Adv. Janet Belleli Goodvach is head of E-Discovery and Document Review at LawFlex. She has spent over 20 years managing large and complex litigation, arbitration and document review projects across many jurisdictions including in Australia, Europe, the USA, Africa and the Middle East. Janet graduated with a BA/LLB from Monash University Australia and worked with Australian and British law firms, before joining LawFlex. Janet has managed many international LawFlex project teams for eDisclosure / eDiscovery and document management projects. She looks forward to the days of fully autonomous vehicles, so that she can stop saying “I wish there were more hours in a day.”
Croatia recently announced that it will be starting trials in 2024 for fully self-driving cars. Global cities have been conducting similar autonomous vehicle trials for the past 20 years, and despite continued technological improvements, most experts predict that fully self-driving cars are still up to a decade away from a reality on our roads. We have come a long way, but the world is not yet ready for a full adoption of autonomous vehicles.
Similarly, in the legal sector areas of eDiscovery / eDisclosure and document management systems, the technology has developed at an astonishing pace and autonomous review systems are increasingly being adopted, but we are still some years away from computer assisted review (CAR) systems being fully adopted.
I remember working on multiple court disputes relating to an airline’s collapse, about 20 years ago, buried in mountains of boxes of documents for several years straight, working round the clock with large teams of lawyers to try and extract, classify, and make order out of the abundance of relevant documents for various court proceedings.
One of my horror moments as a litigator was when I thought I had completed the review of about 1000 boxes of documents only to discover that there were two extra floors of archives in the client’s office that no one had previously mentioned… The further review took the next six months of my life (together with a document review team of junior litigators) and further costs and court delays.
Over the years, I saw the paper volume of documents for litigation & arbitration proceedings get smaller and smaller, but the digital volume grow to extraordinary levels – from kilobytes, to megabytes, to gigabytes and now in most cases I work on, terabytes of data. I am confident my first petabyte case is around the corner.
I no longer live among boxes of documents when managing document review projects thanks to the myriad of powerful eDiscovery and document management platforms in the market. The developments have been dramatic not just in data storage capacities, but in machine learning capabilities. New technologies facilitate the ingestion, review and classification of data at previously unseen speeds and volumes.
CAR for digital document management and eDiscovery, often also referred to as TAR (Technology Assisted Review) or predictive coding have proved to be faster and more reliable than human-only review and have been accepted by courts as an acceptable way to search for documents. New York courts accepted CAR as an acceptable tool in litigation over a decade ago, with the United Kingdom and other courts internationally following soon after.
However, despite the acceptance of CAR so many years ago, and the improved technologies, CAR has not been adopted as widely and quickly as many expected.
With in-house lawyers and law firms under pressure to find ways to cut costs, and the legal establishment under pressure to reduce the cost of resolving disputes, it was expected that CAR technologies would be the autonomous vehicle that all litigators would be driving by 2024.
However, whilst lawyers and the document review industry have been humbled by what CAR can do, several roadblocks to a fully autonomous system remain.
CAR learns what documents are relevant, confidential or privileged, based on an algorithm that is initially trained by the lawyers who have familiarity with the issues in dispute, and the aspects of confidentiality and privilege. However the computer driven systems often miss the nuances and context of documents, which humans are able to detect and correct. Prior to deciding to adopt CAR, legal teams need to consider the complexity or nuanced categories of search terms that must be reviewed.
Before CAR can be left to run “free”, meticulous planning and human teamwork is required for establishing and working through the training set of documents. With large document data sets, a careful balance is required. Large teams of lawyers may be needed, but it is essential that they are managed carefully to ensure they are synchronised, to avoid variations in coding. If these processes are not carefully managed, and quality control steps put in place, the courts may order that the disclosure exercise be re-done in whole or in part.
Another issue preventing the widespread adoption of CAR, is that the costs remain significant. CAR must be set up, and trained before it can be used, and quality assurance is an on-going requirement throughout any project. There are many cases where the costs of using CAR, are similar to, or more than a manual review process. This is particularly the case where a dispute involves a large number of hard copy documents which cannot be easily ingested into the system in such a way that the AI can read and search the text. If there is a significant volume of such documents, CAR will not be suitable.
Finally, many studies have demonstrated that GPS devices have caused our navigational skills to atrophy. Similarly, reliance on CAR reduces the legal team’s understanding, fluency and familiarity with the case documents, potentially compromising their capabilities. The process of manual review, and then often second or third level review, enables the team of lawyers to become more familiar with the corpus of documents and evidence. A stronger understanding of the document pool enables better dispute preparation and client outcomes.
I am confident that further improved CAR and allied litigation technologies, will continue to address the issues of nuance, inconsistency, reliability, and costs. This will decrease the number of lawyers required to manage large cases or manage large document review projects. However, as in the vehicle context, despite CAR continuously improving, we will need to keep drivers at the wheel for a little longer.