Breaking Down Legal AI

Hi and welcome to the second installment in the ultra series, where we’ll explore what we mean by AI. AI is a term with no definite meaning. The term artificial intelligence is applied when the machine mimics cognitive functions that we associate with human minds. Just learning and problem solving. » The scope of AI is uncertain at the best of times. And to make things even more confusing, it’s a moving target. As machines become increasingly capable, tasks considered as requiring intelligence are often removed from the definition. For instance, optical character recognition is now frequently excluded from artificial intelligence. Having become a routine technology. » It is much more helpful to break it down into different categories.

Different software may use multiple categories, but it helps to understand how the systems work. We will only discuss things relevant to legal AI. For example, robotics is usually considered a part of AI but it is not yet relevance lawyers. Here’s another quote for you to kick things off. It’s only AI, when you don’t know how it works. Once you know how it works, it’s just software. Kurt Keutz. » For our first dive into AI, let’s talk about machine learning. Broadly breaks down into two main parts, supervised learning and unsupervised learning. Supervised learning at its most basic is where a machine has an input to an output. It’s usually a number or a particular tag.

Let’s say you wanted to categorize millions of pictures of pets into subcategories, cats and dogs. You could take the first few hundred pictures and tell the machine, this is a picture of a cat. This is also a cat. This is a dog. The machine would then use that data to categorize the remaining few million pictures. The legal context is may be used to identify a particular clause such as jurisdiction clause, or the amount of salary in an employment contract. » Unsupervised learning is where the machine clusters groups based on a level of similarity, so you may give it a particular clause or a document and the machine looks for similar items.

The machine will find clauses or documents that have similar wording to the original, and you can set the perimeters so you get what you’re looking for. The machine is the one using various algorithms to find the groups rather than relying on human reinforcement. » Semi supervised or reinforcement learning uses unsupervised learning technology with human reinforcement at stages along the way. Machine tries to group things and then the human says, you got this bit right, you got the other bit wrong, learn from it and adjust. With most of these systems, there’s often a trade off between getting everything you’re looking for and getting only the things that you’re looking for.

Imagine a barrel full of marbles, all sorts of marbles, all different colors. I asked the machine to find orange marbles. Now, the machine has a fairly good idea what I mean by orange, but the problem becomes around the edges. What about tangerine, ginger? How about scarlet colored marbles? If the machine gives me everything that is vaguely orange, I may have too many marbles to look through. If its more selective, then it may miss out the marble that I need. To put it all in a legal context, and let’s be extreme about it.

If I were looking for a particular clause in a contract, if the search returns every clause in the document, it will definitely have the one I need, but also lots that’s irrelevant. Even though it’s brought me the right clause, the volume of irrelevant information makes it useless. On the flip side, if it returns no clauses, it will definitely not have any irrelevant clauses, but it will also be missing the right one. There’s often a balance between these two competing objectives, including the right thing and not including anything irrelevant, with a perfect balance being difficult to achieve. » Now let’s move on to talk about expert systems.

Expert systems apply an automated decision tree to a series of data is the equivalent of a complicated flowchart. The data it relies on can either be entered manually, or harvested from the machine reading the documents. So the machine could look for the jurisdiction clause and identify that it’s England and Wales, or you could manually telecontract with the jurisdiction. This data would then be used to complete the decision tree. For example, is the transaction over 5,000 pounds? If yes, move to a step, if no, move to the other step. » The creation of the decision tree requires humans to program it. So the better the lawyer programming it, the better the result will be.

This system offers key advantages over human beings in that it will always apply the decision tree in the same way. Tirelessly without billable hours and without mistakes. However, these systems don’t work well with gray areas or matters that require discretion. Systems can be used to produce particular outputs such as amending a draft documents or they can be used to flag documents for human review.

For example, if you put 20,000 contracts into the system and asked it to check for all of the following criteria, contracts over 50,000 pounds which do not have a limitation clause which were entered into less than six years ago, and are between firm x and firm y, then it could send all those positive matches to a lawyer for review. » One key use of expert systems has been to automate parts of the drafting process. Documents used in law firms almost always follow precedents, and these precedents largely have amendments that can be predicted. By putting in key bits of information, the software can add, or remove certain clauses automatically.

Put in names, dates, facts and figures in all the places that they’re referred to without making cross referencing errors. Some of these contracts can be very long and refer to a party hundreds of times. Removing or adding a clause may have an impact on the numbering or references to lease clauses and having this done for you saves time and mistakes. Most documents however, still need some form of this baking. So the software will not do all the drafting for you, but it will speed up the process a lot. » Now let’s explore natural language processing. This focuses on getting machines to understand what we say, both orally and in written form.

It does use optical character recognition both for typed fonts and handwriting. But that isn’t really the clever bit anymore. What it does do is allows machines to make sense of context and meaning rather than just reading the individual words. The intricacy and accuracy of legal drafting has made this a challenging area. A machines are much better at recognizing what a clause is rather than exactly what it does. So a machine could tell you that clause x contains the names of the parties and who the parties are. It would also recognize now arbitration clause, what it wouldn’t be good at doing is working out if the arbitration clause applies in a particular set of circumstances.

Moving on, let’s discuss data analytics, big data and metadata. Data just can’t be handled in the way it used to be. There is so much more data now than ever before. Nowadays, every two days we create more data than was created between the dawn of civilization and 2003. A Rolodex and a filing cabinet won’t be enough to stay on top of things. Data Analytics, a form of machine learning uses past data to predict future events. Just how long a piece of litigation will last, how much it will cost and how much decided to settle for. For more tangible example, you can work out the cheapest time to buy a plane ticket using big data.

It doesn’t look at any particular plane ticket. It doesn’t know the plane that this ticket belong to was full or empty, or if the price of fuel had gone up or down. However, by putting together trends from millions of searches, it can give an accurate picture of when you’re most likely to get the best deal. Which is 70 days before your flight by the way. Big Data provides a lot of raw material for legal data analytics. Machines can now store and process huge quantities of data. And as more and more of our work is digitized, that data becomes more reliable.

By using thousands or millions of data points, it can find trends without relying on individual examples, it doesn’t understand why a particular case won or lost or why one settled and another didn’t, what it does do is look for trends in percentages. Metadata is usually created when digital information is created. It include things like the time and date it was created, the type of the file, how big the file is, who the creator was, which computer it was created on. How many times it was edited, that sort of thing. It’s often called data about data. It adds to the available pool of information and is a very valuable resource.

Data Analytics within law firms are most easily and most often used to improve admin functions. So timekeeping, billing and costs are high on the priority list. However, the more difficult areas such as using it to help predict the outcome of cases, or helping a client’s business will be the most fruitful if the law firms can achieve it. » Next up is process automation. Process Automation is where a machine takes over a whole task that was previously done by a human being. In the case of disclosure for example in 2016, a London court allowed the use of predictive coding software to decided which documents to disclose. Time recording software automates the old process of manually recording time sheets.

And voice dictation software like Dragon NaturallySpeaking automate some of the function of secretarial typing polls. One of the key tasks that has been automated, at least partly is legal research. The search functions within legal databases replace human beings walking around libraries and looking things up in different indexes. This is also evolving quickly. Modern AI tools like Ross allow much faster and more accurate research than is possible by humans alone. » Finally, let’s look at collaboration tools. Collaboration tools, like the cloud or video conferencing, allow lawyers to work from different locations and collaborate more effectively on the same project.

This also includes document management systems that create, file and link documents together in a way that’s secure, easily searchable, organized, and with features like version control. They also contribute to providing accurate and complete data for data analytics. » Let’s look to science fiction to round this section off. Arthur C Clarke said, any sufficiently advanced technology is indistinguishable from magic. It may seem that magic is being performed in front of our eyes, when documents can be read, understood and processed by machines. This capability will only improve. But we must remember that there are specific mechanisms at work here. There are systems and software at work. There is no magic.

And there certainly is no human level intelligence in anything other than humans just yet. There is just code and software that does what you tell it to, even if that is very clever.