It’s difficult to discuss the future of legal innovation nowadays without some mention of artificial intelligence, the applications of which are almost ubiquitous in the development of new software solutions targeted at the professional services industries. When you consider the significant productivity benefits boasted by these tools, the hype is unsurprising … but not all AI is created equal, and a lack of clarity around precisely what it is and how it will affect these industries has led to a tendency for its disruptive influence to be unnecessarily sensationalised.
Generally speaking, AI falls into one of two theoretical categories: pragmatic AI or pure AI (often referred to as narrow AI and open AI, respectively). It’s a distinction that has been recognised for some time now, but one that has failed to gain any real traction in the legal tech space in recent years. So what is exactly is the difference?
Pragmatic AI refers to applications of artificial intelligence that are producing real business value for end users trying to work more efficiently and manage their big data challenges. It describes a collection of technologies, or ‘building blocks’, which can be readily deployed to help speed up standard business processes whilst leaving ultimate control in the hands of the end user. Pure AI on the other hand refers to the development of technologies that will eventually be capable of perception, interaction and other human-like cognitive abilities – the somewhat daunting prospect of a future in which artificial intelligence matches or surpasses human intelligence. It’s the apex of AI, the true actualisation of which we have yet to witness.
It’s important to reiterate that these are purely theoretical terms, and whilst they have no technical significance in describing the underlying principles that facilitate the various applications of AI, they do prove useful in mediating some of the hype and describing the state of AI in practice today. Pragmatic AI is about automating straightforward tasks through the use of ‘building blocks’ such as natural language processing and machine learning which, despite the discreet nature of their applications in practice, are still fairly complex capabilities. That’s why we like to take the concept of pragmatic AI a step further to include additional techniques that are even more straightforward, and yet still highly performant. These truly pragmatic techniques include identifying named entities by locating words or strings of words beginning with capital letters, and the use of predefined or user maintained dictionaries which identify entities such as postcodes, email addresses, credit card numbers, and so on.
The relative modesty of this kind of AI, at least when taken at face value, is not to say that its implementation cannot be substantially beneficial. In fact, in a legal context, the combined time savings of automating simple business processes like clause analysis and document comparison for contract review can be extremely valuable. The careful and considered application of pragmatic AI is, to us, the embodiment of the argument that technology should support lawyers, not replace them, and it’s a big part of the work that Nalytics is doing in the legal space.
The distinction may seem obvious, yet all too often the boundaries between pragmatic and pure AI seem to blur and, for one reason or another, potential business users are discouraged from adoption. Whether due to a misguided fear of replacement by technology, concerns relating to the perceived complexity of implementation, or a hesitance to give machines the ‘final say’, the same question ultimately has to be asked: if we were more transparent about what AI is and exactly how it is being used, would scepticism be as much of an issue? Professionals (legal or otherwise) don’t want machines to make big decisions for them, and why would they? That’s what humans are for; it’s what we’re good at. This is what pragmatic AI is all about: helping users take care of the menial yet time-consuming work faster so that they can get to the decision-making stage much quicker, not to mention encouraging the development of more informed decisions when that time comes.
As a society, we have a tendency to label things in order to better comprehend and process the relentless flow of information coming at us from all angles in an increasingly digital world. That’s why it’s somewhat surprising that this important distinction hasn’t been made more explicitly in discussions around recent AI developments in the legal space. Artificial intelligence is a topic which has caused considerable controversy within the industry - controversy which might be more effectively moderated with a little bit more transparency from industry vendors and critics alike.
When we start looking at AI for what it really is, rather than allowing an exaggerated perception of more revolutionary AI developments to distort our understanding (developments which are at least decades away from fruition anyway), the hype surrounding the burgeoning legal tech movement might soon be hype for all the right reasons.