The use of artificial intelligence (AI) in legal services is increasingly employed in various legal contexts. As digital technology performs ever more complex functions, the trend is poised to accelerate and widen in scope. Since the direction of travel is clear, we need to consider the likely long-term consequences.

AI intelligence refers to the broader science of designing computers to act intelligently without necessarily being explicitly programmed to reach particular outcomes. Many tasks requiring familiarity with data and contextualised judgment in an infinite combination of circumstances can now be carried out by AI systems.

The push for increasingly more ambitious use of technology in legal services comes from both low-end and top-end users. Policy makers hope to harness computer technology in order to facilitate access to justice to those who cannot afford to pay for legal services, by now the great majority of the population. Top-end users who see the advantages of AI in their own businesses expect lawyers to likewise exploit its potential in order to increase efficiency.

The courts have been rather slow in taking advantage of digital technology. But now paper is being replaced by digital format, and soon much interaction with the court will be online. In tribunals, digital technology is employed for starting and processing applications in administrative appeals, in bail applications, for tax challenges, and in employment litigations. Users of tribunals can submit online arguments and evidence, participate in dispute resolution, follow the progress of their case, and engage in continuous online resolution to the point of final determination. Such uses of technology do not exploit AI, they merely use old fashion digital systems.

A bolder move in this direction is the Online Court, designed to facilitate access to justice to unrepresented litigants bringing modest claims. Computers operating symbolic AI, or expert systems, will guide lay litigants to frame a coherent legal claim or defence. Such systems are not AI based but operate with logical rules or decision trees, consisting of rules, or instructions, determining the result of given inputs. The reasoning is deductive (if X, then Y), followed through ‘yes’ or ‘no’ answers to pre-programmed questions designed to arrive at a predetermined final output.

AI systems are, however, employed at top-end litigation and for other demanding legal tasks. In commercial litigation machines are used to identify relevant documents. In complex transactions machines can now be used for document review. Automation of such tasks and others besides, which were traditionally performed by lawyers on hourly fees, may lead to a reduction in the number of practitioners.

So far, such developments have had limited impact. But as AI systems acquire greater legal skills and perform more complex tasks, a point may be reached when technological impact on legal services begins to change the legal landscape more profoundly. The legal profession may see a fall not only in size but also in the skill-sets lawyers are required to have. Furthermore, new entrants are likely to develop AI systems to compete with traditional providers of legal services, offering alternatives to human-centred services at lower cost.

It is not only the economic interests of the legal profession that are at stake. A leaner legal profession may no longer be able to produce sufficient numbers of lawyers possessing the kind of experience and skill-sets required for judicial roles. A legal profession that is smaller and that possesses more limited skill-sets may not be able to make its traditional contribution to the development of the law or to holding the executive and the powerful to account. The loss of its contribution to social wellbeing may be significant.

Digitisation seems likely to affect the nature of legal institutions, and in particular the court. If the technology employed in the Online Court develops to take advantage of the AI potential, much of the litigation process could be transacted without face-to-face interaction in the public space. This is not a farfetched prospect. Machines already perform judicial tasks in determining disputes between sellers and buyers on Amazon and eBay. As most people cannot afford to pay for professional legal representation in court proceedings, it is conceivable that before long similar forms of machine adjudication may replace judges in low value claims. The gulf between the legal services available to the rich and those affordable by persons of moderate means may grow wider still if the latter and larger group of litigants are no longer dealt with by human judges but are processed by machines.

If this came about, the adversarial system as we know it would disappear from large swathes of litigation because the decision-making process (data input and resolution output) would effectively take place in the interstices of a computer. Yet the adversarial system, involving opponents presenting their arguments and evidence to an adjudicator sitting in a court open to the public, underpins court legitimacy. Legitimacy in the sense of a public perception that the court has the right to demand compliance. It is this belief which inclines people to comply with the law, independently of reward or punishment.

The legitimacy of judicial institutions is founded to a large extent on their moral authority. Courts command moral authority because they are seen to respect the individual, engage with individual concerns, with persons’ moral outlook, beliefs, ambitions, emotions and their social consciousness. Since machines lack first-person subjectivity, machines know nothing of such things, algorithms consist of statistical patterns and little more. As a result, AI decision-making may lead to an ever-widening gulf between machine law and human conceptions of justice and morality, to the point where legal institutions cease to command loyalty and legitimacy.

     Despite such risks to the integrity of our legal institutions, the use of technology in legal services is set to expand due to its undoubted advantages. Court digitisation can achieve savings and greater efficiency. Computer adjudication could offer access to justice to the great part of the population which cannot afford to employ lawyers. Machines lack the personal touch, but as people get used to digital technology in all sorts of contexts, they may come to prefer, at least in the short term, machine adjudication because it strips out factors that algorithms consider irrelevant, such as emotions, and provides standardised outcomes faster and at lower cost.

Notwithstanding short-term benefits, a gap between AI and human wants and expectations may widen to a point where people may become alienated from legal institutions, leading to loss of legitimacy and damage to the rule of law. To be able to moderate the risks posed by AI, we must consider how best to benefit from what AI offers without undermining the foundations of our legal and social institutions. We may not be able wholly to control the course of events, but discussing the options now may give us a chance to avert some unwanted consequences.

Adrian Zuckerman, Emeritus Professor of Civil Procedure, University of Oxford.

(Suggested citation: A. Zuckerman, ‘Artificial Intelligence – Implications for the Legal Profession, Adversarial Process and the Rule of Law’ U.K. Const. L. Blog (11th March 2020) (available at https://ukconstitutionallaw.org/))