Whenever the US Justice Department reviews a merger between two major companies, it often requires careful analysis of millions of electronic documents. In the past, this task has typically fallen upon contract workers, earning between $25 and $40 an hour, but in a recent case involving two beer makers, US authorities decided to replace these employees with a more automated solution: computer software.
As the Wall Street Journal reports, lawyers involved in a proposed merger between Anheuser-Busch InBev and Grupo Modelo were recently given permission to use so-called predictive coding software to sift through the millions of strategic plans, pricing data, and other corporate documents from Constellation — a spirits company that had been selected to buy assets in the deal. Lawyers fed some of the documents into the program, developed by kCura, and trained the software to recognize relevant materials. Once US authorities and Constellation representatives were convinced of its ability to do so, they allowed it to take over the process.
Predictive coding offers significant savings
Ultimately, lawyers for Constellation and Crown Imports (a joint venture between Modelo and Constellation) ended up disclosing hundreds of thousands of documents to the Justice Department, which dropped its challenge to the merger after all parties agreed to sell Modelo's US operations to Constellation. They also saved a lot of money. "Something that would easily cost three, four, five million dollars, you can do in the range of one to two," Warren Rosborough, a lawyer who represented Constellation and Crown Imports, told the Journal.
Thus far, however, there aren't any hard regulations on the use of predictive coding across other agencies. The Securities and Exchange Commission doesn't bar the practice, though a spokesperson declined to tell the Journal how often it is used. A spokesperson for the Federal Trade Commission said the software is used on a case-by-case basis. The Justice Department says it has used similar software to review internal documents, though it declined to comment on this particular case.
Carl Walworth, a partner at Paul Hastings LLP, expressed skepticism about a computer's ability to completely replace lawyers, noting that a thorough review of documents is often critical to familiarizing oneself with a given case. "More often than not, you're trying to learn your case through the documents," Walworth told the Journal, "and how will we substitute that function of learning from the documents when you're using predictive coding?"