How to Get Artificially Intelligent in Litigation

You may not always notice it but our daily lives are entwined with Artificial Intelligence (AI).

Visit Amazon and you will be presented with a list of other products you may like to purchase, browse Netflix and the platform will highlight shows you may be interested in watching, log into Facebook and you will be delivered information on our electronic friends based on our perceived likes and dislikes. AI learns from our behaviours and our interactions and then uses this information to point us towards other things that might grab our attention. What’s more, it does all this with such subtlety that we begin to take its convenience for granted.

So how does AI work? Well, it’s nothing to do with individuality. AI uses algorithms to assess us, not as unique snowflakes but as a collection of people with similar interests. By analysing patterns within this collection, the algorithms predict the likelihood of other individuals sharing these interests and behaviours. As consumers, we respond well to the helpful nature of these algorithms, and use them to facilitate the discovery of other things of interest.

In litigation – and in electronic data discovery in particular – any tool that assists us in getting hold of the relevant documents sooner rather than later is going to be helpful. Less time spent reading irrelevant documents means more time spent focusing on valuable information and substantial issues. The wasted time and associated cost of perusing these blind alleys has been criticised by many, including the Australian Law Reform Commission, who cite it as evidence for discovery reform. Would the implementation of AI algorithms in this capacity reduce if not eliminate this cost?

In fact, these algorithms are deployed already. Predictive coding is the application of AI in electronic discovery and is already being put to use in the field, although you would be forgiven for not noticing it. The fact is, software development is an incredibly competitive industry and the leading vendors need to find ways to deliver an enhanced experience to their users. Integrating predictive coding protocols like Continuous Active Learning (CAL) into document review work-flows is a great way to achieve this. 

The benefit this brings cannot be understated. Vast, sprawling documents – which will be familiar to anyone involved in litigation – can now be approached very differently. Once upon a time, the sheer weight of all the data made it prohibitively expensive and time-consuming to deal with. Now we can easily manage documents, ordering them from most to least relevant and scanning the information within them quickly and effectively. Proportionality arguments against dealing with large document collections now hold less weight as summarised in Pyrrho Investments Ltd v MWB Property Ltd &Ors[2016] EWHC 256.

So what are the next steps? What do you do if you are interested in adopting AI within your own field of litigation, or if you have your doubts regarding its effectiveness? You need to start by gaining experience in using AI and by following these four steps to efficiently integrate smart algorithms into your matter.

  • Find a service provider who can offer CAL as an integrated AI solution. The technology should be seamless to use and come as a feature of the review software, not as an additional expense that must be sold to your client without first understanding its benefit.

  • Establish review work-flows that incorporate the CAL inputs and outputs. A recent study has shown that starting with known relevant documents and then continuously trying to find other relevant documents is superior to other approaches to predictive coding. Take advantage of CAL’s prioritisation function and review documents from highest ranked to lowest. This will let the system guide you to other relevant documents similar to the ones you have already found.

  • Read up on CAL and gain an understanding of its strengths and limitations. CAL should not be viewed as a total replacement for document review and it is important that you understand the relationship between precision and recall and know how to use both in formulating proportionality arguments.

  • Work only with the legal and technical teams who are prepared to incorporate CAL into a search and culling strategy and who are comfortable with not reviewing documents even when they contain apparently relevant keywords. Once you and your team understand the CAL process, you will warm to this proposition and having the right support around you will help integrate AI with the search process.

Electronic information is now ubiquitous in litigation. It's not uncommon to deal with data sets of too great a size for traditional linear search methods to handle. AI in litigation provides an agile and inexpensive alternative to effectively combat oppressive levels of data. These new methods are here to stay, so lawyers and consultants must become familiar with their benefits. To start building your knowledge and your experience, begin with the first three steps outlined above. Make yourself comfortable and confident with the concepts and outcomes of CAL. Once you become clued up on the subject, you will have no trouble in applying AI in your relentless progression towards innovation, efficiency, and cost-reduction.

Benjamin Kennedy

Benjamin Kennedy
Manager, eDiscovery & Forensics

Filed under: ai, artificial intelligence, data discovery

The contents of this article are intended to convey general information only and not to provide legal advice or opinions.

The Epiq Angle brings you our thinking on topical issues in eDiscovery, bankruptcy, corporate restructuring, data breach response, global business transformation solutions, class action, and mass tort administration.

By continuing to browse and accepting this banner, you consent to the storing of first and third-party cookies on your device to enhance site navigation, analyze site usage, and assist in Epiq’s marketing efforts. Read more on our cookie notice.