Lines of binary code spiraling into an image of the globe

The Rise of the Machines – Predictive coding and disclosure

Elon Musk has described Artificial Intelligence as “summoning the demon”. Others see it as one of the biggest threats facing humanity today, alongside giant asteroid strikes and nuclear armageddon.

Far more importantly, how will the Internet of Things affect litigation lawyers in this brave new world?

The disclosure of documents remains the bedrock of litigation. It usually makes or breaks a case. But finding these case winning documents is often a long and arduous task.  Many a lawyer will recall sleepless nights trawling through hundreds and thousands of pieces of paper to extract relevant documents. This was just as painful an exercise for those clients who had to bear the cost.

As technology has advanced the volume of potentially disclosable documents has also increased exponentially. Emails and SMS sent frequently and often informally within a business, have taken the place of the traditional letter or memo. Relevant documents are usually now longer archived in boxes or files but instead might be stored on a server, a USB stick, in a cloud or in numerous other electronic locations.  This has meant that parties to litigation have needed to adapt and change the way in which searches for documents are undertaken.

So far machines and lawyers have been rather suspicious and uncomfortable bedfellows. Although technology is available to assist in this quest to find disclosable documents, this has been embraced too slowly by lawyers and the courts.  Specialist e-disclosure platforms can be engaged to collect and analyse electronic documents. Searches and filters are applied to the data using key words and date ranges to find those potentially of most significance to the dispute. But this can still be a fairly crude process which fails to pick up key documents and is still heavily reliant upon a lengthy further review by a legal team so that the documents which should (or should not) be disclosed.   Those who are undertaking that review process are often inexperienced junior paralegals who consider the relevance of one document at a time (much like going through a lever arch file of printed material).  Although the filters will have removed some of the irrelevant documents this task can remain a time consuming and expensive exercise and there remains a large capacity for human error to fail to identify key documents.

Signs of a more collaborative approach between the lawyer and the machine are however starting to emerge. In a recent court judgment (Pyrrho Investments Limited v MWB Property Limited) it was held that the parties could use predictive coding or a “technology assisted review”. Predictive coding is a bit like online shopping – once you have made your product selection up pops further suggestions of other similar items. In simplistic terms, predictive coding involves:

  • Filtering the documents collected as described above to get a target set of documents;
  • A lawyer, reviewing a sample set of the documentation, “teaches” the software how to assess the relevance of the remaining documents;
  • Analysis of the documents is undertaken by the software which applies algorithms to the coding decisions to identify those documents of relevance;
  • Reviewing the results and refining the coding until the results which the computer returns are satisfactory.
  • Ensuring, via human review, that the documentation selected falls into the client’s disclosure obligations.

In the Pyrrho case there were a staggering 3.1 million documents and the disclosure process would have been impossible without the assistance of technology.  More recently in Brown v BCA Trading Ltd the Court ordered predictive coding after hearing that it would cost around £132,000 compared to  a traditional keyword search which would cost at least £250,000. There was nothing to suggest that predictive coding would be less effective.   The Court noted that predictive coding was new so there was bound to be some concern that it might be less effective than more traditional, but more expensive, methods. However, the parties had to do their best to achieve reasonable and proportionate results.  

We are likely to see predictive coding be used more in the future and it is certainly something to be embraced. Key documents will still need to be reviewed by a lawyer.  But technology can help speed this process up, and remove the likelihood of human error getting in the way. This of course will save a vast amount of time and cost, but the lawyer still remains a central and key component to a process which is quicker and more cost effective than a full human review

As Aldous Huxley wrote in “A Brave New World” We are not our own any more than what we possess is our own. We did not make ourselves, we cannot be supreme over ourselves. We are not our own masters.” At least for the short term, the future for litigation lawyers is not servitude or extinction but a more collaborative relationship with the machines and technology for the ultimate benefit of our human clients.