The day when robots start doing legal work might come sooner than you think. Indeed, courts and litigants across the country appear to have a laserbeam focus on the newest entrant to the E-Discovery scene: computer assisted review.
In a nutshell, computer assisted review (like Predictive Coding®) is an automated doc review process. In lieu of the traditional “linear review” that most litigators are used to, computer assisted review employs algorithms and other formulations to review massive swathes of documents in just a fraction of the time. But, how does a computer know what is relevant and what is not? What is hot and what is not? Here’s where the lawyers do come in: they train the computers by reviewing a small segment of the corpus themselves. The computers then crunch the data, and begin their own review of the documents based upon the sample set.
Needless to say, the efficacy of technology assisted review could change the landscape of E-Discovery. However, both courts and skeptics have helped to slow the tidal wave created by the marketing of savvy E-Discovery companies who are chomping at the bit to get in on the next big mega-case. Although some courts have condoned the use of technology assisted review for various portions of the discovery process (see, e.g., Da Silva Moore v. Public Groups & MSL Group, Case No. 11-CIV-1279), very few have actually required it—perhaps with good reason.
While many have continued to expound on the virtues of computer assisted doc review, no one can deny that there are certain associated risks. For instance, the computer’s review will only be as good as the review established by the “training sample” conducted by the original attorneys’ eyes. A poor sample can result in both overproduction of non-responsive documents and the under-production of relevant information. Computer assisted review may also have more difficulty making correct privilege calls, which amplifies the risk of producing privileged information. In addition, the cost of technology assisted review may make it prohibitive in smaller to medium sized cases, and courts will have to grapple with developing a consistent jurisprudence on these ESI issues notwithstanding.
These are just some of the concerns that keep both practitioners and the bench on the proverbial fence. Thus, although the role of technology assisted review in any given case remains murky, one thing remains pretty clear: it doesn’t look like either technology assisted review or the good old days of human linear review are going anywhere anytime soon.