Chief Justice Sundaresh Menon: "The role of technology in justice systems: a vision of accessible and collaborative justice"

Session 3 of Sejong International Judicial Conference (Justice in the Age of AI: When Law Meets Technology)

“THE ROLE OF TECHNOLOGY IN JUSTICE SYSTEMS: A VISION OF ACCESSIBLE AND COLLABORATIVE JUSTICE”

Chief Justice Sundaresh Menon^*

1. This paper discusses two global trends that have intensified over the past decade, both of which have profound implications for our justice systems. The first is the challenge of access to justice. The size and scale of the ever-widening justice gap have reached such alarming proportions that it should be acknowledged as a crisis that demands new and innovative solutions. The second trend relates to the rapid and dramatic technological advancements that we have witnessed, most notably in the emergence of generative artificial intelligence (“Gen AI”) and the remarkable growth of its capabilities and sophistication. While Gen AI is not perfect and has some serious limitations, it is here to stay and courts need to be thinking proactively about how, rather than whether, they can incorporate Gen AI into their workflows. 

2. Against that backdrop, the central thesis of this paper is that courts should harness the transformative potential of Gen AI to enhance access to justice. This should be seen as an essential step given the sheer size and scale of the global justice deficit. In particular, courts should take a more active role in leveraging on Gen AI to provide practical assistance to litigants who are not legally represented (referred to in this paper as self-represented persons or “SRPs”), in ways that were previously unimaginable, so that SRPs can navigate our courts more easily and effectively to obtain justice. If we do this in a measured and thoughtful manner, it will help us advance our shared mission of securing and maintaining public trust and confidence in our justice systems.

B. THE CHALLENGE OF ACCESS TO JUSTICE

3. The challenge of access to justice – the inability of people to obtain justice for their legal problems – can be understood in terms of the three principal dimensions of what may be termed the “justice gap”:¹

a. The physical gap represents the literal distance and other physical barriers that litigants may encounter when they try to access the courts.

b. The resource gap refers to the costs of navigating the justice system, particularly the escalating costs of legal representation. 

c. The literacy gap encompasses at least two elements. People who face legal problems often fail to recognise the legal dimension of their problem and the potential legal remedies; and even for those who appreciate that they face a legal problem, they may struggle to understand the relevant laws as well as the applicable procedures and processes. 

4. Unquestionably, the challenge of access to justice has worsened over the past decade. This can be seen in at least two notable ways. 

5. First, there has been a dramatic increase in the number of SRPs who appear before our courts. For example, in the United States, studies have shown that SRPs were a small minority in the 1970s, appearing in less than 20% of cases in the US state courts.² Today, the situation has changed to such an extent that the US state courts have been referred to as “lawyerless courts”, with more than 75% of cases involving at least one SRP.³ This may be attributed at least in part to the broader societal trend of lay persons becoming more confident of representing themselves because of the growing availability of and accessibility to legal knowledge and resources. But the rise of SRPs may also be indicative of the rising levels of inequality and the increasing cost of legal representation, which have contributed significantly to the access to justice deficit. 

6. Second, the growing number of SRPs belies a significant global predicament – the number of people who effectively live outside the protection of the law. This is a stark reality that threatens the sustainability of the rule of law and the well-being of our societies, because nothing can be more corrosive to public trust and confidence in our justice systems than the perception that justice is the privilege of an entitled few. 

7. The size and scale of this global crisis are starkly illustrated by the following statistics. In 2019, the World Justice Project estimated that 5.1 billion people – or two-thirds of the world’s population – had “unmet justice needs”;  in 2021, it observed that the access to justice crisis had been intensified by the COVID-19 pandemic;  and in 2023, it found that in seven out of ten countries, at least 62% of people who needed access to a dispute resolution mechanism could not obtain such access. 

8. It is important to appreciate that many of these “unmet justice needs” relate to relatively simple or routine legal problems. The Hague Institute for Innovation of Law has estimated that 60% of legal problems fall into just five categories – family, employment, land, crime and neighbour disputes. Further, many of these cases involve recurrent fact patterns.  There is therefore great scope for us to proactively search for new and innovative solutions that can help bridge the justice gap. While legal aid and pro bono efforts will continue to be important, the reality is that there are simply not enough lawyers and judges to meet the justice gap, leaving aside the fact that conventional solutions such as these are often expensive, reactive and resource-intensive.  

C. THE RISE OF GEN AI

9. It is in this context that courts ought to harness the transformative potential of Gen AI to enhance access to justice, especially to help those who would otherwise be completely shut out from our justice systems. Indeed, for all that Gen AI has shown to be capable of in the last three years, and the unprecedented speed and scale of its development and public adoption, could it be that we are still only at the “foothills” of what AI can potentially do for us? 

10. Indeed, some experts have said that we are on the cusp of a new era in AI that “promises to achieve an unprecedented level of ability”.⁹ The early signs appear to already be before us. For example, in July 2025, AI models developed by Google and OpenAI clinched historic gold medals at the International Mathematical Olympiad, surpassing the silver medal that was obtained in 2024. This has led to predictions that AI is less than a year away from cracking unsolved mathematical problems at the frontier of the field.¹⁰  

11. Notwithstanding these breakthroughs, the reality is that Gen AI – at least in its current state – still possesses significant limitations and risks. A recent study by the Harvard Business School and the Boston Consulting Group (or “BCG”) found that AI’s current capabilities “cover an expanding, but uneven, set of knowledge work”, which they described as a “jagged technological frontier”. Through randomised experiments conducted among 758 BCG consultants, the team found that AI can complement and even displace human work within the growing frontier. However, outside this frontier, its output is “inaccurate, less useful, and degrades human performance”.¹¹

12. The challenge therefore lies in understanding where precisely this frontier lies at any given point, given that it is rapidly expanding and constantly shifting. Further, the contours of the jagged frontier are likely to vary depending on the specific field of knowledge in question. For instance, a recent study found that while large language models (“LLMs”) were capable of basic legal analysis, they were also “limited by brief responses lacking detail, an inability to commit to answers, false confidence, and hallucinations”.¹² We know, however, that Gen AI is highly capable in other areas, such as the summarisation of large quantities of information and data, the recognition of patterns and the extraction of meaningful information based on defined parameters. 

13. How then should we think about the role of AI in our justice systems? I suggest that we can consider this in three distinct parts. 

a. First, AI should not be seen as a single agglomerated concept. We must disaggregate the compendious notion of AI into its distinct capabilities. AI can be used in a wide range of settings and its abilities are not uniform across these settings. We should therefore carefully examine the range of capabilities that AI possesses, and initially at least, be bolder about harnessing its strengths in the areas where it can – with relatively limited risk – have a positive impact on the experience of our court users as well as the productivity of our staff and their quality of work.

b. Second, given its limitations and risks, we should adopt an incremental and measured approach towards the adoption of AI. Given that the technology is constantly evolving, we should therefore engage in extensive testing and experimentation to better understand how it can be best applied safely. 

c. Third, the adoption of AI must be accompanied by broader efforts to enhance AI literacy, both in terms of enhancing our understanding of what AI can do and what it cannot do, as well as in practical knowledge such as in how to craft effective prompts with sufficient context to generate helpful outputs (also known as prompt engineering). This applies equally to judges, lawyers and SRPs, all of whom stand to benefit tremendously from learning how to work effectively with AI. Relatedly, we must establish a culture of accountability by developing the relevant frameworks to ensure that users recognise the need to verify the accuracy and relevance of AI-generated output. To that end, the Singapore Courts have, like many other judiciaries, published a “Guide on the Use of Generative Artificial Intelligence Tools by Court Users”, after consultation with our stakeholders.¹³

D. BRIDGING THE JUSTICE GAP THROUGH GEN AI 

14. Returning to the challenge of access to justice, Gen AI can help to bridge the justice gap by mitigating the resource and literacy impediments that have been outlined (see [3] above). Gen AI is perhaps less suited to bridging the physical gap, which is better addressed through other technological initiatives such as remote hearings and asynchronous hearings, both of which are used in various ways in the Singapore Courts. Thus, we have launched online portals that guide SRPs through the entire process of preparing and filing court papers for certain family cases, and to do this from any place of their choosing. We have also made selected frontline court services available at decentralised community touchpoints. 

15. The transformative potential of Gen AI to bridge the resource gap and the literacy gap stems from two key factors:

a. To begin with, unlike legal aid and pro bono services, once AI systems are developed and adopted, the marginal cost of using them is typically relatively low. They can therefore potentially be applied at scale to assist SRPs without appreciable increases in cost. This is crucial because the access to justice crisis is fundamentally a problem of scale. 

b. Furthermore, unlike other technological innovations such as electronic filing and remote hearings, Gen AI has the ability not only to replicate real-world processes in the digital space, but more importantly, to create novel solutions that were previously unimaginable. 

16. Take legal research for example, which can be challenging and time-consuming for SRPs without a legal background or access to subscription-based legal search engines. Drawing on the strengths of AI in distilling and summarising large bodies of information, and in collaboration with the Open Government Products unit in the Government Technology Agency, we have introduced Pair Search, a freely available AI-powered tool that allows any user to research Supreme Court judgments, legislation and parliamentary debates.¹⁴ Pair Search helps users identify the relevant cases that are responsive to their keyword search (for example, “division of matrimonial assets”), but more interestingly, it can also extract the key principles from the identified cases and provide case summaries and thematic or chronological syntheses of areas of the law. This is an example of how Gen AI could help to deliver granular and salient legal information to SRPs and help them better navigate our justice system. However, to enhance the prospect of this, we may also need to provide some basic training to help SRPs understand the method and limitations of such tools. 

17. Looking ahead, we think that Gen AI may be capable of providing practical assistance to SRPs in other ways. For example, Gen AI may help in the generation of court documents through suitable user prompts, and it may even, in time, be able to provide reasonable predictions of the likely outcomes of certain types of cases, to allow SRPs to make a more informed decision on whether to pursue their claim, as well as to facilitate settlement discussions. In particular, the latter capability appears to be suitable for fairly straightforward matters with recurrent fact patterns, or which involve no significant normative judgment. One possible example is in maintenance applications in family proceedings. In this regard, our Family Justice Courts are working on the development of an Online Dispute Resolution platform which will contain an algorithm that can suggest reasonable maintenance figures, so as to encourage the expedient resolution of maintenance applications.¹⁵ However, this does not presently entail any application of AI based on LLMs but is instead designed using decision trees that have been validated with actuaries. 

18. Lastly, we have experimented with Harvey.AI, the AI start-up behind the eponymous tool, to use Gen AI to assist users in the Small Claims Tribunals (or “SCTs”). By way of background, the SCTs’ jurisdiction is confined to certain types of lower-value civil claims, such as those relating to contracts for the sale of goods or provision of services, and contracts for residential leases not exceeding two years. The SCTs hear claims of up to SGD20,000, which may be increased to SGD30,000 if the parties consent. Significantly, all parties in the SCTs are self-represented because legal representation is not permitted there. This is to minimise costs and avoid the risk of a party being disadvantaged because of an inability to afford legal representation.¹⁶ Many of the parties are ordinary citizens seeking redress for the legal problems that they face in their daily transactions and interactions with one another. 

19. Although cases in the SCTs are often relatively straightforward both factually and legally, it is ironically in such cases that the justice gap may be at its widest, because of the barriers that could stand in the way of obtaining justice, in particular the disproportionate cost of legal services in this setting. We therefore considered that Gen AI would have the tremendous potential to significantly enhance access to justice in the SCTs. In line with the incremental and measured approach that has been advocated, our initial deployment of Gen AI in the SCTs has been to provide on-demand translation of court documents into Chinese, Malay and Tamil (the other official languages of Singapore apart from English). In September 2025, we announced that we would use Gen AI in the SCTs to summarise the parties’ documents, so that they would be able to better understand the other side’s position and hopefully reach an amicable resolution of the dispute. In time to come, we hope to provide Gen AI assistance to guide SRPs to draft and file their claims and defences, understand the evidence they need to produce, organise their materials, prepare their submissions, and possibly even provide suggestions for potential settlement or compromise.¹⁷

E. CONCLUSION

20. The challenge of access to justice is a global one that is in urgent need of innovative solutions. This paper suggests that Gen AI holds great promise in shaping some of these potential solutions. But in order to realise this promise, we must adopt and embrace the need for a collaborative approach:

a. First, we must work together with other stakeholders, such as governments, lawyers, technologists and SRPs themselves, to better understand the different facets of the challenge and their possible solutions. 

b. Second, we must also adopt a transnational outlook and examine the measures introduced by other judiciaries, including in the use of Gen AI. This is especially important given that Gen AI technologies are developing very quickly, and it would serve us well to draw upon the combined wealth of our experience and the learning we have each gained. 

21. Indeed, it is this spirit of collaboration that makes platforms such as the Sejong International Judicial Conference so valuable and timely, because they allow judges and technologists from around the world to gather and explore the role of technology and Gen AI in our justice systems. To that same end, my court regularly convenes the virtual Meeting of Chief Justices and Judges in Charge of Technology to discuss the interface between technology and our justice systems, and we would be happy to share more details about this forum to those who might be interested to find out more. 

Sundaresh Menon CJ

22 August 2025 (updated on 25 September 2025)