picture of Rajah

Rajah Lehal

No More Wax On, Wax Off

June 26, 2017

Links from this article:Blue J LegalSentiment Assessment TechnologyIdeal.comPlaybooks

I was recently asked to speak at Big Data Toronto on the topic of machine learning and AI, and I decided to speak to a question on my mind that I have been struggling with: what are the implications of the replacement of junior staff in an organization by software-based solutions? To be more specific, if machine-learning and AI are doing the easy or tedious work reserved for untrained hires—and the same tedious tasks are an important building block for up-and-coming team members—then how do we make sure that in the future, our juniors will receive proper training alongside this new technology? In other words, what are the implications of collapsing a category of work that fosters on-the-job training?

As the operator of a business, there’s nothing more pleasing for me than an efficient, consistent system that saves time.

My passion has been developing a legal document comparison software called Clausehound.com, which is deeply investing in machine learning that has led to significant progress in the annotated language generated by the tool. I am also the founder and on the board of directors of Cobalt Lawyers, a law firm that hires and trains young lawyers. As can be seen, I have considered this problem of training young lawyers using the technology of Clausehound.com from a few angles.

Wax on, wax off.

Clausehound hits the bullseye when it comes to using machine learning, by the way. Clausehound maps knowledge into smart contracts—we intersperse thousands of bits of learning material into contracts so that drafters are immediately connected to clause definitions and lawyer-written articles. Our algorithm, which is getting smarter, is intended to automatically match learning materials to our taxonomy of contract clauses.

By the numbers:

Number of clauses that we mapped into our database in our first week: 400

Number that we mapped this week: 4000

As a lawyer, I have spent years training hundreds of law students and young lawyers on how to research and read contracts and case law, to pull out “business” insights, and to draft and review agreements. I appreciate how important it is for newly-minted lawyers (or staff in any services role) to work through the basics, make mistakes, and do the same tedious tasks over and over to “practice” in order to learn how to deal with context-based scenarios.

Although work probably only needs to be done by one person, the end-customer might get two or three lawyers on a file (or accountants, or investment advisors, or cooks in a restaurant), in a workflow that has the junior doing the bulk of the work and the senior reviewing it all in a fraction of that time.

From restaurant meals to law firms bills, this “apprenticeship” is built into the pricing of services.

Whether it’s a restaurant chef, a lawyer, a doctor, an accountant or an investment professional, the matching of context or fact pattern-based concepts is something that is typically learned from experience and historically has been difficult to transfer to a categorized database. Software that is adept at pattern-matching is getting rid of all of the tedious “staging” work that consumes hundreds of hours of the time of junior staff.

The death of the apprenticeship model of services training.

So what happens when machines start to do the work of juniors? Thinking about this from the perspective of a law firm, training time will shrink, and training budgets will shrink as customers will demand lower fees. Customers will self-educate and “Do It Yourself” (DIY) everything except specialist work, relieving junior staff of their responsibilities, or at the very least, challenging junior staff members as to the accuracy of their work. This is not a new problem—for years, medical professionals have had to diagnose patients who arrive armed with research. Extending to other professions, high-stress encounters will likely commence between junior staff members and customers who self-educate in advance of meetings, knowing as much about a topic as the junior team because they are learning from similar systems and materials.

Faster training also means potential integration of tasks that were once outsourced. This integration may be customer-driven: companies can hire internal clerical staff who can be trained onsite instead of using expensive outside law firms. Customers will start to develop loyalty to their chosen software in favour of certain lawyers and law firms.This integration may also be market-driven, meaning that in the future, companies need to make this shift, because law firms will not be able to afford extra staff if their training budgets disappear.

*More numbers:** *

With practice, 15 minutes is the time that it takes to read a legal case. It takes an additional 45 minutes to create a briefing of the relevant cases to share with your senior counsel. (Longer if you’re not sure what you’re looking for.)

  • **Cloud-based research tools like *Blue J Legal* can now analyze thousands of legal cases in a single request, and some tools will prepare a complete legal memorandum.*

Customers performing their own research? This is not ideal for the up-and-coming lawyers who benefit by reading all of these judgements. A junior lawyer learns a lot of skills in researching cases, including:

  • asking the client the right questions,
  • framing the search properly, and
  • writing clearly and effectively.

Reading primary materials also gives lawyers the confidence to support their position.

Using legal software also undermines the specialist skill development of lawyers. Their natural exposure to the problems faced by a diverse set of customers expands their understanding by shifting a lawyer’s learning experience to the customer or to in-house personnel. This creates a dearth of learning experience for the junior lawyer.

Will the categorizations and summaries created by software oversimplify the problem? Spotting a needle in a haystack or outlier situations is part of the learning process for a junior lawyer.

So the core question that remains is whether legal software is the right solution to replace some of the tedium of being a lawyer while delivering efficient legal solutions for the client. Don’t get me wrong, I’m not frowning upon software as a solution for tedious tasks. Most of my day is spent trying to improve the DIY software process for our users and customers. This is something I think is necessary to ensure that the software we build is sensible.

Human factors in the client-lawyer interface?

Customer intake is another junior task. Whether a junior at a law firm or a server at a restaurant, is it better to have the customer interview with a human, or is a ‘bot more suited to patiently asking every question and assessing sentiment, worry, and concern?

Client interviews help new staff in understanding and assessing how the customer thinks and in managing the time-wasters and the tricky clients, which historically can be “an art”. Client intake meetings are a training ground for managing clients and improving “detective skills” that you learn from meeting clients. However, early sentiment assessment technology is already in the market, and chatbots are definitely more patient than most human interviewers.

Is the “human touch” a necessary part of the restaurant experience? Would you prefer a robot server at the restaurant, or a human? I guess it depends, but I know that I go straight for the self-serve kiosk at McDonald’s.

Should services businesses start to spend less time worrying about a trained front-line staff, and instead replace this category with software solutions and a friendly host/relationship manager (a fungible skill)?

What’s the risk of replacing a junior legal team with data scientists? Statistics-driven solutions will reduce creativity in problem solving, as the solution will be checklist-based and FAQ-based. As noted above, the risk of oversimplification or bias should be taken into account when developing a checklist.

**So who benefits from the shrinking of junior professional service? **

Software engineers and data scientists will benefit. The software providers will benefit. End customers will benefit. Organizations using the software as part of their customer experience will benefit by on-boarding customer materials into their software.

There’s a wave of transparency coming. The use of online tools will help give guidance on the choice and rejection of options when data is collected and examined in aggregate.

For professional services organizations, there’s a wave of new customers coming. More DIY means more clients. Many underserved in the community will become partly-served by software and partly-served by specialists. By analogy (and by anecdote), the introduction of automated teller machine ATMs led not to a reduction of bank employees but rather to more bank branches.

Specialist law firms will benefit—lawyers that were “stuck in the weeds” on tedious details will have more time to focus on working on the non-tedious matters. Generalist lawyers and law firms will struggle (and possibly be absorbed into larger law firms), as specializations will likely be required in order to successfully compete with software solutions.

Education emphasis should be on critical thinking and analysis.

From robot burger-flipping to automated general ledger updating, there’s also a wave of categories of work in which underemployment is coming.

Machine learning is removing tasks that in many cases are jobs they don’t want to do, to save money and then reinvest their money so their family can move into a higher paying job. Efficiency through technology is removing the rungs of social/economic mobility. Educators will need to emphasize building the skills that allow for creativity, analysis and the fast development of specializations (assuming that machine learning will tackle the generalist skills).

How do humans compete?

  • Keep learning: For example, think of computer-based training to prepare students for the workforce, and solution providers creating DIY offerings likely also creating training materials (like our Playbooks).
  • (My friends at Ideal.com suggested this): To remain competitive, students should focus on complex skills that require cognitive reasoning, abstract reasoning and emotional intelligence. Robots haven’t replicated the skill to create arguments by analogy and imagination, at least not yet.
  • Stay “warm”: Many will argue that human factors, emotional intelligence, and even “warmth” and humanness are irreplaceable. So stay spunky and keep ‘em guessing.

Although AI is disrupting the legal industry and will be a force for change, lawyers (especially junior ones) can and should find skills that will differentiate them from the ‘bot. This can be anything, from intuitive traits that a machine cannot possess to new skills that aren’t skills commonly possessed by lawyers. A machine may be faster and smarter than you, but at the end of the day, only you can relate to your human client’s needs!


Written by Rajah. Rajah Lehal is Founder and CEO of Clausehound.com. Rajah is a legal technologist and technology lawyer who is, together with the Clausehound team, capturing and sharing lawyer expertise, building deal negotiation libraries, teaching negotiation in classrooms, and automating negotiation with software.