Artificial intelligence and the art of noise over substance
This is the first in a series of articles, initially published in The Global Legal Post, on various technical themes in language which can be understood by those who prefer to use technology rather than immerse themselves in it. These are written by Paul Longhurst at 3Kites with Kemp IT Law contributing.
For our first topic, we are discussing Artificial Intelligence or AI. The reason for the slightly contentious title is that many firms make a great deal about their AI initiatives whereas in truth this is often a case of catching the attention of prospective clients by appearing to be investing heavily in this field when the reality is often little more than a pilot project – or ‘innovation theatre’ as it’s unkindly, but sometimes accurately, called. Of course some firms have gone much further than this and invested heavily in real line-of-business solutions which seek to reduce the time taken to respond on matters or to lower costs where this can be done without affecting the quality of advice. However, all too often we hear about AI initiatives without seeing the real benefits – it is as if someone has arrived at an answer without first hearing the question. So what is all the noise about and where can it really help?
A definition is often the way to a lawyers’ heart so here is one for AI – systems that are able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages… the apparent ability to think and learn. Machine learning is a form of AI whereby systems adapt to reflect what they ‘learn’ about a specific area. One example of this would be having a system analyse large numbers of MRI scans along with the results of any abnormalities found so that it can understand what a clear scan looks like as opposed to one that clearly identifies, say, malignant growths and, most importantly, those giving an early indication of problems thereby providing the greatest chance of successful treatment. Humans will struggle to analyse as many images and identify subtle patterns which point to early diagnosis whereas a system accelerates this process by using human-learned skills in volume and at speed.
Translating this to the legal sector, systems can be tuned to analyse case files in order to identify patterns which are present whenever these cases are clearly lost or won allowing firms to arrive at a decision for, say, insurance clients which save the expenditure of time on cases that are likely to be lost. It may be that some assessments are incorrect but if, on balance, the vast majority are right it will be an effective tool in an increasingly competitive market. And what of those cases where the outcome is not clear ? This is where judgement can be applied by the experienced lawyer, where their skills really make a difference and where clients will see the benefits of extra effort. In 3Kites’ AI seminar in 2017, we presented an example of this approach whereby Allianz in France analysed case patterns to provide an indication of settlement costs and timescales as opposed to trial durations and damages if the same cases were contested – this provided case handlers with helpful, experience-based guidance when deciding which path to take.
AI can also be usefully applied to the extraction of details from documentation in order to shortcut the (initial) analysis that would otherwise require a degree of human effort which can be subject to increasing inaccuracy as data volumes increase. Our 2017 seminar covered examples from:
• Vodafone (where contract management uses automated triage to help determine if lawyer input is needed),
• Riverview Law (which, before its EY acquisition, automatically allocated inbound documents to the right department based on extracted data), and
• Brandwatch (where the use of self-service contracts in this social intelligence business is supported by a system identifying what is outside pre-approved parameters).
Other relevant examples of AI include legal research (IBM’s Watson has been used in a US initiative called Ross which has ingested bankruptcy and IP data to reduce research times from c. 10 hours down to a few minutes), due diligence analysis and abstracting key terms from complex agreements such as ISDAs to provide improved visibility of obligations and risks. These are some of the practical applications in use with law firms today, and that really is the point here – they are practical and providing real value at scale… not simply making noise.