In 1818, Mary Shelley published her novel, Frankenstein. This was perhaps the first modern conception of an ‘artificial intelligence’. Of course, when it comes to the legal services industry, when we talk about AI and the ‘new digital frontier’, as the World Intellectual Property Organisation’s (WIPO) Director General Francis Gurry described it, we are thinking less about an artificial ‘sentience’, and more about how computers can help improve process. AI in law means: developments in machine learning like NLP, intuitive software that allows firms to pool and manage content and resources, new workflow tools to aid the administration of multinational businesses, and the automation of court processes. Still, concepts of a walking, talking AI are no longer the stuff of fiction. In China, AI-powered judges are already a reality, able to voice questions that flesh-and-blood judges would normally ask and call upon algorithms to give out judgements.
The gap between what we can conceivably create and the speculative worlds of science fiction authors continues to narrow, and the future of law is likely to contain more AI, not less. Is this a good thing or a bad thing — and how can the legal services industry harness the best bits to enhance the services it provides? A good place to start is WIPO’s 2019 report on trends in artificial intelligence. This report gives a breakdown of the number of patents filed in the various sectors of AI, which allows a direct insight into what types of AI are the market’s busiest. Unsurprisingly, machine learning is highest, representing 89 per cent of filings mentioning this AI technique and 40 per cent of all AI-related patents. How this breaks down and how wide the net is cast when talking about machine learning is uncertain. However, the report also notes that computer vision was mentioned in 49 per cent of all AI-related patents. Deep learning is the fastest growing technique in AI, with multi-task learning second.
This analysis reveals a great deal for the legal community. While NLP and machine learning can help to streamline processes and improve workflow, of more interest is the effective rollout of deep learning and multi-task learning which could herald a seismic shift in the legal services industry, as many of the tasks traditionally given to junior lawyers and paralegals can be done just as effectively (and, crucially, more cheaply) by AI. Deep learning is the gateway to self-improving AI that learns as it goes. This could mean that in law firms image analysis, contract review, and due diligence even of a more complex sort could be hoovered up by AI within a number of years.
On computer vision and image processing, this could prove crucial for both commercial and criminal lawyers. For the former, more developed AI could mean processing case materials that contain elements other than pure text more quickly and efficiently. For the latter, better AI could mean the end of poor facial recognition software that has been known to foster dangerous and damaging racial bias and discrimination due to imperfect algorithms and data input that retains historic bias from police forces.
On algorithms, this then begs the question: to what extent can they benefit or hinder not only the backroom functions of law firms, but their journey through the judicial process in the courts? Litigation prediction machines that have already been built claim 90 per cent accuracy in determining the probable outcome of a case based on the facts.
In turn, this is fuelling the litigation funding sphere, which can increasingly rely on data-driven and statistical analysis of whether a case is worth financing. If we take as our starting point that all should be equal before the law, and justice should be accessible to all, how do we square this with a marketplace in which it is increasingly easy to hunt out the likely winners, and conversely identify those where it is almost impossible to develop a winning strategy and therefore may not represent good investment? Seasoned litigators may argue that litigation prediction machines simply allow any potential client the best possibility of winning their case by identifying the strongest strategy early on, but others may suggest that this form of AI opens the door to scrutinising would-be clients on the basis of potential success and revenue for the firm, determined by an algorithm.
Of a greater philosophical concern is the use of such predictive machinery to analyse the likely outcome of a judicial decision based on that individual judge’s past decisions. The possibility of ‘gaming the system’ has led France to legislate against this sort of technology, whilst many multinational law firms are in the process of perfecting their own in-house software to offer them exactly this kind of data-driven analytics. Again, seasoned practitioners might argue that even in the old days you knew your chances against one judge versus the other and would plan accordingly; codifying this via statistics is not a quantum leap. The French decision, to allow such data to be ‘open source’ but provide regulatory safeguards on its usage, sounds like a reasonable middle ground.
Given such developments in AI over the last few years, it is unsurprising that the number of ‘AI in Law’ seminars, lectures and University courses has boomed. The University of Oxford’s Unlocking the Potential of Artificial Intelligence for English Law project, which includes research backed by UK Research and Innovation (UKRI), gave birth in December to an education programme consisting of six modules designed to help the legal sector fill gaps in their AI knowledge. The ultimate aim of this programme is to facilitate and accelerate the uptake of technology by the legal industry, and this is apparently just the start.
But let’s not get ahead of ourselves. Research conducted in March 2020 showed that only 25% of solicitors surveyed had used AI-backed software. Whether this has changed thanks to Covid-19, and whether this is partly due to some solicitors having used AI-backed software without having known about it is uncertain. One of the biggest barriers to increasing this number seemed to be a lack of adequate training — and so we can be thankful for the boom in new initiatives from reputable providers.
Luciano Floridi, Professor of Philosophy and Ethics of Information at the University of Oxford, wrote: ‘the digital revolution transforms our views about values and priorities, good behaviour, and what sort of innovation is not only sustainable but socially preferable – and governing all this has now become the fundamental issue’. Such a conversation is one that legal service providers who want to be at the cutting edge will make sure they engage in.
All opinions my own.