Bloomberg Law
Jan. 14, 2016, 2:17 PM UTC

To Achieve Real Innovation, A.I. Must Survive the Hype

J. Stephen Poor

Editor’s Note: This post is written by the chairman of Seyfarth Shaw.

By Stephen Poor, Chairman, Seyfarth Shaw

Over the holidays, I learned that my daughter and son-in-law recently disconnected their cable in favor of a combination of streaming services. It seems like only yesterday that most of us drove to our local Blockbuster to rent movies. Then Netflix came along and, in what seems now like the blink of an eye, poof – no more Blockbuster.

The legal industry is rather preoccupied with the question of whether it is likely to go the way of Blockbuster. Phrased in attention-grabbing terms, we debate the impending death of the industry. Typically, technology features heavily in these discussions. Watson . Ross . Riverview’s Kim . The arrival of each new platform touting machine learning capabilities injects new life – and new levels of hyperbole – into the ongoing debate.

Amidst those discussions, a recent study by KPMG on the current state of eDiscovery and data management provides some much-needed perspective on the realities of tech-driven change in the legal industry. While eDiscovery cost is decidedly not a new issue, it is still a source of real and increasing pain for a significant number of companies. Although review costs have fallen on a per-unit basis, the continuing increase in volume and types of data keeps this problem relevant. Meanwhile, the cost drivers are well-understood. In KPMG’s survey of global survey of general counsel, compliance and risk officers, 43 percent of respondents characterized manual review as “extremely costly.” Further, a 2012 report by the RAND Institute for Civil Justice indicated that manual review accounts for 73 percent of all eDiscovery costs.

Despite all this, only 13 percent of KPMG respondents indicate making investments necessary to support technology-assisted review and data analytics. Why this gap? On one hand, we hear it argued that artificial intelligence will wholly displace lawyers; on the other, at least according to KPMG, little material change has taken root in how we use technology to conduct document review.

Hype – and its unintended consequences – may be one reason. Commenting on the “striking” level of inertia in the eDiscovery study, KPMG notes that technology-assisted review has “clearly failed to live up to expectations and its value is questioned.” By setting unrealistic expectations around the potential impact of an emerging technology, we might doom it to failure. Despite the fact that eDiscovery was once widely touted as the most likely starting point for the A.I. revolution in legal, eDiscovery technology languishes for the now in a “ trough of disillusionment.

Asking whether artificial intelligence will replace lawyers exemplifies how the overheated rhetoric of hype turns out to be not only reductive but also ultimately unhelpful. McKinsey’s recent study on work automation characterize this “focus on occupations” as “misleading.” Rather than the potential of technology to displace entire jobs, the McKinsey report places more emphasis on capacity of technology to automatetasks.

McKinsey estimates that up to 45 percent of current workplace tasks could be automated with current technologies, and their published data indicates that about 23 percent of time now spent by lawyers could be automated. This, of course, leaves out the question of readiness. Before any of this can actually happen in the real world, the legal industry must first contend with the need for “entire business processes to be transformed, and jobs performed by people to be redefined.”

Too often, the legal industry fails to do this difficult underlying work: articulating the logic of our decisions, mapping the processes that comprise the delivery of services, and identifying the right bottlenecks to automate. Instead, we want the Easy Button. When this or that technology fails to deliver on this fantasy, it feeds the vicious cycle of inflated expectations and ultimate disappointment. In other words, the legal industry has such low levels of tolerance for “failure” that the inability of emerging technology to deliver on hyped levels of expectations can sound its death knell.

Particularly because emerging technologies typically need iterative experimentation with real-world users to test for product-market fit, the resistance of lawyers creates a particularly harsh environment for technology advances to survive to maturity. The applications of artificial intelligence currently available in our industry are still extremely immature – and likely to stay that way unless they are refined and iterated out in the wild.

The industry needs to take the A.I. rhetoric down a notch. The practitioner’s fears of complete displacement are understandable but facially silly. To think more clearly and productively about the future of law, we need to shift our focus to the client. Rather than asking how artificial intelligence will affect the role of lawyers, we might benefit from asking how that technology might improve the experience of clients.

The truth is that Blockbuster wasn’t disrupted by video-on-demand technology. They were disrupted by a better business model built on a deeper understanding of the customer. Netflix gained a foothold in the market by offering subscription pricing to customers fed up with Blockbuster’s due dates and late fees. Long before Netflix had the massive infrastructure and customer data that now enables instant streaming and intense personalization , the company had an edge over a massive incumbent: Netflix understood the need to provide a better customer experience. Our industry needs to develop the same understanding.

For more essays from Stephen Poor (@stephen_poor) and Seyfarth on change in the legal industry, visit Rethink the Practice.

Learn more about Bloomberg Law or Log In to keep reading:

Learn About Bloomberg Law

AI-powered legal analytics, workflow tools and premium legal & business news.

Already a subscriber?

Log in to keep reading or access research tools.