Will Predictive Coding Spell Trouble for Doc Review Attorneys?

In the past few years, thousands of attorneys have relied on temporary document review jobs to help them survive financially during layoffs, law firm hiring cutbacks, and slow starts for solo practices. As a document reviewer, attorneys assist the ever-growing world of e-discovery by sifting through millions of documents, organizing them by responsiveness, privilege and issue tags. Traditionally, manual review has been the standard in document production requests, but the introduction of predictive-coding software products could change that trend.


In a recent study released in the Richmond Journal of Law and Technology, attorney Maura R. Grossman and computer science professor Gordon V. Cormack, discovered that software using predictive-coding can do a more efficient and accurate job of tagging documents than humans. While some firms may be hesitant to embrace new technology, recent malpractice lawsuits regarding the diligence of human reviewers may make firms think twice. With predictive-coding, the software delivers results based on keywords that are set by seasoned attorneys, and then the software can be tweaked to improve accuracy over time.

By utilizing predictive coding technology, law firms can prioritize documents by subject matter and responsiveness, and decrease the amount of time reviewers spend coding documents from scratch. That means firms can get more done in less time with fewer human reviewers. Plus, they can worry less about human error due to boredom and fatigue after long stretches of reviewing similar documents all day.

Fortunately for many doc review attorneys, software hasn’t completely eliminated the need for human reviewers, and predictive coding raises at least two important issues for attorneys: (1) their legal obligations to conduct a reasonable search for responsive documents under federal discovery rules, and (2) their ethical obligation to safeguard a client’s privileged information. As with any new technology, courts and bar associations will have to address these issues before software can significantly replace human reviewers. In the meantime, I would like to hear from you. Who can do a better job: software programs tweaked by one attorney or a room full of manual reviewers using traditional coding platforms?

  • Lawrence

    Kate – interesting article. But I think the bigger problem for attorneys or at least attorneys that practice in the U.S. is that our jobs are going overseas. Document review has been outsourced to India for a while now. Next up if not already started will be legal research and drafting. This type of outsourcing should not be permitted. It’s not good for U.S. lawyers or our clients to have untrained and nonlicensed practitioners doing our legal work. Our clients will suffer by receiving a poor workproduct that could harm their case and US attorneys will suffer from the lack of jobs.

  • George Diaz

    Kate, the future is now. Data will continue to grow exponentially. There is no way to review that much information unless you throw more bodies at the data. This means more money during litigation. It is unsustainable.

    Predictive coding uses math and science so the computer learns from the expert making decisions on responsive / non-responsive documents. The process is fully defensible.

    The choice is actually which technology does this best job. OrcaTec is worth taking a look at because they take advantage of newer code and high performance hardware.

    In any event, the evolution of document review is now at the point where we will still need expert reviewers, just not that many of them.