By Stephen Turner, founder – Lawyers of Tomorrow
“In the future robots will do anything a humans can come up with and the most successful people will be the ones who can optimize their robot’s activities”
Kevin Kelly, writing in ‘The Inevitable: Understanding the 12 Technological Forces That Will Shape Our Future’
This is the third part of a six-part series looking at the question, ‘Will artificial intelligence (AI) replace lawyers?’. In Part 1, I looked at the views of the experts and in Part 2 I looked at when the wave of new technology might sweep through the legal sector.
In this part, I look at legal technology in overview and identify recent technological innovations and the drivers behind them. I will also look at what we actually mean when we refer to ‘artificial intelligence’ – an umbrella term – which is used to cover a range of systems: robotic automation, expert systems, cognitive computing and natural language processing. I will look at each of these in turn later in this post.
The billable hour and other barriers to innovation
Some will tell you that lawyers are simply not entrepreneurial. Others will say that lawyers are by nature resistant to change and are wedded to the ‘billable hour’ model, which discourages innovation.
Earlier today (as I type this post) I attended a fascinating webinar hosted by Ivan Rasic of LegalTrek when we addressed this issue: “Is the Billable Hour Slowing Down Innovation?” I’ll nail my colours to the mast for the record: there is a big incentive not to innovate with fixed fee pricing and new technology (which could improve your efficiency and lower the cost to the client) when you are on £500 per hour. In addition, many law firms are organised as partnerships – a structure not inclined to facilitate quick decision making – and are often controlled by senior solicitors, perhaps not that far from retirement, who may have no interest in investing millions of pounds in new technology which “might not work”.
In conclusion, I don’t think I’m sticking my neck out here if I say that, left to their own initiative, the vast majority of law firms would not innovate.
More work for less money
The financial crisis changed everything and left everyone short of cash. Clients wised up quickly and became irresistible drivers for change. They saw that law firms were essentially doing the same things, with little or nothing to differentiate one from another, and one thing to unite them all: the longer they took to do the work, the more they got paid.
Clients now demand that law firms do more work for less money and smart firms have realised that they have to innovate – or at least they have to appear to innovate. Some firms have improved efficiency via the use of technology and have passed the cost saving on to their clients but many firms have found it very hard to move away from the billable hour model. Others have only pursued innovation as a marketing tool. For these firms, the front end of the website may ‘talk the talk’ as far as innovation is concerned, but on the back end, it is business as usual.
A new attitude towards genuine innovation
Nevertheless, there is evidence that over the last two years attitudes are changing. Dentons have taken a truly innovative approach in setting up a wholly-owned subsidiary, NextLaw Labs, as a law-tech incubator, where research teams are free to try out ideas, see what works and learn from what doesn’t.
Law firms from an accountancy background have brought a new attitude of genuine innovation and many law firms now have dedicated innovation teams with considerable budgets.
There is also evidence of innovation in working preferences. Lexoo’s business model is rooted in the idea that there are some magic circle lawyers who do not wish to be partners or work ‘magic circle’ hours and who, for lifestyle reasons, decide to set up on their own. Lexoo has pre-approved a panel of 200 ex-magic circle associate lawyers and provides a platform for small and medium-sized businesses to access these lawyers – but at trainee rates – providing a compelling offer to those businesses using the platform.
“Add AI to it to make something new”
Increasingly, for every lawyer out there saying that AI is not going to take work away from lawyers there are others saying that this is exactly what was going to happen, particularly in relations to tasks where the work requires a low level of skill, but a high degree of productivity and or efficiency.
Kevin Kelly, the author of the bestselling book – The Inevitable: Understanding the 12 Technological Forces That Will Shape Our Future says that:”
“AI is as fundamental as the invention of printing, easily – maybe even as fundamental as the Industrial Revolution.”
“It will be a utility like electricity that you will just purchase from the cloud and the real opportunity for entrepreneurs is to take something and add AI to it to make something new.”
In addition, Kelly is in no doubt about the effect that the robots will have on the work that we do:
“All the jobs that are going to the bots are jobs that can be defined in terms of efficiency and productivity. If efficiency is all critical to this task – that is a task that goes to the bots. If productivity is all concerned with this task, it goes to the bots. What humans will be about is other things where inefficiency is a large component…human relationships are inherently, terribly inefficient. That kind of inefficiency is actually what we are good at.”
And of the question, ‘Will AI replace our jobs?’, Kevin says:
“No, AI will replace tasks. Jobs are made up of a bunch of different tasks and a lot of different tasks will go but we will have to work with the AIs because they think differently than humans.”
“You will be paid in the future for how well you work with AI. We are going to work with them.… There will be AI whisperers and they will be highly paid for being able to work with AI.”
Types of legal technology
In part 4 of this series, I’ll take a look at the legal sector and identify the innovators and early adopters and AI technology that each is currently deploying and assess the degree to which the innovations are now doing work that was previously performed by humans.
For now, in order to put the innovations into context, here’s a summary of the different types of legal technology currently in deployment and some working definitions.
These include cloud storage tools and security solutions. Essentially, the position here is that having such solutions is now non-negotiable and so firms have adopted such solutions in order to retain their current relationships
Support process solutions
These provide law firms with increased efficiency in case management, business development, customer relationship management, finance, billing, accounting and other areas.
Most, but not all, law firms use these solutions but there is considerable variation in the degree of sophistication of the systems in deployment. Within any particular firm, you might find a patchwork quilt of solutions which may be integrated to a varying degrees. In some firms, each element may operate more or less independently of the others with little or no integration.
The legal profession has been very slow to adopt support process solutions that are the industry standard in other market sectors.
Support process innovations can be enhanced by robotic automation such that an automated virtual system interacts with the user interface of a third party application (e.g. to enter data, or issue purchase orders) as a human would. The only difference is that the automation uses a virtual mouse and keyboard rather than a physical one.
The key point to appreciate with robotic automation in this scenario is that, rather than configuring the robot automation with code-based instructions and having it communicate with the third party application in code (which would require the skills of an IT professional) a robotic automation can be trained and configured by anyone so that the automation follows the exact same steps that a human would follow. This makes the robotic automation very easy to train since the non-technician human user can do the training in an intuitive manner, just as they would when training a colleague.
The big difference will be that the robotic automation will then perform the process solution much faster and it won’t make any mistakes, or get tired.
Substantive law solutions
These are highly innovative systems (incorporating AI – e.g. expert systems or cognitive computing and natural language processing – see below) which either support lawyers in key transactional or litigation tasks or in some cases replace them altogether.
There are a wide variety of solutions on the market, though the adoption rates for this category of technology are far below enabler technologies and support process solutions.
Some substantive law solutions deal with standard/commoditised tasks which might typically be performed by low-skilled legal workers.
In contrast, other solutions deal with much more complex tasks such as data analysis and summary, including the identification of risk – say, the percentage chance of a client finding a successful outcome in litigation, based upon a statistical analysis of previous decisions of a court.
Substantive law solutions will be the focus of discussion in part four of this series of posts and it is here that the cutting edge AI innovations are being made.
At this stage, it will be helpful to distinguish between expert systems, cognitive computing, and natural language processing (NLP) since all play key roles in delivering the various forms of substantive law solutions and all fall under the umbrella term, AI.
For a more detailed analysis of the different types of legal technology and the likely impact on law firms, see the excellent guide published by the Boston Consulting Group in conjunction with Bucerius Law School – How legal technology will change the business of law.
Expert systems are computer systems that mimic the decision-making ability of human experts according to ‘if X then Y’ style rules which are programmed into the system by human designers along with the system logic.
Expert systems are designed to operate within a ‘domain’ of expert knowledge which comes from human experts and also the internet, magazines and books. Automation is crucial to the functioning of expert systems since automation provides large amounts of data upon which the system will base its decisions.
Humans then set the expert system a problem and it follows the system rules and logic and uses the knowledge and ‘inference procedures’ to solve problems by applying reasoning to the facts and data. The expert system then reaches conclusions, much like a human expert would.
An example of an expert system at work would be one performing research tasks, say multi-jurisdictional surveys, and then delivering tailored reports.
Whilst cognitive computing systems – known as ‘true AI’ or ‘hard AI’ – with their self-learning algorithms and self-directed study, get most of the media coverage, the vast majority of AI solutions now in deployment in the legal sector fall into the category of expert systems – known as soft AI – (e.g. those provided by Neota Logic) and I will look at these systems in Part 4 of this series.
Cognitive computing systems
These systems are often referred to as ‘machine learning‘ or ‘true AI‘ or ‘hard AI‘ and include platforms such as IBM Watson, the most well-known of the cognitive computing systems, which simulate human thought processes and utilise self-learning algorithms that mine data and recognise patterns.
Advances in NLP mean that it is no longer necessary to force humans to think like a computer because cognitive computing systems interact with humans on their own terms and understand a human speaking in their own natural language. IBM Watson can read and it can understand tweets and articles – it understands all the written data which makes up 80% of all knowledge in the world.
IBM Watson is famous for winning the quiz show, Jeopardy, in 2011, by beating the best human players of the game. In order to win, Watson had to be able to understand the cryptic and contextual questions asked on the show and it had to be able to give very specific answers to those questions – and it had to do this quicker than the two best players that ever played the game. The questions asked on Jeopardy can be on any topic, so it was impossible to teach knowledge to Watson with a view to it being able to regurgitate that knowledge. Instead, what Watson did was to extract the knowledge from huge volumes of diverse and unstructured documents. It then discovered patterns and relationships within the data and revealed its insights based upon its self-learning of the knowledge buried within the data.
In contrast with expert systems, the developers of cognitive computing systems do not give the system any rules to follow nor is there any domain knowledge which is taught to the system – nor do they explain a set of steps that the system should follow. Instead, the system is left alone to learn under its own direction. Such systems are therefore entirely self-taught.
Automation in relation to data flows is equally crucial to cognitive computing as it is to expert systems. The more data a cognitive computing system is exposed to, the more it learns, and the more accurate, faster and smarter it becomes over time. Marketing Week reported in May 2016 that Watson is
“2,400% smarter today than when it achieved the Jeopardy victory five years ago.”
The IBM Watson Developers Cloud invites developers to:
“Enable Cognitive Computing Features In Your App Using IBM Watson’s Language, Vision, Speech and Data”.
All the technology and tools that developers need to take advantage of Watson’s cognitive power are a click away. That’s a hugely powerful support network for any developer who wishes to work with Watson to break new ground.
However, so far, only US law firm Baker & Hostetler has taken up the challenge, signing up with Ross Intelligence to use their Watson-based platform in bankruptcy matters.
Ron Friedman, Ph.D., the award-winning social psychologist and author has highlighted what he believes are “business model challenges” of deploying Watson, which he sees as a “big time investment”. He concludes that when measuring the cost/benefit of investment in Watson against investing in available alternatives – i.e. expert systems such as Neota Logic – Watson comes up short. However, as I explained in Part 2 of this series, we are still in the early stages of the adoption curve for AI and so things are inevitably going to develop.
In part 4, I’ll look in detail at the innovators and early adopters in the field of AI, including taking a look at the platforms provided by Kira Systems, Ravn, Ross Intelligence, Riverview Law, and Neota Logic.
Will Artificial Intelligence Replace Lawyers? Part 4 – Innovators, early adopters and AI whisperers – coming soon!