterday, I sought to defend the value of my law school course on E-Discovery & Digital Evidence to a law Dean who readily conceded that she didn’t know what e-discovery was or why it would be an important thing for lawyers to understand. It was a bracing experience.

My métier has always been litigation, to the point that everyone I work with sits in and around trial practice. My close colleagues recognize that 90% of what trial lawyers do is geared to discovery and motion practice, and much of that motion practice is prompted by discovery disputes. So, hearing how a tax lawyer and academic viewed litigation was eye-opening, and troubling to the extent it impacts what’s taught to new lawyers.

Do you agree about the centrality of discovery to litigation, Dear Reader?

The Dean shared her sense that discovery is being replaced by AI and that “soon AI will handle the production of relevant information instead of lawyers.” I replied that I expected the review phase to be abetted or supplanted by AI in the near term—that’s here—but it would be some time before all the tasks that come before review would be fully AI-enabled.

The idea that there are crucial tasks requiring lawyer intervention before review was surprising to her. For those who don’t manage electronic discovery day-to-day, electronically stored information seems to magically appear in review tools. But for e-discovery folks, the march through identification, preservation, collection and processing is our path, and we know that no one, and no AI, can undertake an assessment of the evidence without facing the data.

You’ve got to face the evidence to assess the evidence.

That’s axiomatic; but it’s downplayed by those shouting “AI! AI!” As they say in these parts, “you’ve got to put the hay down where the goats can get it.” Until AI is embedded in everything, until AI faces the data in every phone, cloud repository, storage medium and database in ways that support discovery, the goats can’t get to the hay.

The evidence in our cases is not a “collection” until it’s collected. That doesn’t necessarily mean a copy must be made to isolate data of interest, but that remains the prevailing way that a discrete assemblage of potentially responsive ESI is marshaled before it is processed for search and review. Not until that occurs does the evidence face human or AI review.

We must never forget that the goal is to isolate information that is relevant, responsive and not privileged. Perhaps e-discovery teams should assemble each morning and ritualistically chant in unison: RELEVANT! RESPONSIVE! NON-PRIVILEGED! Hey! It couldn’t hurt!

RELEVANT: The touchstone for what makes evidence probative and material; having some tendency to make a fact of consequence to the case more or less probable.

RESPONSIVE: The duty to disclose presupposes a proper request for production or the obligation to comply with rules requiring initial disclosure like Fed. R. Civ. P. Rule 26(a)(1)(A).

NON-PRIVILEGED: Our law recognizes that certain relevant and responsive information may nonetheless be logged and withheld from production if it serves to reveal information privileged from being disclosed, such as confidential attorney-client communications, attorney work product or other sensitive or protected categories.

AI Large Language Models are proving extraordinarily adept at surfacing relevant, responsive and privileged content from large collections of data when a skillfully prompted AI tool faces the evidence. If your firm’s business model hinges on lawyers billing time for document review of information that may or may not be relevant, your model is in peril. But let’s be honest; it’s been imperiled for a long time.

Currently, the sea change in e-discovery wrought by AI is largely confined to the review phase, which looms large in lawyer psyches (and larger in the minds of certain law deans) but it is not the whole of e-discovery. The hardest part—the human part— comes before review. Until nettlesome issues of privacy, cost and feasibility are sorted, the meatware remains key to a functional e-discovery process, and the skills required to initiate, oversee and manage the e-discovery workflow are essential to our system of justice.

What do you think? What’s your projected timetable for massive change to the left side of the EDRM?

Read the original release here.