Blog Post

Why Machine Learning is Key to Legal Transformation

November 19, 2020

It sounds intimidating and highly complicated, but is machine learning all hyperbole or could it be the hero you need for your legal department? Read our 101 guide to find out.

One of the biggest problems with technology is that it never quite lives up to the hype. Anyone who’s tried to connect their game console to their TV or their printer to their Wi-Fi knows nothing is ever as simple as the tech vendors would have you believe.

And it’s not getting any easier. We’re constantly lambasted with new words and shiny concepts, and told how amazing they’ll make our lives, one day. When was the last time you made any money out of bitcoin or been shopping and used augmented reality to grab a bargain?

But for every nanotech or Second Life (remember that!) there’s cloud computing and smartphones – technology that’s taken off so widely most of us would consider them an essential part of our lives. Machine learning is (and will be) one such tech.

Search for machine learning and the legal sector on Google and returns about 91 million results. And according to Zion Market Research, the global legal tech AI market was valued at $3B in 2018. By 2026, the industry is expected to generate around $38B, reflecting a CAGR of 36 percent over seven years.

The reason for this boom is that machine learning is a technology that will change our lives – and already has been for the past decade or so. Satellite navigation, email, social networking, Google maps, online banking, Alexa – the list goes on. And its grip on the legal sector is only just taking hold.

But like most technology these days, it’s complicated to understand and that makes it intimidating to use. Not anymore. Below we give you Machine Learning 101, what it is, how it works, and why it’s time everyone in the legal sector started to use it.

What is machine learning?

Machine learning is often used interchangeably with Artificial Intelligence (AI). This is incorrect – it is a subset of AI, which is a broader term used for intelligent machines that can mimic human understanding.

Machine learning is one specific element of AI that centers around data. It uses algorithms and statistics to find patterns within huge datasets. It requires humans to train it on past experiences, using huge amounts of data. For example, a law firm using machine learning to undertake due diligence would require its lawyers to train the system on industry-specific clauses, policy, procedures, IP, and compliance.

The machine learning algorithm learns from this dataset and repetition of information and then applies that knowledge to any new data it receives to discover any patterns or anomalies. The training is repeated to ensure more accurate results, and the more data you feed the algorithm the more precise your results will be.

How it works – three different flavors

To ‘learn’ machine learning algorithms are divided into three flavors. The one you use will depend on how your data is classified, what dataset you have, and the condition of the data you use.

Supervised learning
The most popular archetype for machine learning is supervised learning. Most machine learning is based on supervised learning. This is where the data is labeled or classified to tell the machine exactly what patterns it should look for.

The machine then learns from this classified data to predict and classify unlabelled data. By repeating this process (aka training the machine), the algorithm learns and can then apply this knowledge to any other dataset.

For example, this could include dates or specific clauses in a legal context. Alternatively, when you press play on your favorite Netflix show, you’re telling their machine learning exactly what types of shows you’re interested in, and the more shows you watch, the more accurate it will become in predicting what shows you would more likely be interested in.

Unsupervised learning
Unsupervised learning is the opposite of supervised learning. All the data is unlabelled or unclassified and the machine looks for whatever patterns it can find and groups the data according to those patterns.

The idea is that the machine learning algorithm organizes huge data sets into groups that are related or corresponding. The challenge with unsupervised data machine learning for many businesses is that it’s tricky to find an application for it.

One area that has seen an uptake in unsupervised machine learning is where organizations need to group or cluster customers by purchase behavior. In a legal context, such as legal document review, the unsupervised machine learning will cluster similar documents or clauses, along with clear anomalies from those groups. This would therefore reduce the time spent on manual review and ensure more time could be focused on reviewing those outliers where clauses or documents were different.

Reinforcement learning
Reinforcement learning is where the algorithm uses a trial and error method to come to a clear objective. It’s a game-like scenario where the machine is rewarded or penalized for coming up with the right solution.

The programmer has set the reward policy (or rules of the game) but machine learning has no idea what the correct outcome should be. The machine understands that the aim is to achieve the most rewards and so continually will try different ways to achieve the greatest reward – hence eventually finding a solution to the problem. Like using treats to train a dog, reinforcement learning is the tool used by Google’s AlphaGo program used to beat human chess champions.

Why use machine learning in legal?
While there will always be some laggards, (and there are still many in the legal sector), law firms across the world are starting to reap the benefits of using machine learning. From document review to due diligence, pre signature contract review to post signature contract analysis, we’ve talked about how legal departments are finding a way to use machine learning to augment their services for several years.

And that’s the point about, it’s not designed to take over the role of lawyers or paralegals – quite the opposite. It is a technology that’s designed specifically to work alongside human intelligence.

It’s designed to alleviate the laborious and tedious work, leaving lawyers free to focus on the reason they became lawyers in the first place – to provide the best client counsel and advice and ensure a successful outcome every time for every client.

Want to find out how machine learning could power your law firm or take your legal team to the next level? Contact us now to find out more.