I, (Ain’t No) Robot: Automation’s Role in Lit Support

HaystackID Blog | April 4, 2017

by Andy Parrish, HAYSTACKID Director of Client Services

Automation is an increasingly powerful and pervasive force in the public and private sectors. Now, in a debate that has raged on for decades and spanned virtually every industry, the argument regarding whether automation and robotic technologies will replace human employees has made its way into litigation support.

Top firms and organizations are already using automation to handle many back-office tasks that were previously thought to be immune. Lit support software start-ups are heralding how automation within their products eliminates human workflow steps and simplifies discovery management. So, the question I’m hearing a lot these days is – should lit support professionals be looking for a new line of work?

Short answer: No.

Dystopia, here we come?

Consider some of these statistics and studies related to the war between automation and the human workforce:

• Recode reported that one National Bureau of Economic Research study found each industrial robot implemented in a business reduced demand for an average of six human jobs.
• The World Economic Forum cited physicist Stephen Hawking’s statement that middle class jobs and wages will be laid to waste by waves of robotics and automation deployments.
• Last July, the White House forecast 83 percent of jobs with wages fewer than $20 per hour will be at risk of elimination due to automation in the future. An estimated 31 percent of those with median wages between $20 and $40 are also expected to be at higher risk.

And for all of you who might not be at all concerned about the above, consider that in December 2016, a team of then President Barack Obama’s top economists released a landmark report titled Artificial Intelligence, Automation, and the Economy, which had this little tidbit in the Executive Summary:

“Although it is unlikely that machines will exhibit broadly applicable intelligence compared to or exceeding that of humans in the next 20 years, it is to be expected that machines will continue to reach and exceed human performance on more and more tasks.”

If that does not get you thinking of some dystopian landscape à la Kurt Vonnegut’s Player Piano, I’m not sure anything will. Those of you reassuring yourselves that you’ll simply become one of the novel’s Engineers, though, might have the right idea. There are plenty of arguments against the belief of automation replacing highly skilled labor, especially those positions involved in decision making and other complex, non-repeatable tasks.

Before going any further into how this battle is playing out in litigation support, specifically, we should discuss the concept of routine vs. variable labor.

Routine vs. variable

There is no doubt that automation has reduced manual labor and the need for human intervention in manufacturing, finance, IT, agriculture, and even litigation support. However, the one element shared among all of the tasks being automated is predictable repetition. Tasks that must be taken care of the same way on a tight, everlasting schedule are well-suited to automation, and generally defined as routine.

A Special Report composed by The Economist last July argued that the level to which a job is routine will be the best measure of its susceptibility to automation, rather than its classification as “white or blue collar.” The authors noted that a 2003 study from Carl Frey and Michael Osborne placed about 47 percent of the American workforce in this at-risk category.

On the other hand, those tasks and functions considered to be variable will be either extremely difficult or impossible to automate. Robotics, automation software, artificial intelligence, and similar technologies are not that great at thinking on their feet at this stage, and they might not ever be able to handle highly complex, variable tasks that have many moving parts or require advanced thinking.

Impacts in law

The New York Times reported that automation has had a mixed impact on law firms in the past two decades, in some ways reducing staff sizes but not offering the capacity to completely dismantle the profession. For example, the news provider pointed out that attorney James Yoon used to need a combined 12 professionals on a patent trial compared to the four required today.

Still, Yoon told The Times that analytics, automation, and other technologies have largely acted as support systems rather than methods to entirely replace workers. One study cited in the article found that only 23 percent of the average attorney’s job could be fully automated.

What you have is the ability to create a leaner company, replacing workforce hours devoted to routine tasks with software and allowing professionals to focus on more strategic matters. In litigation support, the song remains the same – automation is having a clear and massive impact on the workforce, but not altogether replacing professionals.

The lit support landscape

Will we likely see more left-side EDRM tasks pertaining to Information Governance, Identification, and even rote Preservation subjected to enhanced automation? Sure, that’s happening already and forcing professionals in these areas to re-engineer their workflows and controls.

However, while such work can lend itself well to automation, automated sweeps – no matter how sophisticated or well controlled – are not today (nor will they likely be in the future) inured from talented human oversight, analysis, and control.

The EDRM is a continuum, not a checklist of individualized or dissociated tasks, and as such, the devil remains squarely in the details. The connective steps between various EDRM tasks are largely variable and reflective based upon upstream results and downstream goals, and these areas are where litigation support professionals remain unabashedly necessary.

New tools touting the ability to simply drag and drop ESI into a landing page and let software take it from there may be useful in treating small, rote, or otherwise unextraordinary data sets. But how often do high-exposure/high-risk investigations or litigations limit themselves to only homogenous data?

The ability for next generation tools to quickly ingest this data is useful, but the idea that the story ends there is at best dangerous and, at worst, negligent.

The speed and reach of automation is compelling; one need look no further than the rapid acceptance of analytics-enabled review and TAR to recognize the value. However, TAR without proper AI training is chaos, and that training and selection process at the front end of a predictive coding project is fully leveraged into human decision making.

Consider the SEC’s protocols for Model Second Request – a frequent case type for automation. The production must include not only the document identified as responsive, but also a detailed sampling of those that were not, including the parameters used by the respondent to make such determinations.

Why? Because even with artificially intelligent tools doing much of the iterative heavy lifting, a single overlooked misstep can result in downstream failures that might have aligned with what we asked automation to do, but are nonetheless incorrect.

Stimulating workforce transformation

There is an argument to be made here that, so long as litigation support professionals continue to learn and progress alongside technology, they will not be replaced by it.

Keep in mind here too that IBM Chief Executive Officer Ginni Rometty argued that automation will actually increase the number of available jobs in the future at the World Economic Forum in January. Rometty is not alone in this assertion, either.

The Economist’s Special Report cited data from the Federal Reserve Bank of St. Louis that showed a steep and sustainable climb in the number of non-routine cognitive employment positions spanning back to 1983. This supports the idea that, in many industries, automation is going to transform the workforce rather than entirely replace it.

Right now, demands are higher than ever for lit support staff with a deeper, broader knowledge of tools and technologies, as well as how best to use, monitor, and QC automation-driven projects. As more sensitive tasks get handled by automated tools, it is hard to imagine how that human element will be completely decommissioned in the near future.

Lit support circa 2020

Looking toward the end of the decade, I would have to argue that litigation support professionals will still be just as in-demand as they are today, if not even more so. The exponential increases in data, constant transformation of technology, evolution of regulations governing the litigation community, and other massive trends will only work to increase the size of the litigation support workforce.

Right now, project managers, digital forensics specialists, data analysts, and myriad other positions could not possibly be entirely replaced by technology. What’s more, it would be arguably impossible for them to do their jobs without automation. This is reminiscent of autonomation, also known as Jidoka, which Toyota defines as “automation with a human touch.”

This autonomation of litigation support will likely continue on, increasing the demand for well-rounded, highly skilled, and knowledgeable staff members.

So, long story short, I expect automation to continue stimulating change and transformation in litigation support, especially with respect to the workforce therein. Those organizations and individual lit support professionals that keep pace with automation and other technological advances will have strong prospects going forward.

Andy Parrish is HAYSTACKID’s acting Director of Client Services in Chicago