Created with RPC

Why the ascendence of AI can benefit young lawyers

Avatar photo

By Thomas Connelly on

Ahead of ‘The future of litigation’ on Thursday, RPC’s Daniel Wyatt tells Tom Connelly how the Pyrrho case is redefining what it means to be a litigation lawyer

Everyone’s talking about artificial intelligence (AI), about what it can actually do, about how far it will go, and about whether trainee lawyers and paralegals’ jobs are at risk. Daniel Wyatt, a litigation and arbitration specialist at RPC, has witnessed the rise and rise of AI within the legal sector first-hand.

Wyatt — who joined RPC as an associate in 2012 — recognises that AI’s impact on litigation and the legal sector is hugely important but stresses that it is more about a change in focus, not the death knell for junior posts. Even better, it could mean less time spent in gloomy datarooms:

There will always be a need for junior lawyers. I suspect that the impact of AI will be more a case of re-distributing the work that paralegals and trainees undertake. This will actually allow them to focus on the more interesting aspects that a role in litigation can bring.

Wyatt was first introduced to AI when he acted for the claimants in the landmark 2016 case of Pyrrho Investments v MWB Property. With over three million documents to be considered for relevance and disclosure, the High Court’s Chancery Division ruled that RPC’s clients — in what was a first for the English legal system — could use predictive coding software to sift through the vast quantities of potential evidence.

“Behind the scenes, predictive coding is essentially a set of techy-algorithms”, Wyatt says. Broadly, it involves a computer being ‘trained’ by a lawyer who reviews batches of documents for relevance, usually around a few thousand per batch, before the software extrapolates those decisions on relevance across a wider collection of documents. Revealing that this was his first hands-on experience of this form of legal-tech, he explains:

The lawyer, who will usually be quite senior and who will have an in-depth knowledge of the issues in the case, will review each document in the initial batches assessing whether each is relevant to the particular issues of the case. These human decisions are logged by the predictive coding software, allowing it to learn what sorts of concepts and phrases the lawyer considers relevant. Methodologies vary but essentially this learning process — which was approved by the US courts in 2012 — is repeated a number of times to ensure the system fully understands what to look out for — it can then go out and look for similar documents in a far wider dataset.

Prior to the development of AI and predictive coding the disclosure process was becoming extremely laborious. The “old fashioned way”, as Wyatt describes it, was to “use human review”. Utilising an army of paralegals and trainee lawyers, documents would be checked for relevance and disclosure by hand. This was all very well when our world was largely paper based. But as electronic documents became more and more prevalent, it has become less and less manageable and is now a process that could take several months to complete (or even, in the biggest regulatory investigations, years).

A step up from this — which is less favoured now thanks to AI — was to conduct “keyword searches” against electronic documents. Using Boolean codes (true or false responses) based around the issues of a case to identify documents that might be relevant to the case, Wyatt explains that this rather “simplistic approach, whilst pretty unreliable, was often the default way for parties to seek to reduce the number of documents to be reviewed”.

For example, if lawyers were looking for documents containing the name of a particular company, and that company has its name in the footer of all of its staff emails, early keyword searches would flag up every piece of correspondence. This is clearly not helpful.

The Pyrrho case was particularly suited to predictive coding because, Wyatt reveals, “almost all of the documents were held by one party, one of RPC’s clients.” With large volumes of data to process, and disclosure obligations very one-sided, AI software was a practical and sensible way to lighten the load for RPC’s litigation team to ensure their clients could comply with their disclosure obligations in a cost effective manner, whilst maximising the prospects of finding material relevant to the case.

Predictive coding will lead to greater consistency and cost-savings, the judge in the Pyrrho case found. It is clear then that AI is here to stay. But the message from Wyatt — who studied law at the University of Newcastle and graduated in 2007 — is that concerns among law students about AI are understandable but probably overblown.

Aspiring lawyers who are entering the profession in this new tech-age do not need to start learning how to code algorithms just yet, says Wyatt:

Students, as part of their wider commercial knowledge, should be aware that predictive coding is likely to play an ever-increasing role in the litigation process and is likely to be more widely used in other areas of the legal sector too before long. Showing an appreciation for the benefits an AI system can bring is valuable, but knowing how the software works behind the scenes certainly isn’t necessary. Thankfully RPC has IT experts to handle this.

Dan Wyatt is an associate at RPC in London. He was the lead associate acting for the claimants in the landmark Pyrrho predictive coding judgment, and will be speaking about his experience at ‘The future of litigation’ on Thursday evening.

About Legal Cheek Careers posts.

For all the latest commercial awareness info, and advance notification of Legal Cheek's careers events:

Sign up to the Legal Cheek Hub