Back in 2018, the "Horizon Scanning Forward Thinking" report from the Law Society of England and Wales offered ideas for a number of ways artificial intelligence (AI) could be used in the legal sector. This included to predict case outcomes or to power Q&A chatbots that would support client queries.
The report also pointed to examples underway, such as document analysis to draw out key findings, review contracts and then presenting their information in dashboards or even acting as a virtual legal adviser, reviewing relevant cases to find key judgements fast.
Three years on, Stuart Whittle, business services and innovation director at national law firm Weightmans, explains: "As it stands, AI solutions tend to do one thing, really well.”
The challenge for all those in the AI ecosystem is working out how to stitch these single point solutions together to deliver an end-to-end solution that becomes a seamless part of a lawyer's workflow.
"Achieving this will ultimately empower lawyers to spend more time delivering the valuable work they were trained to do and, importantly, that they enjoy doing."
He adds: "Firms have to navigate regulatory, contractual, tortious, fiduciary and commercial constraints when it comes to implementing the technology, and ultimately battle against what can be an inherent scepticism in the industry – and among clients – towards change."
However, there are now many systems already active and in place.
Mark Beer, commercial partner at Keystone Law, highlighted LawGeex for document review, GLS LegalSifter for automated contract drafting and risk analysis, Thomson Reuters Westlaw Edge to give predictive outcomes for disputes and RDO or Jur for resolving disputes online. He also cited an example from China where legal bots are able to address legal issues via massively-popular messaging platform WeChat.
But he said: "AI is already light years ahead of the legal profession's ability to use it. For many firms, Zoom is about as advanced as they want to get right now."
AI implementations in legal are gaining pace
However, Beer is optimistic for the future of AI in legal, adding: "The biggest advancement, and something we are working on at Oxford University's DeepTech Dispute Resolution Lab, is the use of big data to predict disputes years before they crystallise, allowing organisations to avoid [them]. This requires analysis of huge data sets, something quantum processing will assist."
Another example comes from April King Legal, where a digital legal assistant called Amelia has just been launched through work with IPsoft. The first phase of her implementation will see Amelia handle thousands of inbound customer enquiries for free wills, trusts and probate information packs.
The firm's CEO Paul King said the technology will help "improve our customer experiences and reduce administrative costs".
"The massive shift towards digital service consumption over the past year has transformed customer expectations for every industry. The legal sector is no exception; digital experience has become the new customer battleground," he says.
Brennan Ong is founder of LawAdvisor, a company working with a number of top international law firms to develop industry-leading technology that addresses specific issues facing today's legal teams.
Being socially responsible in the developing world
Sample our exclusive Business Briefing content
Ong says: "I think the ability to simplify law for the masses remains one of the biggest challenges of legaltech. We have seen this done successfully (if to a limited extent) in the finance and insurance sectors; however, the legal sector continues to remain inaccessible and expensive for a large part of the population. We are hopeful of addressing this area later in the year with a new solution.
"The legal industry is quite bespoke and risk averse. This means high expectations from AI tools in terms of customisability and accuracy. It’s not easy to achieve a 'plug and play' set up."
Other forward-thinking initiatives include Tech Nation's LawTech Sandbox, which according to Matt Shearer, CPO at Data Language, aims to "fast track transformative ideas, products and services in the legal sector, including AI".
He explains: "Legal firms that build their core expertise and differentiation into the design of their information management systems, at a granular and interconnected level, will create an advantage for themselves – both in their ability to adapt rapidly and their readiness to harness AI for new business models."
Battling bias and trust issues with legal AI
However, legal AI faces many issues, not least if it is programmed by a human with inherent biases. Other problems include trust and security.
Shearer suggests one route forward may be what’s known as explainable AI, as it’s able to offer "complete visibility of how and why an AI-powered decision or action is made".
He explained: "This is an approach to tackling the ethical challenges around machine intelligence in the legal sector, by making the decisions involved in AI-powered processes transparent.
"The challenge here is the tendency, which we have seen in other sectors, to create 'black box' AI services: A monolithic prediction approach that uses training materials to create a singular machine learning model to power a business process.
"This leads to decision opaqueness, which becomes an issue when we want to see why a particular decision, such as a mistake, was made. To mitigate this risk in legal AI use cases, an 'Explainable AI' approach can provide transparency and adaptability."
Sally Mewies, head of technology at international law firm Walker Morris, suggests one of the biggest issues for AI in legal is "that judgement is a big factor in advising a client".
She explains: "For example, there are often many solutions to a problem and in certain circumstances the correct solution can come down to an emotional issue, rather than a logical one.
"There are also potential security concerns around granting third party AI tools access to confidential information and it's unclear whether it’s yet widely trusted in the industry. It becomes a sub-optimal tool if the AI's output needs to be checked by a lawyer prior to sharing with a client."
Some within the profession have also admitted many solutions need a huge amount of effort to tailor products to be useful to clients, wiping out the benefit. So in an industry with so many manual processes, how easy is the transition?
Mewies adds: "It is not easy because the law has been practised in the same way for many years and law is probably one of the least developed industries from a technological perspective.
"Many of the most successful AI tools will target specific challenges, such as streamlining aspects of the time recording or file opening process, but right now it will not be used to introduce sweeping changes to the way that legal advice is given."
