Artificial Intelligence in practice
A considerable amount has been written about Artificial Intelligence (AI) and the way that it is going to change work habits and practices.
To put AI into context, it is no more and no less than an aspect of the digital paradigm which itself is disrupting and transforming society. AI is something of a “dog-whistle” issue, giving rise to images of robots and disembodied voices telling humans what to do. This is an aspect of what the science fiction author Isaac Asimov referred to as “The Frankenstein Complex” – the atavistic fear of the created being.
Asimov did very well out of the Frankenstein complex. His series of “robot stories” were premised on some of the paradoxes that arose from his “Three Laws of Robotics”, which were designed to keep humans safe from the machines. All his robots were programmed with the Three Laws, which read thus:
- a robot may not injure a human being or, through inaction, allow a human being to come to harm;
- a robot must obey the orders given it by human beings, except where such orders would conflict with the First Law; and
- a robot must protect its own existence, as long as such protection does not conflict with the First or Second Laws.
Asimov’s paradoxes in his stories were in fact not science fiction, but exercises in statutory interpretation. Others have also done rather well from the Frankenstein complex (one only has to look at the success of the “Terminator” franchise). But rather than speculate about where AI is going, perhaps we should look at the AI that is with us and how that is going to impact on legal practice. Rather than worry about how we are going to regulate, inhibit or otherwise emasculate AI, we should be asking how lawyers can use AI systems to improve their practice and their services to their clients.
The deployment of AI into law has been with us for some time. That it should extend further is inevitable. But this does not mean decisions by Terminator J. Rather, use of AI systems will enable the smarter use of lawyer’s time and expertise. It will free lawyers up from repetitive tasks and enable far more targeted advice based on more accurate data analytics. AI is already being used in e-discovery using a number of different systems, of which predictive analysis is becoming well-known.
I want to briefly describe one subset of AI – legal expert systems – and place it within the context of the law office.
Legal expert systems
An expert system is a system that is “capable of functioning at the standard of human experts in a given field” (John Zeleznikow and Dan Hunter “Building Intelligent Legal Information Systems in the Law”, H.W.K. Kasperson et al. eds., Kluwer Computer Law Series 13 1994 (see pages 4 and 69)).
Expert systems enable many people to benefit from the expertise and judgement of experts anytime, anywhere, cost-effectively. They create leverage at Internet scale. However, one must use the term with some care, for it may encompass a number of different ways in which computer algorithms may be deployed.
Expert systems fall into four major areas:
- Analysis and advice – Systems basically set up to provide answers to questions based on an “IF THEN” model. A fact-specific analysis is required and it must be clear how the system reached its conclusion.
- Intake and assessment – These guide users through a system that collects data, evaluates facts and issues, and recommends actions to the user. Examples may be an incident reporting system, a compliance review system, a claim evaluation system or a due diligence guide.
- Intelligent workflow – These can be long running sessions. Rules are applied and messages are sent to multiple parties who contribute to the system and, when all the facts are gathered, reasoning is completed and the workflow is completed. Examples may be a process management system, a leave request manager or a compliance authorisation system.
- Document automation – These leverage the software to create complex documents of many types, including complex legal documents.
Generally, fact values may be obtained from the user or sourced externally from databases, files, web service or other applications. The expert system software applies fact values to reasoning and sets conclusion values. This process continues and, when all the required values are generated and sent, databases are updated and the session is complete.
Applications for legal expert systems
Information retrieval systems and expert systems comprise two types of AI applications used in law. Legal expert systems’ designs are categorised as either case-based or rule-based systems. Often, researchers build systems on a combination of the rule-based and case-based approaches. Rule-based systems are the most prevalent legal AI expert systems. These systems store legal knowledge as rules. The rule-based systems reason directly with these legal rules through formal logical deductive and inductive methods. Case-based systems operate by comparing the intersections of facts in a database of past cases, called exemplars, with the facts in the present situation. The case-based system attempts to draw analogies between the exemplars and the present case in order to retrieve the most on point cases.
Lawyers were originally identified as primary target users of legal expert systems. Potential motivations for this work included:
- speedier delivery of legal advice;
- reduced time spent in repetitive, labour intensive legal tasks;
- development of knowledge management techniques that were not dependent on staff;
- reduced overhead and labour costs and higher profitability for law firms; and
- reduced fees for clients.
Later, work on legal expert systems has identified potential benefits to non-lawyers as a means to increase access to legal knowledge. Legal expert systems can also support administrative processes, facilitating decision making processes, automating rule-based analyses and exchanging information directly with citizen-users. The benefits for clients are improved outcomes, reduced risks and reduced costs. For the experts in the domain new revenue streams are generated, strengthened and improved client relationships and replacement of billable hours with applications.
As noted, legal expert systems allow the repetitive aspect of legal work – gathering information and applying fixed criteria to ascertain rule application – to be automated. But the automation process is not “bespoke”. It has standardised elements to it and is therefore reusable. Because it is reusable, it can be considered a commodity.
This concept of the commoditisation of legal work is discussed by Richard Susskind in Tomorrow’s Lawyers, 2nd Edition (Oxford, 2017) (see chapter 3, page 25 et seq). The standardisation element means that repetitive tasks can be systemised, because in many respects the processes that are undertaken by legal expert systems are based on workflow systems. This means that the provision of this part of the service to the client comes at a significantly decreased cost.
Susskind gives the example of the insurance industry, where there is automation of high volume, low value tasks and activities. This way of automating workflow can enhance the efficiency of legal work to the point where, using a web-based service with the legal expert system available to the client on a 24-hour basis, the lawyer can literally make money while asleep.
The systemised approach can be applied to document drafting (an example of an automated document drafting system may be found at Automnio which is based on process flows – see https://autom.io/). Document automation requires users to answer a series of questions on a screen and after completion of the online form a first draft is made available. In none of this process has a lawyer been involved, unless the user inputting the necessary information is a lawyer.
This technology is not new. It has been around since the 1980s and it is a legal expert system in that it uses a rule-based decision tree. Susskind then takes the use of these commoditised systems a step further. If the drafting of certain types of contracts can be done online using a web-based interface, could this not be done within a client organisation? Why employ an expensive lawyer to draft “bespoke” standard form employment contracts, when the process could be undertaken within the human resources department of the organisation?
Does this mean that the lawyer gets cut out of the loop? Not necessarily. Susskind suggests that the lawyer “externalise” the service. “This occurs when lawyers pre-package and make their experience available to clients on an online basis,” he says.
This is a different way of obtaining the expertise possessed by lawyers and presents a number of different or alternative business models. The externalised service can be made available as a chargeable one, albeit at a rate less than for the bespoke product. There may be advantages to a “per use” charging model at a rate that encourages reuse of the system. It may well be that it could be made available at no cost – a model favoured by government and charitable organisations such as law clinics. Alternatively, it could be made available on a “commons” basis in the spirit of the open source movement.
The advantages for the client are clear. The cost of legal services comes down. The price of those services – freed from the tyranny of the hourly rate – becomes more certain. The time to complete the work reduces. The quality of the output increases, because sitting behind the system is the collective expertise of a number of professionals which outclasses that of the individual.
John Zeleznikow and Dan Hunter “Building Intelligent Legal Information Systems in the Law”, H.W.K. Kasperson et al. eds., Kluwer Computer Law Series 13 1994;
Richard Susskind, Tomorrow’s Lawyers, 2nd Edition (Oxford, 2017);
https://autom.io/ (last accessed 6 July 2017).