Robots Are Taking White Collar Jobs, Too

If you find yourself in court in the near future, your lawyer might be a robot. 


We’ve all heard the predictions: robots are coming for our jobs. And not just factory work, service jobs, or deliveries—we’re talking white collar jobs as well.

One of the industries the World Economic Forum predicts will take a hit across the world’s biggest economies is the legal field. While on the surface it might seem impossible to automate a job that requires problem solving, critical thinking, and persuading judges and juries, when one considers the mountains of paperwork and research involved in lawyering, it’s easier to see where machines might have a leg up, so to speak.

For years e-discovery software has helped lawyers manage the electronic records that need to be produced at the beginning of an investigation or law suit. Different e-discovery programs help process, search, review, produce, and store documents, and have become an industry staple, but such automated legal assistance is just the beginning.

Last fall, Joshua Browder, an 18-year-old computer science major from the UK who had reached his wit’s end with “trivial” excuses for levying fines, created a website for friends and family plagued by parking tickets. Within a month, donotpay.co.uk logged 6,500 complaints from disgruntled motorists. The site pulls together everything necessary for drivers to send complaints and defenses to the local council, including photos of road marking and ambiguous signage, and it can “generate winning appeals to parking tickets in seconds.” After helping drivers avoid thousands of dollars in fines, Browder upgraded the service.

Billed as the “world’s first robot lawyer,” DoNotPay now does more than register complaints for parking tickets. Browder got so many emails asking for help that he modified the site to more closely replicate a human lawyer. An interaction with the DoNotPay robot resembles a text conversation, but unlike Microsoft’s recent debacle with the Tay chatbot, Browder’s AI doesn’t insult or seduce people. Instead, it gives real-time legal advice, answer questions, and files claims for basic legal situations, all for free.

And now there’s ROSS, an automated attorney whose artificial intelligence integrates legal knowledge with natural language capabilities. The AI can understand complex legal questions in context and then offer advice, additional readings for more information, and citations for legal statutes. An iteration of IBM’s Watson, which rose to stardom when it decimated Jeopardy’s most impressive human contestants and has since been used to create recipes, analyze genetic information to help treat brain cancer, and to develop healthcare, education, and agricultural systems in Africa, ROSS’s AI functions like an indefatigable pack of paralegals.

ROSS keeps itself and its clients updated on legal changes that might bear on one’s case. It also has the ability to learn by experience, consistently refining its knowledge based on the questions and situations it engages. ROSS’s ever-increasing skill set has turned heads in the legal world, and the international firm Baker & Hostetler recently announced that it will bring ROSS onboard to work in its bankruptcy practice.

Right now, it’s unclear whether ROSS will replace human workers. Given that it will primarily perform drudge-work, it may bump a paralegal or young associate onto other assignments, which may sound good, but critics worry that denying lawyers this type of practice will limit their formative hands-on experience. Time will tell whether future AI will be able to handle a lawyer’s full breadth of casework.

According to a paper published on the Social Science Research Network, the varied nature of legal work might be lawyers’ saving grace. While AI continues to advance when it comes to specific skills, such as winning the game Go, artificial general intelligence (AGI), or the ability for AI to learn and apply knowledge to other situations it wasn’t expressly programmed for, remains out of reach. Most lawyers complete a variety of different tasks, only some of which involve research, so it’s unlikely that a single system would be able to do everything a lawyer does on a given day. Beyond the tasks themselves, the paper also points out that machines can’t embody values and ethical codes, which apparently some human lawyers have.

On the upside, robot lawyers would make sense given the tricky road we’re inevitably facing when it comes to robotics law and maybe even robot rights. If robots become the subjects of laws and protections, then perhaps they should learn how to navigate the system. Of course, as with everything else, we run the risk of being surpassed in skill and acumen by our robotic counterparts.

Although by then, robot judges may be banging gavels and silencing courtrooms, as well as naysayers.