I very much doubt that Master Matthews regards himself as a visionary of a dystopian future in the style of George Orwell or John Wyndham, irrespective of some portrayals of his recent judgment in Pyrrho Investments Ltd v MWB Property Ltd.
The issues in Pyrrho were relatively narrow. The court approved the use of predictive coding to sort over three million documents for the purposes of disclosure. The parties had agreed on the use of predictive coding and there is still the opportunity to use other review methods prior to trial if this proves ineffective.
Predictive coding and other forms of technology assisted review (TAR) rely on machine learning techniques that have been around for a long time. The rules enabling litigants to use TAR have been around for at least a decade since the introduction of Practice Direction 31, which was drafted deliberately widely to accommodate the emerging use of clustering technology at the time and to anticipate the evolution of new techniques without the need to subsequently change the rules.
The court, therefore, has always had the power to endorse the use of TAR, but Pyrrho is the first such formal endorsement; it is helpful in confirming that TAR is a valid adjunct to mass human review.
Rise of the machines
The clamour of portentous commentary saying that Pyrrho signals the sudden ‘rise of the machines’ perhaps reflects a thirst for better guidance around this issue in comparison to the relatively large number of decided cases in the US.
In reality, clients have been using TAR for some time, notwithstanding the absence of specific judicial endorsement. This reflects a broader corporate interest in artificial intelligence and software robotics to automate processes previously conducted as mass human exercises (both in legal and in the wider day-to-day commercial operations of the business).
Some corporates suggest that the use of such techniques has been relatively effective in litigation but those techniques need to be applied to other areas of legal activity to reduce significant legal spend.
For example, we are seeing significant interest in the use of these techniques in contract review and remediation exercises driven by current bank ring-fencing and other regulatory requirements.
For businesses, while litigation is a concern, they are much more concerned about the inexorable rise of regulation and the need to drive down costs in the legal function while delivering greater consistency and quality.
“While litigation is a concern, businesses are more concerned about the rise of regulation and the need to drive down legal costs”
And therein lies a paradox. In litigation, by the time you get to disclosure, the close of pleadings has crystallised the issues so it is easier to apply TAR to a defined set of issues within a reasonable deadline set by the court. There is time for the machine to learn.
In regulatory reviews, production is front loaded, timelines are aggressive and the issues are rarely crystallised. Clients and regulators are in investigation mode. In the race to gain the trust of the regulator, there is no time for the machine to learn.
So the opportunity to use TAR is, in practice, significantly curtailed. In those situations, clients need a different approach to the mass human review requirement, often using outsourced managed document review (MDR) to reduce cost and improve consistency.
The end is not (yet) nigh
What does this mean, not for the lawyers, judges or regulators but for the people who matter most: the clients who pay the bills? In our experience, it means that there is no single silver bullet. TAR does not signal the end of lawyers (yet).
Many companies are looking to further disaggregate, disintermediate, outsource (and ultimately automate) the provision of some legal services. They need a balanced armoury, from the rapier of artificial intelligence (including TAR) to the bludgeon of MDR. Clients are looking to apply those techniques in combination across the portfolio of their legal risks – not just litigation and regulation – in order to more effectively manage their costs while delivering greater consistency and quality of output to the business and to external stakeholders.
So the answer to the question posed in the title is: no. Robolawyers are already here. But Pyrrho might just create a few more.
Nah. This just means fewer trainees and paralegals with eyeball screen-burn and extreme boredom trying to review 20 documents an hour coding them for relevance, issues and privilege (and getting it only partially right most of the time). At least the errors the predictive coding makes (and it will make a lot) can be corrected by code refinements. With trainees/paralegals you are in the hands of [please insert Deity of choice according to diversity policy].