A featured contribution from Leadership Perspectives, a curated forum for enterprise technology leaders, nominated by our subscribers and vetted by the CIOApplications Editorial Board.

Dorsey & Whitney

A Legal Perspective: Machine Learning and AI in the Legal Sphere

Everywhere you turn, someone is talking about machine learning, predictive analytics, and artificial intelligence, or “AI” as it is commonly referenced. The legal industry is no exception. In fact, I had an attorney stop me in the hallway the other day to ask whether AI technology was going to eventually eliminate his job. While machine learning, predictive analytics, and “AI” have been around for a while in legal, the possibilities for further implementation of these tools abound. Nonetheless, I do think the human element continues to be required…and will be for some time to come.

Since back in the early 2010’s, predictive coding, a form of machine learning, has been used in e-discovery in the legal sector. As email volumes grew, legal practitioners turned to AI technologies, like predictive coding, to help efficiently manage the large volume of documents to be reviewed as part of discovery. By reviewing a sampling of documents, and then applying an algorithm that enabled the machine to review the rest of the documents, discovery time-lines could be expedited. Since then, courts – and not just in the United States – have endorsed the use of this AI technology to help assist in moving the discovery process forward more quickly. I can attest that at our firm, we are strong believers in the value of machine learning in discovery – CAL, as we call it (continuous active learning); we rely heavily upon it to manage document review costs, often cited as the most expensive aspect of discovery.

While we have also implemented other AI-like tools in discovery to assist in identifying PII and PHI or privilege content, for example, we have also seen growing adoption of AI-infused technology on the non-litigation side of our Firm. With the pandemic, we started to see an increase in requests related to using AI to help with analyzing contracts for a variety of business reasons. With the AI technology, just drag and drop a bunch of contracts into the AI platform and ask it, “Do these contracts automatically renew?” and the platform will provide a yes or no answer! Or, ask the AI platform to find all of the clauses related to the ability to increase prices pursuant to the contract terms. No longer do we need human eyes reviewing every page of every contract to locate that information…what a time saver!

Of course, the big fear with machine learning and AI is whether it will replace the need for a human to do any work. I would say, however, that we are a way off from machines replacing humans. Take, for example, the discovery process described earlier and the use of CAL. We still need humans to teach the machine in each case what is relevant to the case and what is not. This is an iterative process. And while we find we can reduce the amount of human review, we do still require a level of human review.

Likewise with the contract analysis platform. We find that in some instances, the AI is spot-on in identifying key contract provisions or answering those yes/no questions. But, we also find that sometimes the technology is over-inclusive or under-inclusive in what it identifies and, sometimes we need to review 25-100 contracts to teach the platform what we are looking for, in order for it to find the language on its own. Humans are still required as part of the process.

There remains the question, on how far AI will go, and warnings against AI and the unlawful practice of law. For this reason, alone, the human element is still needed


AI and machine learning technologies are present in legal research, as well. At our firm, one of our legal research platforms allows us to pose natural language statements to find relevant case law. Just yesterday, I typed a statement regarding the obligation to retain videos as part of a legal hold if litigation is anticipated. The platform was able to return for me, cases that discussed a party’s obligation to retain video if potentially relevant to discovery. This same platform allows you to drag and drop a document - even a news article - to find relevant case law. This type of technology can really help to expedite the legal research process and save clients money.

Recently, I had the opportunity to play with some beta AI software that allowed me to automate various drafting tasks. I even asked it to “write me a poem about electronic discovery” and minutes later, it produced a haiku for my review. This sort of technology may eventually be invaluable in efficiently drafting legal documents. In fact, there are a growing number of companies that are offering AI contract drafting software. While we’ve long used technology integrated into our Microsoft Word deployment to proof the accuracy of already drafted contracts, the newer AI technologies take drafting to a new level.

Of course, AI and machine learning are finding their way into legal in other ways, too. Technologies exist to assist with predicting legal outcomes, to determine whether settlement is a preferable option than going to court. Other platforms can help predict when – and how – a particular judge may rule on a case. There is AI to help with jury selection, promising to help select an impartial jury. We have even developed an internal tool to help us in blind matching of junior associates to case assignments by utilizing an algorithm to match the associate to the assignment, in an effort to ensure all junior associates have equal access to the same career opportunities.

All of this is very exciting, but of course, there remain questions on how far AI will go, and warnings against AI and the unlawful practice of law. For this reason, alone, the human element is still needed. So, in answer to the attorney who stopped me in the hallway the other day, “Your job is safe, for now!” Although, I did just read an article about a robot lawyer arguing in court…

The articles from these contributors are based on their personal expertise and viewpoints, and do not necessarily reflect the opinions of their employers or affiliated organizations.