What If Judges Were Replaced by AI?

Whilst some may argue that Judge AI may be more predictable and accurate than human judging, there are numerous questions relating to transparency in decision making, procedural justice and issues relating to substantive justice when data and systems are used to determine a dispute particularly if they are developed outside the judicial system. At present, as digitization of court records and general dispute records is at an early stage in most jurisdictions, there are short term risks that Judge AI will result in perverse outcomes, as the only data to be considered is from cases finalized by judges which form only a small percentage of disputes that are finalized. From a longer term perspective, the risk is that faster and cheaper outcomes may be obtained at the expense of justice and the ongoing development of the law.

One significant issue emerges in societies where independent judges assist to develop the law. They do so in many democratic countries by dissenting, developing, innovating and considering law and justice within changing contexts. Whilst it is likely that AI can assist with this task (supportive judge AI), it is not likely, at least in the shorter term, that Judge AI (which involves the complete substitution of AI) will result in creative or innovative outcomes that can aid in the development of the law. Also, although justice is difficult to define, the growth in therapeutic jurisprudence, problem solving courts as well as the recognition that many human disputes cannot be defined by reference to legal issues alone means that AI does not and cannot currently address a myriad of human justice issues. This reality also means that a human being, who does much more than sort, categorize and produce a written outcome to support justice in both criminal and civil dispute arenas is unlikely to become redundant. The human face of justice, as characterized by a judge, includes empathy, compassion, reflection and the acknowledgement and support of the human struggle. 

Democracy and Judge AI

Whilst objections to Judge AI can be summarized as including: (i) a missing ‘human element’; (ii) transparency issues; (iii) issues associated with system transformation and change; (iv) the challenges of coding law; (v) issues surrounding Judge AI in innovative and novel situations; and (vi) issues associated with creating new law in precedent based systems, perhaps the greatest concerns where judges are replaced by AI relate to threats to democratic systems of government.[1]

In most democratic countries, the concept of separation of powers (sometimes called the Westminster system in the UK) is intended to ensure that no branch of government becomes too powerful. In such systems, the judiciary is one of the three arms of government comprised of the judiciary, the executive and the legislature. The judicial arm is intended to support a system of checks and balances, to enforce legislative requirements and to also support the legality and appropriateness of government decision making. Whilst some might question the extent to which the doctrine is realistic,[2] constitutional arrangements can require that each of the three arms of government have distinct roles and function independently of the other. 

It is in this context that additional concerns regarding Judge AI may arise. Most commentators suggest that central to the establishment and maintenance of the rule of law is the concept that the judiciary remains independent and the extent to which the judicial arm is ‘independent’ may be questionable, particularly if other arms of government are involved in the creation of Judge AI. Supportive Judge AI arrangements could also be perceived to be problematic in some countries, particularly if such supports are regarded as having an ‘improper’ influence on a judge. 

A central tenet of judicial independence that is also challenged by Judge AI relates to how an AI Judge could really be said to serve as a significant ‘check’ on the executive and legislative branches of government.[3] For example, Zalnieriute and Bell have identified the potential for automation to undermine the independence of the judiciary from interference or usurpation by the other branches of government.[4] In this regard, there are a number of assumptions that are made about automation and these include the view that the judicial role is only to ‘robotically apply the law to the facts.’[5] Judges do far more than this, and an analysis of the judicial role and function indicates their important role in settling disputes as well as determining them.

There are other issues linked to judicial independence that are not related to potential executive control or interference. For example, there are concerns regarding digital overreach in that many digital corporations that will develop Judge AI may not understand neither democratic notions, nor the rule of law, and may reshape judicial activities in unforeseen ways. The law as code movement is gathering momentum around the world and there are questions about in whose interests a mechanical application of the law might serve:

‘To be very clear: requiring that law be either as precise as code or be rewritten as fast as code is updated is simply anti-democratic, as this ignores the need for deliberation and compromise in democracy as well as the time required for due process under the rule of law’.[6]

Online Courts and Judge AI

To some extent, at least at present, Judge AI relies on there being written rather than oral evidence which is less likely to be the subject of exploration and interrogation. In countries where Judge AI is developing quickly, such as China, this may mean that it is unlikely that the position put forward by the State or a powerful party can be challenged, and inequity in terms of resources or coding variables may be more extensive. In jurisdictions where there are significant social issues that may have political ramifications, an independent judicial inquiry can be an important mechanism to support a functioning democracy.

In addition, online ‘courts’ that are created as separate tribunals or entities may raise issues of independence, particularly where they are developed and maintained without judicial supervision or monitoring. For example, both online courts and Judge AI raise concerns about the extent to which judicial independence is supported where court processes are not ‘open’. In this regard, judicial independence, impartiality and ‘open’ courts are often perceived to be essential to the maintenance of the rule of law.  Further, some commentators consider those online court developments which involve ODR constitute a potential encroachment on judicial independence by limiting the matters that may come before a judge or by otherwise impacting on the business of courts and therefore judges. 

“When justice slips out of sight … .the prospect of arbitrary, incompetent or unlawful conduct raises its head. Again, if we simply accept the argument that private online dispute resolution is the way in which the majority of disputes, and in some areas all disputes, may be resolved in future we accept this loss of accountability; we further accept the growth of a democratic deficit. And the same is the case if we divert public justice to an unobservable online forum. Our digital courts must be open courts”’.[7]

Some of these challenges can be overcome by a clear and consistent focus on using technology to support ‘open’ court hearings however the use of online hearings and hearings ‘on the papers’ raise issues about the extent to which procedural and participatory justice is supported.  

Conclusion

Is it realistic that AI will replace anything other than simplistic judicial decision making? My response is yes, although it may take some time and there are many hurdles to overcome and also questions about why we would want to do so.[8] Inevitably however, there is room for improvement. Inconsistent, slow, laborious justice processes can be unjust and are inappropriate in a modern world where AI offers options to enable justice improvements to thrive.

It seems clear that, in the coming years, technological developments will have an impact on the role of judges within democratic and non-democratic societies in terms of the retention of human judges, the function of judges, the definition of a court, the ‘openness’ of justice and also the humanity of the justice system. However, these impacts need not be negative. In many ways, technology can enable a better appreciation and understanding of the judicial role and judges can play a central role in the redesign of the justice system to ensure that technological reform efforts reflect current and potential future judicial functions. Ensuring that judges are engaged in reform is necessary, not only because of the issues that can emerge in democratic countries that relate to the separation of powers doctrine, but also because judges on the whole are more likely to be able to manage the balancing process that must be undertaken to ensure that the key tenets of the rule of law and inherent in the formulation of justice are maintained and supported. 

This means that the design of the justice system of the future must incorporate a nuanced understanding of the role of judges in society and enable judicial activism in an extended public form to assist to guide that design process. In addition, it is important that there is an understanding that judges as ‘guardians’ of the justice system may play a significant role in relation to the judicial supervision, monitoring, engagement and development of online courts, and in creating clear boundaries in relation to Judge AI with an understanding that Judge AI will have differing evolutionary stages.