Justice Aidan Xu: Speech at the National Conference on Judicial Process Re-engineering and Digital Transformation

National Conference on Judicial Process Re-engineering and Digital Transformation

I. Introduction

1. The Honourable the Chief Justice of India, Honourable Judges of the Supreme Court and High Courts of India, distinguished delegates, ladies and gentlemen.

2. It is indeed a great honour to address this distinguished gathering. India and Singapore share deep institutional and legal bonds. Both our judiciaries are rooted in the common law tradition, and both are now grappling with one of the defining questions of modern judicial administration: how to harness technology in a way that genuinely strengthens justice, while preserving the values of fairness, legitimacy, due process and public trust.

3. If I may say so, it is indeed fitting for the Singapore judiciary to have been invited to speak today. This is the 200th anniversary of the establishment of the court system in Singapore. When the Second Charter of Justice was granted on 27 November 1826,¹ Singapore was administered from Calcutta, as part of the Bengal Presidency. It remained under the authority of British India until 1867.² The ties between our two countries run deeper than the common law and our occupation by a common colonial power. Singapore’s very name is Sanskrit — Singapura, “Lion City,” as I understand it from ‘simha’ and ‘pura’. Thus, when I address judges of India, I am conscious that I speak from a jurisdiction whose earliest recorded name and whose foundational legal institutions both trace connections to this subcontinent.

4. Before I turn to Singapore’s experience, it must be recognised that India’s judiciary has achieved a tremendous amount of progress in adoption of technology, and at a scale that very few judiciaries in the world could contemplate. I note that over 39.7 million virtual hearings have been conducted.⁴ The National Judicial Data Grid holds over 320 million orders, judgments and case records, publicly accessible and updated daily.  SUVAS has translated over 31,000 Supreme Court judgments into 18 Indian languages.⁵ Virtual Courts have processed more than 101.3 million traffic cases through fully automated 24/7 adjudication.⁶ Adalat AI transcription is deployed in thousands of courtrooms, with Kerala becoming the first jurisdiction in the world to mandate AI transcription for all its trial courts.⁷ And the e-Courts Mission Mode Project⁸ represents a programme of judicial transformation extraordinary in its ambition. These are not aspirations. They are achievements.

5. India’s challenges are of an entirely different scale from Singapore. India is a very large federation, with great linguistic diversity, varying levels of connectivity, hundreds of millions of legacy paper records still awaiting digitisation, and a large volume. Singapore is a single small city-state, with one unified court system, one national language of record, and a caseload that is much, much smaller. Much of our experience must be understood in that light.

6. This speech is therefore a sharing of our experience, in our context. It may not be directly applicable to what you have embarked on here, but we hope that some of it, including our framework, our experimentation, and our own ongoing challenges, may be of interest. If anything in Singapore’s experience is of use to India’s judges, I hope it may be found in the detail of how particular problems were framed and approached, rather than in any claim to a superior method.

7. I have been asked to speak about our efforts in harnessing technology in the judiciary, as well as the steps we have taken to ensure that the wider organisational environment facilitates the use of technology, through rules, processes, structures and culture. I will speak under five broad themes: first, the institutional architecture of disciplined innovation; second, enhancing efficiency through digitalisation; third, expanding access to justice through technology; fourth, tackling culture change; and fifth, the emerging challenges posed by generative AI and data governance.

8. Let me first emphasise the underlying principle. It is never technology for technology’s sake. Technology is just a tool, so the first question must always be what the problem is. Sometimes the answer is technology; sometimes it might be AI. But sometimes the answer may be less glamorous: better signage or a simpler form. Where rules are needed, they should be facilitative, helping to create space for better ways of working, not adding layers of red tape. Similarly, technology should aim to help, not create more hurdles to trip up the users.

9. As our Chief Justice has observed, the Singapore Judiciary sees technological innovation not merely as another operational tool, but potentially as a force multiplier and a strategic driver of transformation. But a force multiplier must be directed carefully, or it will simply multiply confusion, risk or inequity. That is why, in Singapore, the story of technology in the courts is really a story about design of the institutional framework, rules and culture.

II. A Framework for Disciplined Innovation

10. The first theme is the framework that enables disciplined innovation. We try to start off on the right footing, making sure that we explore options fully. We aim for iterative development and pilots over single leaps, and establishing structures so that reform does not depend on the enthusiasm of any one individual. Beyond the judiciary itself, I will explain the enabling national structure that Singapore provides — the government infrastructure, executive branch support, and the broader legal sector. I will also speak about a relatively recent but important development: our Access to Justice Programme Office, which examines whether the courts are truly serving all who need them. Finally, I will address stakeholder engagement — including some candid observations about where our own systems have fallen short — and the facilitative legal framework that underpins all of this.

11. The most sophisticated technology is of little use if it is deployed in an institutional vacuum. If there is no clear legal basis for its use, uncertainty follows. If there is no mechanism for reviewing what has been built, early design mistakes become enduring constraints. And if those who must use the system are not involved in shaping it, adoption will be partial at best. For these reasons, the Singapore Judiciary has treated the institutional framework for reform as every bit as important as the technology itself.

A. How Process Re-Engineering Begins

12. The discipline we aspire to apply in initiating a process re-engineering exercise is rooted in a very simple principle: start with the problem.

13. That may sound obvious, but it is often the first principle to be forgotten in periods of technological hype and excitement. Shiny new technologies can be very enticing. One must try to resist and be disciplined. Once a problem is identified with sufficient clarity, a structured assessment follows: Is the heart of the problem legal, procedural, technological, or cultural? Would we solve it faster by amending a rule, redesigning a form, changing a workflow, or building a new system? What risks would each option create? How will success be measured?

14. Our Chief Justice has drawn on Professor Richard Susskind’s distinction between automation, innovation and elimination,⁹ and that distinction is equally useful as a framework for process re-engineering. If a problem arises because a necessary task is done manually, the solution may be automation. If a problem arises because a task is inherently complex but important, innovation may be needed. And if a problem arises because a task no longer serves any real purpose, the best solution may be to eliminate it altogether. This framework represents what we aspire to — a principled, problem-first methodology — even as we acknowledge that real institutional life does not always follow such an orderly sequence.

15. In practice, most reforms involve a combination of these approaches. For example, the move toward asynchronous hearings in certain criminal pre-trial processes required not only technological capability, but also a rethinking of rules and workflows.  The traditional view, especially in common law jurisdictions, is that hearings require presence of all parties before the court, with oral arguments and evidence.  This, in the past, helped ensure participation and the right to be heard.  However, not all hearings these days would necessarily require presence at the same time.  Administrative matters and other proceedings, such as neutral evaluation, could readily be carried out with the simultaneous presence of all.  To that end, we introduced asynchronous hearings, first by way of email, and then eventually through integration in our case management systems. 

16. Perhaps one of our most significant exercises in re-conception and a refreshed approach has been the introduction of therapeutic justice in our family courts. This movement away from the adversarial approach to one grounded in resolving underlying issues between parties in family justice disputes, facilitated by a judge-led non-adversarial process, has marked a sea-change in our justice system.  The benefit has been reduced conflict, and hopefully better welfare and less acrimony.  This is an illustration of re-engineering of the process that has not been driven by technology, but rather by the human touch, though of course technology systems are there to facilitate the work.

17. We  are also exploring the use of initial automatic determination of some kinds of disputes, whether by AI or by rules driven by data, to allow parties to be offered an initial assessment of their claim. If they are satisfied with that automatic initial assessment that will be the end of the matter, saving them time and effort. Should they be dissatisfied, they may have the matter put before a judge for full determination.  It is hoped that this mechanism, which again is a significant departure from traditional methods, would help reduce cost and pain for those pursuing justice.  

B. Iterative Development, Pilots and Review

18. Once a solution path is identified, we generally prefer an iterative approach rather than a single leap.

19. New procedures and tools are often piloted in specific courts or case types before wider roll-out. A pilot might be confined to one registry, one class of cases, or one part of the process. The goal is to learn with lower stakes, to gather feedback, and to adjust the design before it becomes entrenched.

20. For instance, asynchronous hearings in criminal pre-trial conferences were first applied to selected categories suitable for decisions on the basis of written exchanges without oral argument. Only after these trials showed tangible improvement, without compromising fairness, did they become a standard feature.

21. Review occurs on several levels: at the operational level, to see whether the system functions as intended; at the judicial level, to see whether it supports rather than distorts adjudication; at the institutional level, to see whether it aligns with strategic goals; and at the user level, to see whether lawyers and litigants experience the process as an improvement. Review, and rethinking, or revisiting, is not an admission of failure. It is just part of the process.

C. Permanent Institutional Mechanisms

22. We aim as far as possible at the use of permanent institutional mechanisms rather than temporary ad hoc arrangements.

23. The Office of the Transformation and Innovation reflects the importance of judicial leadership in this area. Technology reform cannot be outsourced entirely to technical teams or external administrators. It must be connected to our core judicial, organisational priorities. Internalising these efforts ensures continuity and execution. Reform requires project management, systems architecture, integration, vendor management, testing, training and support. Culture too needs to be changed and fostered.

D. The Enabling National Ecosystem

24. In its technology efforts, the Judiciary is able to build on national resources. The Smart Nation and Digital Government Office guides digitalisation across government,¹⁰ while the Government Technology Agency, or GovTech, manages government technology infrastructure, including that of the judiciary. These shared infrastructure and technology support ensures capabilities that the judiciary could not maintain on its own without much expense and greater resources.

25. In addition, we are fortunate that there has been strong support from the executive branch, particularly the Ministry of Finance and the Prime Minister’s Office, as well as of course the Ministry of Law. This is the product of national commitment to harnessing technology for the benefit of the country as a whole. The Singapore Academy of Law also plays a significant role, bringing together judges, lawyers, in-house counsel, government legal officers and academics, and serving the needs of the entire legal sector in various areas including technological innovation. I serve as the Chairperson of the Promotion of Legal Technology Innovation Committee of the Academy, which spearheads this effort.

26. The national, wider institutions also provide critical infrastructure in the form of the intranet backbone, identity and e-payments systems, and cybersecurity structures. This helps to ensure that whatever we do, we are able to rely on the support of the government as a whole.

27. Beyond this, the presence of foreign law firms, who often bring into Singapore a readiness to make use of technology for increased productivity, has also driven acceptance of technology across the board, and ensured that we are aware of and keep up with international expectations and standards.

E. Judicial Independence and Technology Governance

28. I should address a question that may be on some minds: if the judiciary relies on executive-branch agencies for its technology infrastructure, does that create risks to judicial independence? This is not a theoretical concern. A court system whose digital backbone is controlled by another branch of government must be alive to the implications.

29. In Singapore, the relationship works because of clear governance boundaries. The judiciary retains control over what is built and why — the functional requirements, the priorities, and the data. GovTech provides the technical capability and infrastructure. The judiciary does not cede decision-making over its own processes. There is also a practical reality: a small jurisdiction cannot afford to build and maintain an entirely independent technology stack. The question is not whether to rely on shared infrastructure, but how to structure that reliance so that judicial autonomy in matters of substance is preserved.

F. Access to Justice Office

30. One especially significant institutional development has been the establishment of the Access to Justice, or A2J, Programme Office, specifically to examine and assess the courts’ facilitation of access to justice, and to improve what we do in that respect.

31. Ensuring access to justice is part of the core mission of the Singapore judiciary. Barriers to justice are often practical and procedural as much as doctrinal. We need to ask ourselves not just if we are efficient and prompt; we also need to ask whether we are truly available to all who need our services. Again, sometimes the answer to an access to justice problem may be technology; sometimes, or often, it is not.

G. Stakeholders

32. Technology in the courts affects judges, registrars, lawyers, self-represented persons and the wider public. Stakeholder engagement must therefore occur at every stage: diagnosis, design, piloting, implementation and review. Where file and case management is concerned, often seemingly small design choices can have outsized practical effects. I will go into this a bit more in our journey in the development of our e-filing system, but one issue that I have to grapple with is that our system as originally designed was meant more for the backroom, with relatively little catering for the needs of the judge, district judge or assistant registrar, trying to hear a case. The system currently is less friendly to those trying to hear a case with only soft copies. Twenty-five or so years after our eLitigation system was rolled out widely, we still use a voluminous amount of paper. Sometimes the lawyer may be fully electronic, but the judge still needs to flip volumes of paper. We are trying to remedy that.

H. A Facilitative Legal Framework

33. The institutional mechanisms I have described are supported by a facilitative legal framework. In Singapore, we are able to move relatively faster in pursuing rules changes and statutory enactments where necessary.

34. We generally intend for our rules to be facilitative rather than prescriptive. They create legal space for technology-enabled processes to be adopted where appropriate, while preserving judicial discretion and safeguards. Legislation and rules provide legitimacy and certainty; Practice Directions and Circulars provide the agility to respond to new developments without formal legislative amendment.

35. This layered approach matters because technology moves faster than legislation. If every innovation required a statutory amendment before it could be trialled, the pace of reform would slow to a crawl. The combination of broad enabling provisions in primary legislation, detailed procedural rules that can be updated by the Rules Committee, and Practice Directions that can respond to operational needs, gives the system the flexibility it requires.

III. Efficiency Through Digitalisation

36. I turn then to the second theme: efficiency through digitalisation. We have borne in mind that process re-engineering is not the same as digitisation. Digitisation asks how to make an existing process electronic. Process re-engineering asks what the purpose of the process is, and whether, if designed today in light of present capabilities, it would be designed in the same way at all. A court system can digitise inefficiency just as easily as it can digitise efficiency. What is needed is not mere translation from paper to screen, but a disciplined reconsideration of the workflow itself.

37. I will begin with eLitigation, our principal civil filing platform, and then turn to electronic service and how it connects to Singapore’s national digital identity infrastructure. I will address how we handle original documents and electronic records, before discussing MACO, which is a targeted tool for motor accident claims. I will then speak about what is probably our most significant current challenge: the difficulties caused by the proliferation of different case management systems and our effort to pull things together through the future case management system, or fCMS. And I will close this section with data — not just as a by-product of digitalisation, but as a strategic objective in its own right.

38. Efficiency in judicial administration is sometimes misunderstood as a purely managerial concern. It is not. Delay, duplication, unnecessary complexity and fragmented processes all impose real costs on litigants, lawyers, judges and the public. These affect and weaken justice: a justice system that is unreasonably slow or which throws up obstacles defeats the very reason for its existence. Inefficient justice is simply injustice.

A. eLitigation

39. The backbone of electronic filing in Singapore civil proceedings is eLitigation, which in 2013 replaced the earlier Electronic Filing System. We started the Electronic Filing System on a wide basis in 2000, after pilots starting in 1997. It was one of the first in the world.¹¹

40. We embarked on EFS and then eLitigation, because it enabled the migration of filing, service and retrieval processes into a centralised digital environment, creating a standardised filing environment, improving tracking, allowing search of court records and eliminating loss of physical documents. It is on this foundation of reliable and efficient filing that we built the rest of our technological work.

41. We recognise that India’s e-filing system which is now operational across High Courts and District Courts in bilingual format, represents a comparable commitment across a vastly more complex jurisdictional landscape. I would note in particular that the Supreme Court of India’s collaboration with IIT Madras on AI-powered defect detection in e-filing¹² — to identify errors and deficiencies before filings are accepted — is an innovation that Singapore has not yet pursued. It is a thoughtful application of AI to a very practical quality-control problem.

B. Electronic Service and Digital Identity Infrastructure

42. Electronic service is a particularly important example of how technology can improve both certainty and efficiency. Where service takes place through eLitigation for registered users, the system generates a certificate of service, creating a reliable and reviewable record. That reduces disputes over whether service occurred and when it took effect.

43. At a deeper level, electronic service illustrates how administrative certainty contributes to procedural fairness. Parties should know when they have been served. Courts should know whether timelines have been triggered. At the same time, the courts retain flexibility in relation to substituted service through electronic means where appropriate. This reflects another recurring theme of judicial digitalisation: technology expands available tools, but judicial oversight remains essential.

44. A particularly significant development is the integration of court processes with the national digital identity system, Singpass. Singpass is the digital identity system covering all Singapore citizens and residents bridging access to over 800 government and private sector services.¹³ We have rolled out a Document Signing Service linking to Singpass authentication, and have made use of the Singpass Inbox: letters and notices may now be sent directly to a user’s inbox on the Singpass application. Building on that capability, it has become possible to perform substituted service of court documents through the Singpass Inbox, providing a further avenue where personal service or service on appointed lawyers is not possible.

45. India’s Aadhaar system, which is a national digital identity structure of great scale, offers, I would respectfully suggest, similar potential for integration with court processes. A universal, trusted identity and communications platform, can benefit the judicial system by avoiding the need for duplicative infrastructure. The benefits of national identity systems can and should extend to the administration of justice.

C. MACO

46. Let me turn next to the Motor Accident Claims Online Simulator, or MACO, as an example of a more targeted digital innovation. It was launched in October 2020 in collaboration with the Singapore Academy of Law,¹⁴ to help parties in road accident cases understand likely liabilities and approximate compensation outcomes.

47. Within its first one and a half years, MACO recorded more than 13,000 simulations. Against a population of about 5.92 million in 2023, that figure is revealing. It suggests within our context a strong demand for tools that help parties and lawyers understand likely litigation outcomes at an early stage. MACO promotes informed decision-making, supports early resolution, and demonstrates that not all useful innovation in the justice system consists in building large platforms. Purpose-specific tools can also have significant impact.

48. MACO is also a useful reminder that partnerships matter. The collaboration with the Singapore Academy of Law shows how judicial systems can work with external institutions while retaining a clear public purpose.

49. We note that India’s Virtual Courts for traffic challans go considerably further in one important respect: they involve actual adjudication, not merely outcome simulation. Over 60 million traffic cases¹⁵ have been processed across 21 states through a fully automated system operating 24 hours a day without requiring the physical presence of either a judge or a litigant. That is a more ambitious model of automated resolution, and one that demonstrates what is possible when the right category of case is matched with the right institutional architecture.

D. The Current Fragmentation Challenge and the Future Case Management System

50. If one asks where the next great efficiency challenge lies, the answer is in case management architecture.

51. At present, the Singapore Courts operate four separate case management systems. This arrangement can be visualised as keeping four smaller aircraft in the air simultaneously rather than one well-designed aircraft. Each system has its own rules, data architecture, interface design and reporting capability. Integration between them is limited. The result is duplication, inconsistency and operational burden.

52. This fragmentation is the specific institutional problem that the future case management system, or fCMS, is designed to solve. The fCMS is a single unified platform that will consolidate all four existing systems into one shared architecture, built on three principal capabilities.

53. First, a core CMS providing the common features that every court requires — case records, hearing scheduling, event logs, communications and document management — so that each court is no longer maintaining a separate system for functions that are fundamentally the same across jurisdictions.

54. Second, a flexible forms engine enabling courts to design and deploy electronic forms without custom development. Court forms are central to user interaction. A forms engine that allows rapid creation, version control and structured data capture — including logic-driven branching — removes a significant bottleneck.

55. Third, an advanced analytics engine providing real-time dashboards, resource allocation tools and predictive modelling — capabilities that are only possible when all case data resides in a single, consistently structured repository.

56. The fCMS is being designed from the outset as a modern, integrated platform rather than a legacy system with additional modules. It will incorporate universal design principles and accessibility standards. It does not aim to replicate existing systems in a new skin, but to reimagine how case management should work, discarding legacy features that no longer make sense. Trying to reproduce everything as it was will just create further problems.

57. A critical dimension of fCMS development is stakeholder engagement and leadership support. The engagement of judicial officers and legal practitioners is essential, and there has been emphasis on the need for leadership to provide clear strategic direction, remove institutional obstacles and maintain momentum.

58. What we hope to do as we implement this is to embed AI tools, to make the system ready for paperless hearings, and to address other needs in the courtroom and elsewhere.

E. Harnessing Data as a Strategic Objective

59. Data is an important resource. Properly obtained and analysed, it is crucial in informed decision making. This requires collecting well-structured data, and subjecting it to proper and thorough analysis. In Singapore, the Judiciary has been exploring the use of data for each of these purposes, and fCMS is intended to place these capabilities on a far stronger foundation.

60. The data aspiration is not merely about dashboards. It is about building a data and evidence based administrative culture. Data technology alone will not be sufficient; culture and education are essential.

61. India’s National Judicial Data Grid, which has over 276 million records, already represents one of the most impressive foundations¹⁶ for data-informed judicial administration anywhere in the world.

IV. Access to Justice Through Technology

62. I turn now to the third theme: access to justice through technology. This is, in some ways, the theme that matters most — because a justice system that works efficiently but remains out of reach for ordinary people has failed in its most basic purpose.

63. It is not enough to focus on efficiency and harnessing of data. The justice system aims to serve the people. A justice system that is not used, that is not accessible, is but a shadow of itself. Access requires that the justice system be open, and readily navigable by those who need to use it.

64. In what follows, I will discuss access to justice as a design principle. This means it should be built into systems from the start, not bolted on afterwards. I will then describe the digital services we have developed for court users, before turning to the Community Justice and Tribunals System as an example of access-centric design. I will also cover our partnership with Harvey for the Small Claims Tribunals, the role of remote hearings as access enablers, and the use of video link for evidence. I will close this section with a reflection on why access must remain central to everything we do.

A. Access to Justice as a Design Principle

65. Our Access to Justice office, the A2J office, exists to support the courts in making justice genuinely reachable. It reflects the understanding that access to justice is not simply about whether a courtroom door is open. It is about guiding users through the uncertainty of the process, and helping them move through it without being defeated by complexity. The A2J office examines how things can be improved for our users. This can be through simpler forms, clearer guidance, better signage, or technology where it genuinely helps. Not every access problem has a technological solution, but every access problem deserves institutional attention.

B. The Digital Divide: Equity Risks in Digitalisation

66. Before I turn to specific digital services, I must address a concern that any honest discussion of court technology must confront: the risk that digitalisation creates a two-tier justice system. Those who are digitally literate, well-resourced, and comfortable with technology may find the system faster and more convenient. Those who are not — the elderly, the less educated, those in areas with poor connectivity, those who simply find technology intimidating — may find the system harder to navigate than it was before.

67. This is not a hypothetical risk. In Singapore, even with high internet penetration and a well-educated population, we have encountered users who struggle with our digital platforms. We have had to provide in-person service counters as a complement, not an alternative, to digital access. We have had to ensure that the move to electronic filing does not mean that someone who cannot use a computer cannot file a claim.

68. The principle, as I see it, is that digitalisation must expand access for the majority without closing access for the minority. This requires deliberate, sustained effort — not merely a helpline number on a website. It requires physical help desks, guided filing assistance, and a genuine institutional willingness to meet people where they are, including those who are not online. India’s e-Sewa Kendras, with over 1,394 operational help desks at court complexes across the country,¹⁷ providing human-assisted physical access points for litigants who cannot navigate digital platforms independently, represent an exemplary response to this challenge.

C. Digital Services for Court Users

69. The span of digital services for users is very broad. Not every access problem arises at the hearing stage. Many issues arise earlier. Someone with legal difficulties may not know what information is needed, what forms must be completed, or what type of claim is appropriate. Digital tools can reduce that uncertainty and lower the intimidation barrier.

70. Digital tools must also be designed with awareness that many court users are not comfortable with technology. While we must not allow the benefit for the majority to be erased by over concern for edge cases, common concerns about dealing with technology, such as an unfamiliar web interface can be as intimidating as the legal process itself. Wherever possible, the design must account for varying levels of digital literacy. Plain language, clear navigation, the ability to save, are among the measures that might help.

71. In this regard, we have often run into issues about the proper scope of assistance by the courts. There is an understandable reluctance or fear of trespassing into giving advice, i.e. telling a court user what is better for them to do, rather than just giving information for them to then formulate their own course of action. The line between the two may be thin. But it is something that calls for considered engagement which is beyond the scope of this paper. As an institution, we do think it is appropriate in our context to give some informational assistance as part of our functions.

D. Community Justice and Tribunals System

72. Let me take our Community Justice and Tribunals System, or CJTS, as an example of our attempt to be access-centric. It provides online filing, case management and forms of online dispute resolution such as e-negotiation and e-mediation, with self-represented persons in mind.

73. A platform designed for self-represented persons cannot simply be a lawyer-facing system with fewer buttons. It has to be built around very different assumptions: that the user may be encountering legal process for the first time; that legal terminology may be unfamiliar; that confidence may be low; and that the system must support rather than punish ordinary misunderstanding. The justice system must meet users where they are.

74. We aim to ensure that our future case management system, into which the CJTS system will eventually be folded, is simple to use and navigate, so that self-represented persons can engage with it with ease.

E. Harvey and the Small Claims Tribunals

75. Another important initiative is the Judiciary’s collaboration with Harvey, which involves providing AI tools for document translation in the Small Claims Tribunals and summarisation of case documents. We embarked on this as the users of the Small Claims Tribunals, from which lawyers are excluded, often struggle to organise facts, identify relevant documents or understand what the tribunal will consider material. Often a mass of materials, namely documents, social media messages and photographs are filed into the system. These may not be readily understood by the other side or by the judge. We thought that an AI tool could assist in digesting these materials, allowing the users to understand the case against them, and thus be prepared to respond better.

76. At the same time, safeguards are essential. Users must not be encouraged to think that AI output is itself judicial guidance. Tribunal Magistrates must retain control. The risk of hallucination and overconfidence must be managed. The Small Claims Tribunals are a sensible place to begin — a high-volume environment with many self-represented users and recurring procedural needs, where deployment can remain controlled and learnings gathered before broader expansion.

77. On translation and transcription, India has moved much faster than Singapore. SUVAS has translated over 31,000 Supreme Court judgments into 18 Indian languages.¹⁸ Adalat AI has been mandated for transcription in all trial courts in Kerala,¹⁹ and is deployed across more than 3,000 courtrooms in multiple states. Singapore remains at an earlier stage on transcription. We are still experimenting and testing reliability in our courtrooms. India’s experience will be highly relevant as we develop our own approach.

F. Remote Hearings Facilitating Access

78. The Courts (Civil and Criminal Justice) Reform Act 2021 created permanent statutory authority for remote hearings, documents-only hearings and asynchronous hearings.²⁰ When remote hearings were rolled out in the midst and aftermath of the Covid pandemic, we worked to provide video-conferencing facilities within court premises so that self-represented persons who lacked equipment or internet access could still participate fully.

79. While we have moved away from the levels of remote hearings since the pandemic, they are still used in many settings. Remote hearings reduce time away from work, and improve scheduling flexibility.

80. The scale of India’s remote hearing programme must be acknowledged. We understand that over 33.8 million virtual hearings have been conducted.²¹ India’s Nyaya Shruti app, launched in 2024,²² enables virtual appearances and testimonies by accused persons, witnesses, police officials and prisoners directly from jails and hospitals.

G. Evidence by Video Link, Transcription, and Online Affidavits

81. The use of video link for witness testimony reflects the intersection of technology, procedure and judicial craftsmanship. Singapore has long had legal provisions permitting evidence by live video link: s 62A Evidence Act 1893.  We have also permitted electronic signatures for a number of years, through the Electronic Transactions Act 2010.  The legal permissibility of remote evidence, however, is only part of the story. One must also ensure reliable platforms, secure record-keeping, clear protocols and fairness in the conduct of testimony.

82. I understand that there is interest in how we manage online hearings. I will focus on civil matters.  As I have mentioned online hearings are permitted, and were widely used during the pandemic, both for trials as well as application hearings.   Since the end of the pandemic, we have continued use of online hearings for applications²³, i.e. those matters determined solely on affidavit evidence with a full trial with cross examination. For trial matters, we have generally pivoted back to physical courtrooms as lawyers and judges find it easier to manage the interaction physically. There has also been a concern that younger lawyers had lost much inculcation of etiquette and behaviour because of lack of exposure to physical hearings.  We think we have now largely struck the right balance in our context. 

83. When we do have trials online, various arrangements will be made through pre-trial conference directions, covering the arrangements for testimony by witnesses from remote locations and the use of trial bundles.  Some of these are adapted from arbitration proceedings.  Witnesses will generally testify either from a lawyer’s office, or some other agreed premises.  The opposing party will be given the option of having a solicitor or other representative present to help ensure propriety. This option is not always taken up.  Agreement will also be reached on the trial bundles, containing the affidavits and documents, to be provided to the witnesses: sometimes, this will be in hardcopy, sometimes in soft versions, often through the assistance of litigation management companies retained for the hearing. The same will apply as to the provision of simultaneous transcription.

84. Evidence for trials, whether online or physical, will in civil proceedings, generally have been filed online through our eLitigation system.  Trial bundles, such as bundles of documents, bundles of affidavits, and bundles of authorities, are marked as usual as in physical hearings, with the marking pronouncements read into the record.  At present, there may be considerable duplication among these bundles. To reduce that, we are working on a bundle application as part of improvements to eLitigation, called Smart(er) Bundles, that will allow the judge and parties to construct virtual bundles from individual pages in a pool of filed documents. 

85. We generally discourage late filing or introduction of documents. Where this is necessary, in online hearings, arrangements will be made for the soft copies to be distributed to the parties and witnesses usually through email or some file-sharing device, and will be marked in the usual way.  The document will also need to be filed on eLitigation for the record.  

86. We are aware that some jurisdictions have introduced various more elaborate mechanisms, including the use of blockchains, but we do not think that this is necessary at present.

87. What we are examining are better ways of capturing digital evidence. Video or audio recording will generally need to be transcribed, with the transcriptions filed together with a copy of the recording, usually in a CD-ROM. We are looking at alternatives, though this poses some challenge as we can foresee that digital storage may become burdensome if uncontrolled. 

88. Additionally, in some types of cases, we may need to allow software evidence to be captured more fully. Currently again, these are mostly adduced through screen shots captured in pdf, together with a CD-ROM of the software. This may not be sufficient, so we are looking at various alternatives. We have used video capture software to record software demonstrations on screen, but in some contexts, we do need to enter digital evidence into the record without altering it. 

89. The remote swearing of affidavits raises a related point. While it offers convenience and speed, safeguards must be maintained. Identity must be verified, deponent and commissioner must see and hear each other in real time, documents must be properly identified, and records must be kept.

90. The legislative framework spans a number of statutes and rules:  s 7 of the Oaths and Declarations Act 2000, s 18 of the Electronic Transactions Act 2010, O 15 Rule 22 of the Rules of Court 2021, and practice directions, including para 81 of the Supreme Court Practice Directions 2021.  In broad terms, remote attestation is available where both the deponent and the commissioner for oaths are in Singapore. Cross-border attestation is not permitted.  Among the requirements are that the commissioner for oaths must be able to maintain visual contact and communicate with the deponent and any interpreter; confirm the identity of these persons;  verify by visual inspection, read, interpret and explain the document; and confirm that the document with the commissioner is the same document that the deponent affirms and signs. 

H. Open Justice in a Digital Age

91. Digitalisation also raises important questions about the open justice principle. Courts have traditionally been open to the public, and this openness serves fundamental constitutional values: accountability, public confidence, and the disciplining effect of scrutiny. The question is how digitalisation changes the practical operation of that principle.

92. On one hand, digital platforms can make justice more open than physical courtrooms ever were. Electronic access to judgments, published court records, and online filing systems all increase transparency. India’s National Judicial Data Grid, with its 276 million publicly accessible records,²⁴ is arguably the most ambitious implementation of open justice through technology in the world.

93. On the other hand, there are countervailing concerns. The ease of digital access means that information which was always technically public but practically obscure — an old bankruptcy filing, a dismissed charge, a family dispute — can now be surfaced instantly by anyone with a search engine. The practical obscurity that once protected individuals is eroding, and the law has not fully come to terms with the implications.

94. The Supreme Court of India’s decision to livestream proceedings²⁵ is a bold step that Singapore has not taken. Different jurisdictions will reach different conclusions on these questions, and the appropriate answer will depend on context, legal culture, and institutional readiness. 

I. Why Access Must Remain Central

95. Technology will only improve justice if access remains a central design principle rather than a rhetorical afterthought. Institutional users, lawyers, judges and administrators often adapt. The harder question is whether the ordinary person, encountering the justice system perhaps once in a lifetime and under stress, can make sense of what is required. Access to justice cannot be treated as an optional supplement to digital transformation. It must be part of the transformation’s logic from the beginning.

V. Culture Change: The Human Dimension of Reform

96. I turn now to the importance of addressing organisational culture. One can build excellent systems, amend sound rules, and establish sensible governance structures. But if the people within the institution do not change how they think, work and relate to the technology, the reform will underperform. The management of cultural change will require shifts in perceptions, beliefs and practices, and that uncertainty, insecurity and sometimes fear will be part of the process. In what follows, I will address why culture matters and the challenge of reimagining habits, the practical requirements of training and support, the real problem of change fatigue, and the need for an outward-looking institutional perspective.

A. Why Culture Matters and the Challenge of Reimagining Habits

97. Technology reform often appears to be about systems. But systems are always embedded in human behaviour. It is only human, when new things are introduced, for people to still stick to tried and trusted ways of working and thinking. The institution must move beyond merely digitising old habits, transferring from paper in the real world to paper on screen. For true gains to be achieved, it must be prepared to reimagine what it does, and discard unnecessary aspects, while keeping the essential.

B. Training and Support

98. Culture does not change through exhortation alone. It changes through sustained investment in training, support, and the creation of environments where people feel safe to learn. In our experience, three aspects of training deserve particular attention.

99. First, training must be role-specific. The training a judge needs is different from the training a registry officer needs, which is different from what a self-represented litigant needs. Generic training sessions that attempt to cover all users tend to satisfy none. We have found that judges respond best to training that is closely tied to the workflows they actually use — how to navigate a case file digitally during a hearing, how to annotate documents on screen, how to use the tools that are directly relevant to adjudication.

100. Second, training must be ongoing, not a one-off event. A single training session at the point of system launch is insufficient. People forget. Interfaces change. New features are added. The institution must commit to continuing education, refresher sessions, and readily accessible help resources — whether in the form of help desks, quick-reference guides, or peer support networks.

101. Third, there must be a culture of psychological safety around technology use. People must feel able to say that they do not understand something, or that a system is not working for them, without embarrassment. This is especially important for judges, who may be reluctant to admit difficulty with technology in a professional environment. The institution must create space for that honesty.

C. Change Fatigue

102. I should also address change fatigue. When reform is continuous and overlapping, there is a real risk that those within the institution reach a point of exhaustion. New systems are introduced before the previous ones have been fully absorbed. Training requirements accumulate. Familiar workflows are disrupted repeatedly. The result can be resistance not to any specific change, but to the idea of change itself.

103. This is a genuine institutional challenge, and one that Singapore has not been immune to. We have had to learn that the pace of reform must be managed, that the sequencing of initiatives matters, and that there is a limit to how much change any organisation can absorb at once. The answer is not to stop reforming, but to be deliberate about what is introduced, when, and with what level of support. Transparency about the reform roadmap — so that people can see what is coming and prepare — also helps.

D. An Outward-Looking Perspective

104. Finally, cultural change requires a shift from an inward-looking institution to one that genuinely sees the system from the user’s point of view. Courts have historically been designed around the needs of the institution: the registry’s filing conventions, the judge’s scheduling preferences, the clerk’s administrative requirements. Technology reform both provides an opportunity as well as obligation to ask, as well, what the users, lawyers but increasingly self-represented persons, might need.  

105. This means testing systems with all actual users. It means listening to complaints not as resistance but as data. It means being willing to change something that the institution finds administratively convenient, but the user might find arbitrary or confusing. Our eLitigation system was designed more for the backroom or for chamber hearings than for the judge hearing a case is an example of what happens when the user perspective is not centred from the start. We are now trying to correct that, but correction is always harder than getting it right in the first place.

VI. Generative AI and Data Governance: Emerging Challenges

106. I turn now to what is perhaps the most consequential set of questions facing judiciaries today: the implications of generative AI, and the challenges that accompany it.

A. Generative AI and Judicial Reasoning

107. Generative AI tools are already being used by lawyers in many jurisdictions to draft submissions, conduct legal research, and summarise documents. The same for judges.  In Singapore, we have taken the position that AI tools may assist judges in reading and digesting materials. For the moment, we have not deployed AI in the use of the weighing of evidence, the application of law to facts, or the exercise of discretion. These raise significant issues.  

108. A judge who reads and adopts AI-generated reasoning without independent analysis has arguably not exercised the judicial function at all. In contrast, a judge who uses AI-generated text as a starting point and then applies independent critical judgment is in a different position. The boundary is real but difficult to police, and it requires ongoing institutional attention.

109. We are also acutely aware of the risks of hallucination, namely AI-generated content that is plausible but factually wrong. Several jurisdictions, including ours and India, have already encountered cases in which lawyers submitted AI-generated legal citations that turned out to be fabricated. The risk of this occurring in judicial reasoning is equally real and even more damaging to public confidence. Guidelines and training are essential, but they are not sufficient on their own. A culture of critical verification must be embedded.

110. India’s judiciary has been at the forefront of confronting these questions directly, including through judicial pronouncements on the use of hallucinatory AI-generated content.²⁶ These contributions are valuable to all of us.

B. Data Privacy, Cybersecurity and Court Records

111. The digitalisation of court records also raises significant questions of data privacy and cybersecurity. Court records contain some of the most sensitive information in any society: financial details, family disputes, criminal histories, medical evidence, trade secrets. As these records move from physical files in a registry to digital databases accessible over networks, the risk profile changes fundamentally.

112. In Singapore, court data is protected within the government’s cybersecurity framework, and access controls are applied to sensitive materials. But we recognise that no system is invulnerable, and that the consequences of a breach of court data would be severe, not only for the individuals concerned, but for public confidence in the justice system as a whole. We remain vigilant together with our government partners.  

VII. Sharing with India and All Judiciaries

113. I am very conscious that I speak to judges from a jurisdiction that has itself embarked on ambitious programmes of judicial digitalisation. India’s scale presents challenges very different from Singapore’s. But the underlying questions are the same: how to redesign process without unsettling principle, how to increase access without sacrificing fairness, how to build systems that are stable enough for courts and flexible enough for change.

114. Let me group some final points of sharing from Singapore under three headings: institutional lessons, practical lessons, and relational lessons.

A. Institutional Sharing

115. The first institutional lesson is about political will. Any path of meaningful change will attract resistance. Assistance can be given to those who genuinely need it, but the overall momentum must be forward. The crucial ingredient is the willingness of leadership to address and overcome objections rather than to be deflected by them. The second is a counsel against false contentment.  “If it ain’t broke, don’t fix it” takes the status quo as an untarnished good, ignoring its shortcomings and foreclosing improvement. A justice system may function tolerably and still impose unnecessary cost, delay and confusion on those who need it most. The third is about architecture. Fragmented systems slow everything that follows. Integrated architecture expands and allows for better growth. The investment in unified architecture is not glamorous, but it is necessary.  

B. Practical Lessons

116. One practical lesson is about the limits of consultation. Consultation is valuable and necessary. But there is a temptation to allow it to become a reason for indefinite delay. Consultation should inform reform, not substitute for it.  The second is about pace. You cannot wait for the slowest among the stakeholders to be ready. Accommodation should be made for those with genuine difficulties, and measures taken to bridge the digital divide. But the programme itself must continue to move. The third is about spending. Resources must be deployed. Such expenditure should be treated as investment, with clear articulation of expected returns. Spending must be deliberate and accountable. 

C. Relational Lessons

117. As for relationships, the most important point is to keep the user in view. Not only the institutional user but those outside, especially the self-represented.  Culture is important.  We can buy a digital system. But a proper culture, with readiness for change, accountability and learning can only be built with great effort.  

118. An important relationship in building technology is learning from others. Singapore participates in the Standing International Forum for Commercial Courts, which has been hosted by India; a regional technology meeting, which includes the Delhi High Court, and bilateral engagements including with India. These networks are a genuine source of practical learning, and keeping abreast of what others are doing is essential.

119. The inaugural Singapore–India Conference on Technology in April 2024 reflected a shared recognition that many of the ethical, legal and practical questions surrounding AI and court technology are cross-border questions, and that judiciaries can help one another by learning together.

VIII. Closing

120. Your Lordships, distinguished delegates, I began this address by recalling that Singapore’s very name is Sanskrit, and that its earliest legal institutions were born under the authority of this subcontinent. Two centuries later, the ties between our judiciaries are no longer those within a structure imposed by a foreign colonial power, but of common purpose and mutual learning between free and sovereign states.

121. The challenges we face are not identical, but they are similar in kind: efficient but fair and effective justice, serving all within the community.    The questions that must be faced are not ever that any one jurisdiction will or can answer alone.

122. I hope my sharing today has been helpful. I look forward to the discussion after. 

123. Thank you.