Will Robots Replace Lawyers?

Article by Rei Morikawa | June 10, 2019

The legal system can seem impossible to navigate, but what if a robot could help guide you through the whole process and give you personal, legal advice? AI and machine learning have grown exponentially in the past years, and are increasingly being used to automate processes in various fields including healthcare, transportation, and even law. In this article, we explore just a few areas of law that could be automated using AI.


Machine Learning to Predict Criminal Sentences

An increasing number of criminal sentencing recommendations are now based on AI algorithms.

But the accuracy and fairness of using AI for this purpose is still ambiguous. The US Constitution provides for fair trial rights, but it’s hard to tell whether using AI is fair to the defendant. On the one hand, using computers can be a good way to eliminate subconscious human bias. On the other hand, machine learning models base their decisions on what happened in the past, so there’s actually a risk that computers will pick on past discrimination patterns and actively apply them when making new decisions.

In many cases, judges have no way of knowing how the algorithms work to reach their sentencing recommendations. We might expect that the algorithms consider similar factors as human judges do, such as gravity of the offense, the defendant’s age, education, prior criminal record, and living situation — but we cannot know for sure. This is because many algorithms are developed by private companies that are not required to publicly reveal the technological processes behind how their machines reach sentencing recommendations.


Machine Learning for Divorce Settlements

Almost half of all marriages in the US and Canada end in divorce, so it’s no wonder an app like Wevorce would be popular. Wevorce aims to make every divorce amicable and basically acts as a tax software for divorce. It prompts users to answer questions about their ideal outcome regarding the division of assets, co-parenting plan, and other important decisions. Then, the algorithm suggests a compromised solution based on both parties’ answers. Wevorce also tries to provide emotional support to divorcees going through a tough time, by showing them positive and uplifting articles every time they open the app.

A similar app called Thistoo used to be available for divorcees in Canada, but the company closed earlier this year. It worked by asking users to fill in their basic information such as household income and number of children. Then, the AI algorithm searched through local Canadian provincial case law to make suggestions for how everything should be divided, based on previous court rulings. Thistoo auto-filled the various require forms for divorce, and reminded users to gather other documents such as pension or health benefit forms, that they could have forgotten about. At the end of the process, the app recommended but did not require users to have a lawyer review everything.


AI to Complete Immigration Applications

SimpleCitizen calls itself the “TurboTax for immigration.” Non-US citizens who wish to apply for citizenship, a working visa, green card, or permanent residency can use SimpleCitizen to navigate American immigration laws and autofill the required forms for their applications without an attorney. Currently, it seems that everything is reviewed by an attorney, but it’s possible the company will rely more on machine learning and computers to do the job in the future.


Machine Learning to Predict Trial Outcomes

With machine learning, we can feed computers ground truth data about the individual facts, applicable law, and outcomes of past cases. Then, computers can identify patterns based on that ground truth data, and make predictions about new cases. It’s unlikely that robots will replace human judges in the courtroom, but lawyers can still use this AI application to their strategic advantage. For example, if the algorithm concludes that your client only has a 5% chance of prevailing in court, that would factor into the lawyer’s decision to encourage his client to accept a settlement offer.


AI to Appeal Your Parking Tickets

DoNotPay is a chatbot that streamlines the process of contesting parking tickets. To use the app, the first step is for customers to communicate the issue they’re having by typing statements like “I got an unfair parking ticket.” Then, the chatbot will prompt you for more details so it can direct you to the appropriate issue. Finally, the chatbot generates an appeal letter that you can sign, print, and submit to the proper court or institution. The appeal letters include language like “I believe that the court should exercise fairness in cancelling a ticket that . . . is perfectly justified to be cancelled,” and “I feel that the issue of a ticket is an unlawful action inconsistent with precedent.” DoNotPay can also help users get off due to technicalities, such as if a ticket incorrectly describes a car’s color or make.


Machine Learning for Legal Research and Discovery

Law firms are starting to use natural language processing to predict what kind of documents will be relevant to a case within a few minutes. This saves hours of work for lawyers, and hundreds of thousands of dollars in fees for their clients. In addition, lawyers can use predictive coding in eDiscovery to locate and rank relevant documents, instead of manually conducting word searches on every document. Predictive coding is a form of machine learning, so the computer becomes increasingly accurate as it absorbs additional input data through repeated uses.

The Author
Rei Morikawa

Rei writes content for Lionbridge’s website, blog articles, and social media. Born and raised in Tokyo, but also studied abroad in the US. A huge people person, and passionate about long-distance running, traveling, and discovering new music on Spotify.


    Sign up to our newsletter for fresh developments from the world of training data. Lionbridge brings you interviews with industry experts, dataset collections and more.