Mishcon de Reya chief strategy officer Nick West discusses the “incredibly exciting” field of artificial intelligence in The Lawyer’s 60-second interview on key topics from the upcoming Business Leadership Summit.
Why do you think technologies such as AI and predictive coding are apparently gaining momentum in term of their uptake in the UK legal market?
First and foremost, the legal market isn’t an isolated bubble. There is so much happening right now in the world at large about AI and cognitive computing that the legal market simply can’t be immune. Our behaviours in the workplace are driven (and increasingly so) by our experiences in the rest of our lives, so when there’s a constant stream of mainstream news about AI, it’s inevitable that it must impact the legal market.
Add into the mix a number of other legal market factors – the increasing price sensitivity of clients, the explosion in data, the long-standing feeling that there must be a better way to do repetitive knowledge tasks than simply adding more junior lawyers – and you’ve got the perfect conditions for these technologies to take hold.
Do you have any experience of using AI or seeing it in action? Were you impressed?
Yes, absolutely. At Mishcon, TAR (technology assisted review, including predictive coding) has been a key feature of our approach to litigation for several years and with our recent launch of Mishcon Discover (our managed e-Disclosure offering, in partnership with kCura and Unified), it’s now standard practice on most disclosure exercises.
The technology is getting better with every release, particularly concept searching and clustering using latent semantic indexing/analysis type capabilities. There’s no doubt that TAR drives significant savings in speed and cost and, critically, leads to better outcomes for clients.
Outside the litigation arena, we’re also working with a number of AI/cognitive search players on transactional diligence exercises, to increase the speed and efficacy of contract review. One of the biggest benefits I’m seeing of using these new tools is that the data output is very flexible, so there’s a lot that can be done to integrate the data straight into client systems
What sorts of opportunities do they present for legal services providers to be seen to be at the cutting edge of a market or present themselves as forward-thinking or trail blazers?
Yes, the marketing opportunities are there, for sure, but for me, the more important goal is to get work done better and more productively – that’s a boon for our lawyers and our clients.
What do you see as the biggest concerns or risks about using these sorts of technologies?
At a macro level, I don’t really have any. Clearly, any technology used badly can lead to poor outcomes, so we’re very conscious to ensure that we have sufficient expertise, whether internal or external, to hand when we are working with new technologies. I think it’s also vital for us to test new technologies carefully – there’s often a big gap between the promise and reality – so, we’re careful to pilot things and keep the initial application to limited data before we begin to scale things up.
I’m also very conscious about education – both educating people as to what the technology can and can’t do, and also ensuring that junior lawyers still get sufficient theoretical and practical experience of the underlying legal tasks that we are looking to use the technology for, because in my mind, the winning move is the combination of legal and technology expertise – you need to have a good degree of expertise in a particular legal task to be able to train cognitive technology correctly.
What data exists on how much more accurate AI or predictive coding is than more traditional methods?
There aren’t huge numbers of studies because it’s not an easy task to set up an controlled experiment at statistically relevant scale of human vs technology assisted review, but the work of Grossman and Cormack in the US is worth spending time on – giving quantitative proof that technology assisted review yields better results with less effort and that continuous active learning methods are the best. Grossman and Cormack’s work was heavily cited by Judge Peck in the seminal 2012 US case (Da Silva Moore vs Publicis) in which TAR was first approved.
What do you think the future looks like in terms of the uptake of technologies like AI and predictive coding in the UK and what is that likely to mean for the legal market?
It’s an incredibly exciting field – we’re only just seeing the start of it.