Hardwicke barrister Michael Wheater demystifies the much talked-about artificial intelligence process that is set to sweep the legal profession
With the recent decisions in Pyrrho Investments v MWB Property and Brown v BCA Trading Ltd confirming that predictive coding could amount to a reasonable search and is permitted under the Civil Procedure Rules, lawyers are bracing themselves for change.
At the same time, they are also having to cut through plenty of hype about a process which, despite its use of artificial intelligence (AI), may not quite herald a new dawn of robots taking over the world.
Legal Cheek Careers caught up with Hardwicke barrister and legal tech specialist Michael Wheater to find out about what predictive coding actually means.
Legal Cheek Careers: What is predictive coding?
Michael Wheater: “Predictive coding involves a human reviewer ‘training’ a computer system to identify and classify relevant documents within large volumes of data. To train the system, a more senior lawyer who is familiar with the case reviews one or more initial batches of documents and ‘codes’ each document for relevance, privilege and other specific issues. The system then uses complex algorithms to extrapolate the reviewing lawyer’s decisions and apply them to the remainder of the document set.
“In some ways it’s AI, but a lot of the terminology surrounding its use in relation to edisclosure is unhelpful. ‘Machine learning’ is the technical term that we often hear, which conjures up the idea of predictive coding doing things ordinarily that lawyers would do. That’s not quite the case.”
Does predictive coding require lawyers to be able to code?
“The word ‘coding’ is slightly misleading. In disclosure exercises, coding doesn’t mean writing computer programmes, rather it means tagging documents on a document review platform. The set up is almost a tick box questionnaire on the side of the screen. Is it privileged? Is it relevant? These are the questions that are being flagged up. That process of tagging documents with information is referred to as coding. Accordingly, lawyers will have to be trained to use the software, but that will be no different to, for example, being trained how to use Excel or a case management system.”
Is predictive coding bad news for junior lawyers?
“Probably not. Traditionally, the way of carrying out large scale disclosure and document review has been by trying to identify relevant documents (sometimes millions of electronic documents) by performing a series of key word searches. Whatever results are generated by those searches are then loaded to a document review platform. Then a roomful of people — usually paralegals, trainees and junior lawyers — read through each document to work out what is relevant and what is privileged.
“It’s an extremely costly process and also fairly inaccurate and inconsistent. At the same time, it’s interminably boring. I’ve run disclosure rooms before and it’s seen as almost a necessary evil. Very few lawyers learn their craft that way. Predictive coding allows legal teams to identify relevant documents more quickly. The documents identified still have to be reviewed and analysed as part of the case preparation and this will still require the involvement of junior lawyers. As a result, predictive coding should provide trainees and junior lawyers with a better base on which to build upon the sort of work they are already doing. But if there is a wide uptake of predictive coding, it may well render some of the traditional ‘disclosure room’ roles redundant.
“However, bear in mind that predictive coding is really only being used in relation to litigation. I imagine that transactional work will remain largely as it is, because it doesn’t tend to generate anywhere near the same amount of documents to be reviewed (save in major due diligence exercises). Corporate law firms’ elibraries and knowledge banks are also fairly well developed already.”
What’s the background to all of this?
“Principally, it’s the huge increase in documents that are being created, largely due to email and now smartphones. In the past, people engaged in litigation would write one long letter. Now they write 50 emails, send texts or leave voicemails. In the past, important or significant documents would be retained on a hard-copy file; now the cost of electronic storage is so cheap that virtually everything can be retained. A study by IBM in 2012 showed that 90% of all the documents in existence anywhere in the world had been created in just two years between 2010 and 2012.
“The upshot of this is that disclosure and document review has become a major issue for lawyers, and at a time when the government is keen to bring down litigation costs, the courts are now lending their backing to technology that helps to balance the need to identify relevant documents with spending proportionate time doing so and incurring proportionate cost. It’s worth bearing in mind that some judges are quite ahead of the game on this stuff and are embracing the technology ahead of lawyers.”
There is chatter around the Silicon Roundabout that legal tech could be the next big thing. What’s your view?
“The legal tech market is already huge and a lot of the companies involved — many of which are based in the US — are investing a fortune in new products. So while everyone is talking about predictive coding at the moment, in five or six years there may be something completely new. Meanwhile, the technology is getting cheaper and easier to use. So with lawyers having always been slow to embrace what they don’t fully understand, you’d imagine that there is some real potential there. I would have thought that the next areas ripe for expansion are likely to be in electronic trial bundling and case presentation and in providing streamlined collaborative working environments tailored for the legal market.”
Michael Wheater is a barrister at Hardwicke.