The dawn of the artificially intelligent lawyer
With the best human player of boardgame Go being defeated this month by Google software, lawyers shouldn't make the mistake of believing their skills are beyond the reach of artificial intelligence (AI), write Stéphane Leriche and Michael Stojanovic.
In early March, Google's DeepMind and its AlphaGo software beat the best human player of the ancient Chinese board game, Go, in a series of much-hyped matches. AI-powered legal tools are already a reality, and the success of AlphaGo demonstrates that the technology to make these tools better now exists.
IBM's Watson cognitive computer won the TV game show Jeopardy! in 2011. Watson's question-answering technology has since been adapted to produce ROSS, a legal research virtual assistant that its developers promote as a 'super intelligent attorney'. While there may be an element of mere puffery about that description, any law firm with an eye to the future should take note that such technology is available today, and can only improve tomorrow.
The AI lawyer of 2016 is designed to respond to simple legal questions, and to provide comprehensive supporting references to relevant statute and case law. In short, these systems can perform the sort of tasks which trainee lawyers and paralegals have traditionally performed by poring through text books and case reports over hours or days.
Existing AI technology cannot analyse and give advice in the context of a particular client matter. But one apparent advance of AlphaGo has been the software playing "creative" moves that surprised both its opponent, and its designers. If software can now display unexpected insight and creativity, it may only be a matter of time before legal services software will perform the types of legal reasoning tasks which are currently the domain of senior legal practitioners.
The question is not any longer “if” but “how” AI will affect the legal profession. In the short term, the need for junior lawyers and paralegals to perform research tasks may be reduced, both exacerbating the existing oversupply of law graduates, and reducing the availability of research and procedural work that young lawyers have traditionally "cut their teeth" on.
Law firms will need to consider how they will "skill up" their future junior lawyers, and perhaps accept that junior lawyer training costs will go straight to the budget bottom line, with little scope for recovery from clients.
AI also has financial implications – in a profession built on billable hours, systems that generate near-instantaneous results are inherently disruptive to the law firm business model. It doesn't require a science fiction writer to anticipate systems within the next few years generating first-draft contracts, court documents or advice in the blink of an eye.
While charging for these services on a per-task basis might seem an obvious response, the prevailing trend from the big technology companies is towards providing once prohibitively expensive services for free or at a very low cost. Might AlphaGo lead to AlphaLegal? Will clients always value top-class legal services, or for the price, will they take the view that something generated by an AI system is good enough?
It seems very likely that experienced human lawyers will increasingly spend their time checking over the output of AI systems, and perhaps providing the "human touch" for subjective advice on developing or untested areas of the law. Of course, as any lawyer can tell you, the reasoning of the man on the Clapham omnibus can be elusive to even the finest human legal minds – can we really assume it will never be grasped by artificial intelligence?
Stéphane Leriche (left) is a partner and Michael Stojanovic (right) a senior associate at international law firm Bird & Bird.