Chief Justice Sundaresh Menon: Keynote address at the TechLaw.Fest 2025

Distinguished guests

Ladies and gentlemen

I. Introduction

1. A very good morning. A very warm welcome to all of you, especially our friends who have come from abroad to join us for this year’s TechLaw.Fest. I cannot match the brilliance or the style with which Mr Thomas Chuang welcomed you, but the sentiment is just as sincere. It gives me great pleasure to deliver this morning’s keynote address at the tenth anniversary of TechLaw.Fest. From its early days as the International Conference on Electronic Litigation, the conference has come a long way in establishing itself as Asia’s premier forum on all things concerning law and technology.¹ Each successive edition has raised the bar for what a legal technology conference can offer, from a week-long virtual summit in the depths of the pandemic in 2020, to a metaverse-inspired exhibition in 2022.² This year’s conference builds on this by challenging us to “reimagine legal in the digital age”.

2. This morning, I will respond to this challenge by exploring how we might reimagine the complex and shifting intersections between law and technology. For a start, technology is already being reimagined at an extraordinary rate, and it follows that the legal sector simply cannot afford to fall behind. Just a few years ago, most of us would have found it inconceivable that entire movies could be produced within minutes, that a third of Microsoft’s new products would be coded by AI,³ or that humanoid robots may be deployed for rescue missions by 2027.⁴ The Fourth Industrial Revolution is truly well upon us, driven by the formidable combination of advanced analytics, modern engineering, and unprecedented computational power.⁵ It is therefore entirely unsurprising that many of Singapore’s leaders have described technology (and in particular, AI) as the catalyst for our next revolution.⁶

3. To keep pace with this revolution, which will likely advance faster than our understanding of it,⁷ we must commit to actively reimagining the law. I propose to expand on this theme in four parts.

(a) I will begin by describing two key trends that have come to characterise technological change, and that I suggest we should keep at the back of our minds as we think about the issues that we must confront.

(b) I will then consider how such change should lead us to reimagine the law in three areas.

i. The first is the justice system, where technology can be a game-changer in terms of its potential to enhance the administration of justice through automation, innovation and the elimination of conditions that can entrench the status quo. 

ii. The second is the legal services sector, where technology has already begun to redefine the role of the modern legal professional and the delivery of legal services. 

iii. The third is the law itself. Here, I will map out three ways in which technological change might challenge existing laws, and suggest three corresponding ways in which the law might respond. 

II. The evolving technological landscape

4. Let me begin by discussing two trends that we should keep in mind as we frame our response to technological change. The first is the complexification of technology, and by this, I mean that both individual forms of technology as well as the wider technological space are growing in complexity.

5. In recent years, we have witnessed remarkable advancements in technologies such as blockchain and generative AI.⁸ This is true not only at a technical level but also in terms of what the average consumer can access. When I was working on this keynote address in July this year, OpenAI had just unveiled the agentic version of ChatGPT. This was touted as having the ability to execute complex tasks – from analysing a user’s calendar and preparing briefings for upcoming client meetings based on recent news, to planning and purchasing ingredients for that very complex matter of a Japanese breakfast for four.⁹ These highlight the ability of agentic models to pursue complex goals from start to finish, with limited supervision¹⁰ and curated personalisation.¹¹ Yet, not even a month later, OpenAI launched GPT-5, which it described as its “smartest, fastest and most useful model yet”, and which is said to outperform ChatGPT Agent on a range of economically important tasks spanning across 40 occupations.¹²

6. Complexity emerges not only from the breathtaking pace of these breakthroughs, and the seemingly infinite range of possibilities they hold, but also from certain persistent attributes inherent in AI systems. One such attribute is AI’s opacity.¹³ AI algorithms are often likened to black boxes that defy inspection and generate non-intuitive outcomes that resist our understanding.¹⁴  And while much progress has been made in the field of explainable AI, how AI reasons remains inscrutable and mystifying to most of us.¹⁵ Another such attribute is the apparent inevitability that a specific type of AI, namely large language models (“LLMs”), will sometimes produce responses that are completely false and unusable. Again, while progress is being made to inhibit the incidence of hallucinations, it nevertheless persists and it therefore complicates efforts to harness the many benefits of LLMs. Yet, despite these realities, perhaps the most profound aspect of technological change in general, even going beyond AI, is not technology’s ability to replicate human activity at scale or to do more with less, but to completely revolutionise how things are done in formerly unimaginable ways.¹⁶

7. There is, as I mentioned, a second dimension to this challenge, which relates to the complexification of the broader technological space. As generative AI models exemplify, we are seeing a proliferation of technological solutions on the market that feature diverse uses and applications. This underscores the risk of thinking about such technologies in broad strokes or in reductive terms. Instead, I suggest it would be more accurate and productive to think about different forms or applications of technology and of AI as points along a spectrum of varying complexity. We might think of the autocomplete function on our smartphones as falling at the lower end of the spectrum, when compared to tools that use AI to predict legal outcomes or perform judicial analytics. And we can then calibrate our approach depending on where we are along that spectrum. 

8. The second broad trend that I want to touch on is the ease with which new technologies can be deployed across borders, and the transnational implications this has for the law. Regulatory complexity is a foremost concern, because inconsistent approaches can either lead to weak and ineffective oversight, or it can drive cost up if, for instance, providers were required to develop geo-specific versions of each of their products. It may also promote regulatory arbitrage.¹⁷ By way of example of this complexity, the European Union’s AI Act classifies AI systems based on their risk levels and imposes corresponding obligations, ranging from outright prohibition to lighter transparency requirements and even to no requirements at all.¹⁸ The UK’s AI Regulation White Paper, in contrast, indicates that the UK will take an “agile and iterative approach” based on five AI governance principles, and these principles will not be placed on a statutory footing, at least initially.¹⁹ And this is similar to the non-binding approach that we in Singapore have adopted thus far. One immediate consequence of such regulatory divergence is the challenge posed to compliance work. 

III. Reimagining the justice system

9. So, what do these trends mean for the law? Let me approach this first from the perspective of the justice system. 

10. I have previously spoken about the model of the “extended court”, which sees the court’s role as extending beyond adjudication to helping users understand legal information and navigate the justice system.²⁰ The Singapore Judiciary has for some years embraced this shift towards becoming a more user-centric institution. To this end, we view technological innovation, covering the whole range from simple and reliable varieties all the way to emerging forms of AI, not just as another tool in our arsenal, but potentially as a force multiplier and a strategic driver of transformation.²¹

11. Let me unpack this by considering three different ends to which technology has been adopted in the courts, namely, to automate, to innovate and to eliminate the conditions that can entrench the status quo. This draws on a taxonomy proposed by Professor Richard Susskind that focuses our attention more granularly and contextually on the outcomes that technology is employed to achieve.²² Let me first elaborate on this framework before I consider it in the context of judicial work.

12. Automation refers to the use of technology to perform tasks that used to require heavy human involvement or effort, but with greater efficiency and consistency. Innovation seeks to do things that were previously not possible or even conceivable, but to attain outcomes that have long been sought. Elimination goes one step further by using technology not simply to solve a given problem, but perhaps to remove the conditions that gave rise to the problem in the first place, thereby altering the status quo.²³ These different ways to harness technology might overlap to some degree, but they afford us a useful framework for thinking about how and why we might want to pursue certain uses of technology. I can illustrate this with reference to our evolving relationship with cash and how we access its functionality. Whereas cash used to be dispensed by human tellers at bank branches, ATM machines both automate this process and offer an innovative means for providing convenient and round-the-clock access to bank notes. But even greater transformation has followed the mainstreaming of credit cards, mobile wallets and digital payment applications, which are gradually eliminating our reliance on cash. 

13. More directly, we can see the framework’s utility, at a conceptual level, by considering how it might apply to potential uses of blockchain technology. Smart contracts might be seen as a form of automation, in how they can conceivably trigger the performance of certain types of contractual obligations when predefined conditions are met.²⁴ Innovation may be exemplified by the potential use of blockchain to store electronic evidence, by leveraging the blockchain’s verification mechanisms and decentralised nature, by which we might yet discover new ways to prevent tampering and ensure authenticity.²⁵ Finally, Decentralised Finance (“DeFi”) could offer an example of elimination. By supporting direct, peer-to-peer transactions, the blockchain may one day eliminate some of the conditions that give rise to fraud and manipulation in centralised financial systems.²⁶ The immutable nature of the blockchain means that once a transaction is recorded, it becomes unalterable and tamper-proof. 

14. So, turning to the courts, we have first, as an example, automated a number of our processes, particularly those that are time-consuming or repetitive.²⁷ One example is the Authentic Court Orders system, which enables users to authenticate court orders by scanning a QR code linked to a secured database, and replaces the need to obtain a certified true hardcopy.²⁸ We intend to expand this to other court documents in the near future. 

15. The next area is innovation. Here, I highlight three examples. First, we introduced “e-Services” for more straightforward Divorce and Probate applications. These allow self-represented persons (or “SRPs”) to prepare and file court papers without needing to step into a courthouse or to see a lawyer. We auto-populate data as far as possible, using whole-of-government databases wherever feasible.²⁹

16. Second, we wanted to enable motorists and other road users to understand the sort of quantum of damages typically awarded in different motor accident scenarios. So, we designed a Motor Accident Claims Online (“MACO”) simulator and worked with the Singapore Academy of Law to present this to users. The tool provides a quick, free and anonymous assessment of the likely allocation of fault and quantum of damages arising from typical motor accidents.³⁰ It places would-be litigants in a better position to assess the viability of their claims, regardless of their resources or legal literacy, and this can have an outsized impact for lower-value claims where the cost of legal advice may be disproportionate and even prohibitive.³¹

17. A third example is our Community Justice and Tribunals System (or the “CJTS”). This is used for small claims, employment claims, community disputes and harassment cases, and was designed with self-represented persons in mind. It provides for online filing, case management, and alternative dispute resolution in the form of e-negotiation and e-mediation.³² And through our collaboration with Harvey.AI, we have deployed generative AI tools for document translation in the Small Claims Tribunals. I am confident that we will shortly launch expanded functionalities that will next cover the summarisation of case documents.³³ These are important first steps and we are delighted to collaborate with Harvey in helping SRPs in a very real and tangible way, in their endeavour to maximise their ability to access justice.

18. Finally, we have leveraged technology to question and, in some respects, to eliminate some of the premises and assumptions that have traditionally underlain the design of our court processes.³⁴ Many of us will recall how in-person hearings came to a halt in the wake of the Covid-19 pandemic. Faced with the grim prospect of having to adjourn hearings indefinitely, the courts moved quickly to bring hearings online.³⁵ This ensured that hearings could continue with minimal disruption by eliminating the need for parties to be in the same place, and that, in turn, led us to pilot asynchronous hearings, which eliminate not just the physical gathering, but the need for the parties even to gather at all. Judges can still pose questions and obtain clarifications as before, but without the impediment of scheduling constraints.

19. Adopting this framework of automation, innovation and elimination enables us to think more systematically about the role and limits of technology in the courts and gives rise to at least two specific insights. The first is that despite the allure of high-profile or dramatic change, we must not lose sight of the considerable unrealised value that lies in pursuing low-tech innovation and simple process automation.³⁶ At the same time, we also must not make the mistake of necessarily equating low-tech solutions with low risk. The UK’s Post Office Horizon scandal is a reminder of this. Owing to an undiscovered fault in the Horizon accounting software, hundreds of sub-postmasters were falsely accused of theft and accounting fraud, with many convictions reversed only years later and after devasting personal consequences.³⁷ Automation bias is another related and under-rated risk. This describes the human tendency to attach undue weight to machine-generated results and is one of several reasons why the automation of judicial decision-making continues to be deeply controversial.

20. A second insight relates to innovation and elimination, which have in common the potential to radically transform settled practices. In such cases, we need to be especially mindful of the dislocating effects that fundamental change can bring. So, when we first rolled out remote hearings, we made it a point to ensure that video-conferencing facilities were available within the court premises, so that SRPs who lacked suitable equipment or even internet access could nonetheless continue to participate fully in court proceedings. And in response to increasing reliance on the part of our court users on Generative AI, we issued guidance to clarify our position as to the use of AI in court proceedings and to caution against its risks.³⁸

IV. Reimagining the legal services sector

21. The framework of automation, innovation and elimination can similarly shape how we think about the legal services sector, which brings me to the next part of my speech. But here I want to focus on a different point, about how technology is transforming the role of the modern legal professional and the delivery of legal services. 

22. As I highlighted in my Mass Call address last year, I believe there is growing acceptance of the notion that Generative AI will bring about a paradigm shift in lawyering, and that we should all take this development seriously.³⁹ It used to be the case that while AI’s ability to parse syntax was well-established, it remained less capable at tasks that are thought to require complex semantic reasoning using specialised domain knowledge.⁴⁰ This appeared to change in 2023, when GPT-4 sat for the bar exams – or at least a simulated version of the US Uniform Bar Exam – and passed it by a margin. It excelled in areas such as contract law and evidence, where it achieved over 80% accuracy on the multiple-choice questions.⁴¹ It also produced several long-form answers that were comparable to representative “good” scripts.⁴² Overall, this was assessed as reflecting a 26% improvement from ChatGPT, which was then the best performing model.⁴³ And just this May, the UK Solicitors Regulation Authority authorised the first “purely AI-based” law firm to provide regulated legal services.⁴⁴ The firm, Garfield.AI, claims that it will rely on AI to draft pre-action letters, handle settlement negotiations and undertake trial preparation, to help small and medium enterprises recover their unpaid debts.⁴⁵

23. Of course, there is a difference between being able to pass the bar exam or produce letters of demand, and being a practising lawyer, much less an effective one. But as one IBM report suggests, while AI may not replace people, people who know how to use AI will replace those who do not very quickly.⁴⁶ Whereas lawyers have traditionally been gatekeepers to legal knowledge and expertise, we should anticipate a future where some legal problems will be resolved with the right technological tools, even without lawyers, and other, more complex problems will be referred to lawyers who will be equipped with even better tools.⁴⁷ Let me explore what this will mean for the skillsets and mindsets that legal professionals will need in order to ride the wave of technological change effectively.

24. I suggest that the core skill will be the technical proficiency to understand how technology works and how to use it effectively within their workflow. What this means will likely vary across one’s career. For our junior colleagues, the foremost priority might be learning how individual tools actually work (or not) and how to apply them directly in their daily work. Senior lawyers will have the added responsibility of having to decide, at a strategic level, which new technologies to deploy and how they should be leveraged to build scalable business cases that can justify their initial investment cost.⁴⁸

25. Being proficient with technology entails knowing both what it can do and what it cannot. We often speak of the danger of unknown unknowns, but known unknowns can be just as dangerous if we remain blind to the risks that they pose.

26. The US National Institute of Standards and Technology has identified over 40 ways in which human, systemic and computational biases can affect AI.⁴⁹ Some can be mitigated with prompt engineering techniques, such as instructing a model to approach a problem step by step or to ignore earlier attempts at prompts, but these techniques generally have their limits.⁵⁰ At the same time, as the Senior Parliamentary Secretary (“SPS”) observed, generative AI is often itself unaware of its own limitations, something we see in its uncanny ability to persist in false or hallucinated answers with absolute and remarkable confidence. Further yet, AI may even paper over the limitations of its human user by being unduly agreeable. This is known as sycophancy, or a tendency of some models to align their responses to match their users’ views, as a product of their training to be useful and supportive.⁵¹ And this has the real prospect of reducing AI from a helpful sounding-board into an echo chamber that affirms whatever the user wishes to hear. All of this highlights the pressing need to regard digital literacy as an essential part of the tools of a lawyer’s trade and to seriously invest in building such literacy. One convenient place to start might be the Guide on Prompt Engineering for Lawyers jointly developed by Microsoft and SAL, which sets out prompt engineering techniques with practical examples, and highlights good AI practices for lawyers.

27. Being proficient with technology is important, but this must be anchored in a fundamental transformation of mindset. Critical to this is our capacity to adapt to change and uncertainty.⁵² There are at least two aspects to this.

28. First, we must embrace continuous learning. What we know about technology today will quickly become outdated, a phenomenon associated with the decreasing “half-life of knowledge” and the rapid advance and development of technology. The half-life of knowledge refers to the ever-diminishing amount of time before half of the knowledge in a particular field becomes either obsolete or irrelevant.⁵³ For professional skills, this was once estimated at 10 to 15 years. Today, it has fallen to 5 years and even shorter in the case of technical skills.⁵⁴ As I have observed on a different occasion, in the face of such accelerating change, we can only remain a “learned profession” by becoming a profession of learners.⁵⁵

29. Second, we will have to embrace organisational change. We must acknowledge that the billable-hour model, in making time its currency, has the potential to disincentivise innovation that is designed to save time. Moreover, clients who enjoy a growing list of options – such as technically-skilled in-house counsel or efficiently-priced alternative legal service providers (“ALSPs”) – may no longer accept paying for such time. This is particularly true for high-demand and high-volume legal services, such as document review and IP management. It is therefore crucial that firms carefully evaluate where their value proposition lies and how it can be enhanced. 

30. Indeed, some firms are already responding to these shifts. One such response has been to bring technological capabilities in-house. Just this March, the New York law firm Cleary Gottlieb acquired Springbok AI and welcomed its co-founder, data scientists and AI engineers to form the firm’s new AI acceleration team. The move saw the onboarding of Springbok’s proprietary platform, SpringLaw, and now enables the firm to create custom AI-powered solutions which it says “sets [it] apart from [its] competitors”.⁵⁶ For other firms, I have previously suggested that we might see a shift from the prevailing ”pyramid” model, which comprises a few, more senior partners supported by corps of associates and paralegals, towards a “rocket” model, which comprises a leaner team of digitally savvy lawyers, legal technologists and allied legal professionals.⁵⁷ Pricing structures may have to change, with different structures for standard or routine work on the one hand, and bespoke, complex or strategic work on the other.⁵⁸

31. None of this diminishes the role of the lawyer. To the contrary, it invites us to reimagine the role, to enable the lawyer to focus and capitalise on higher-value, human intellectual work. And to do this well, law firms will need to be more intentional about the division of labour between their lawyers, technology and other allied specialists,⁵⁹ and ensure that lawyers receive proper support to cope with the more complex work that is channelled to them.

32. This mindset change also extends to in-house lawyers, who must see themselves as responsible not only for ensuring corporate compliance, but also as strategic advisors to their companies. This entails being strategic at two levels. First, in-house lawyers should take advantage of technology to automate routine tasks, to deliver more cost-efficient advice, and to better communicate legal concepts to non-legally trained colleagues.⁶⁰ Second, as they are uniquely placed to address growing regulatory complexity, we can expect in-house counsel to assume an expanded role in risk management and business continuity planning.⁶¹

V. Reimagining the law

33. Finally, let me turn from the impact of technology on legal practice to its imprint on the law. I will first outline three ways in which new technologies challenge us to reimagine the law, before suggesting three possible responses to those challenges.

A. Challenges to the law arising from new technologies

34. The challenges can be termed “application uncertainty”, “normative uncertainty”, and “legal lacunae”.⁶²

35. Application uncertainty arises from the inherent difficulty of applying existing law to some of the novel circumstances created by new technologies.⁶³ For example, when electronic communications became a widespread means of conducting business, the question arose as to whether and how such communications would fulfil the requirements for the formation of a contract. This arose because the law on contract formation had developed in an era when goods were offered for sale in physical shop displays, and communications were done by post or fax, and the postal acceptance rule was still a matter of practical interest.  More recently, the development of crypto assets has given rise to considerable academic and judicial debate over whether such assets are property, and if so, what kind of property they are. 

36. The second challenge, normative uncertainty, arises when it is clear that existing rules would apply to the new technology, but it is less clear whether they should, and whether this would produce desirable results. Examples might include the application of rules governing the commencement of claims to anonymous defendants, and the application of principles for the assessment of damages to very volatile crypto assets. In these cases, the generally applicable rules of procedure or of the assessment of damages would apply, but it is less clear if they ought to.⁶⁵

37. The third challenge, the legal lacunae, arises when no legal regime exists that can adequately govern the new technology and its associated social change. Examples include the battery of risks posed by new technology including liability for accidents caused by autonomous vehicles, intellectual property rights in both the data used to train generative AI models and their output, and the data protection risks that arise from the collection and use of vast amounts of personal data.⁶⁶

B. Revisiting and reimagining the law 

38. If we conceptualise the challenges posed by new technologies in the three ways I have described, I suggest we should then think about our responses in three corresponding ways – namely to clarify the law, modify the law and create new regulatory frameworks.

39. Let me begin with clarifying the law, which addresses application uncertainty. This entails resolving any doubts as to whether and how certain legal rules might apply in the context of new technologies. 

40. You heard the SPS speak about the Government’s proactive, yet measured, stance in legislating. A useful illustration can be found in Singapore’s approach to e-commerce. In 1998, Parliament passed the Electronic Transactions Act (the “Act”),⁶⁷ which supported e-commerce transactions⁶⁸ by introducing detailed rules on the formation of contracts through electronic communications, and the use of electronic signatures and records.  While the common law was broad enough to give effect to electronic transactions,⁷⁰ the Act put the legal position beyond doubt. It also clarified pertinent issues, by deeming for instance that a message from a party’s automated computed system originates from that party,⁷¹ and by providing for the time and place that electronic records are deemed to be despatched and received.⁷²

41. In drafting the Act, the Government was guided by four needs:⁷³

(a) first, to conform to international standards so as to be globally integrated; 

(b) second, to avoid over-regulation; 

(c) third, to be flexible and technologically neutral to adapt quickly to a fluid global environment; and 

(d) fourth, to be predictable and transparent. 

42. I suggest that these principles seek to balance the concomitant need for certainty and flexibility when responding to new technologies, and they are emblematic of the approach that Singapore has consistently endeavoured to take. Indeed, it is noteworthy that Parliament re-enacted the Act in 2010, 12 years after it was first enacted, and has continued to amend it regularly to keep pace with contemporary developments and new treaties.⁷⁴

43. Another area where clarification of the law has been key is the application of property law to crypto assets.⁷⁵ In Singapore, the High Court has held that crypto assets are property, and specifically that they are things in action.⁷⁶ Meanwhile, the UK Law Commission has concluded that some crypto assets are neither things in action nor things in possession, and has recommended legislating that they fall within a third category of things, the boundaries of which should be left to development by the common law.⁷⁷ The UK Parliament has introduced a bill to give effect to this proposal.⁷⁸ Whatever position each jurisdiction adopts, it will need to balance these twin considerations of providing certainty and retaining flexibility.

44. The second approach is to modify existing legal rules. This deals with the challenge of normative uncertainty and requires us to first identify the mischief caused by existing rules in the context of new technology, before examining whether extending, varying or creating exceptions to those rules would be most appropriate. 

45. A simple example lies in how the law deals with anonymous defendants. To commence proceedings, a claimant must use prescribed forms⁷⁹ that require the defendant’s name to be stated.⁸⁰ But this is not possible in certain circumstances, such as when online transactions are concluded anonymously. Our courts have held that, where the defendant is unknown, process may nonetheless be issued so long as the description of the defendant is sufficiently certain as to identify the persons who fall within and without its scope.⁸¹ This gives effect to the underlying purpose behind the original rule in a functionally equivalent way.⁸² And this is consistent with the provisions of our Rules of Court, which provide that the prescribed forms are to be used “with such variations as the circumstances require”.⁸³

46. A more complex example concerns the assessment of damages where volatile assets are involved, with crypto assets being the archetype. In general, damages for breach of contract are assessed as at the date of the breach,⁸⁴ and this is known as the breach date rule. Our courts have observed, although not yet settled the point conclusively, that the breach date rule may not always provide the best or most appropriate methodology.⁸⁵ If the asset’s value fluctuates dramatically after the date of the breach, the rule might not properly reflect the quantum of loss that a claimant has suffered.⁸⁶ Indeed, other jurisdictions have adopted different valuation methods to deal with this issue.⁸⁷ Ultimately, it remains to be seen if any modification to the law is necessary. But if it were, one would have to bear in mind the underlying purpose of damages for breach of contract, which is to put an innocent party in as good a position as possible as if the contract had been performed.⁸⁸

47. The third approach to reimagining the law is to create new regulatory frameworks to address legal lacunae. While judge-made law has much to commend it, such as its incremental and casuistic approach, its largely reactive nature is not one of its strengths.⁸⁹ In the face of rapid or systemic technological change, regulatory intervention may often be needed in order to mount a timely and comprehensive response. Whether this should take the form of legally binding rules, soft law instruments or industry standards would depend on a number of considerations, such as: (a) whether formal regulation would be unduly onerous, expensive or inflexible; (b) whether regulators have sufficient expertise and information to regulate the technology effectively; (c) whether robust governance structures and incentives already exist in the form of self-regulating bodies or industry; and (d) whether sufficient transparency and public scrutiny exist for the market to police non-compliance without external oversight.⁹⁰

48. A good example of a non-binding framework is Singapore’s Model AI Governance Framework, which guides organisations in the responsible adoption of AI. This translates ethical principles into practical recommendations,⁹¹ is accompanied by a Compendium of Use Cases⁹² and is regularly updated.⁹³ Together with the Model AI Governance Framework for Generative AI issued last year, these frameworks adopt a balanced approach, addressing concerns such as bias, opacity and hallucination, while facilitating innovation.  

49. An example of a hybrid model is the Personal Data Protection Act (“PDPA”). With the exponential growth in the collection and processing of personal data, Parliament found it necessary to establish a legislative regime to address concerns over the use of personal data and to maintain trust in organisations that manage data.⁹⁵ This omnibus framework replaced a pre-existing patchwork of laws and sectoral approaches, to ensure a baseline standard of protection across the economy. A unique feature of the Act is that it is supplemented by a body of advisory guidelines issued by the Personal Data Protection Commission (“PDPC”).⁹⁶ These guidelines are not legally binding, but they indicate the manner in which the PDPC will interpret the Act⁹⁷ and offer practical guidance to organisations.⁹⁸

50. Having examined the three challenges that new technologies might pose to the law and possible responses to these, let me now draw the discussion together with two observations. 

51. The first is that while I have mapped each challenge to a response, these divisions are neither prescriptive nor exhaustive. A combination of responses will often be necessary in light of the complex and multifaceted nature of technology. What I have endeavoured to propose is a conceptual framework for reimagining the law’s responses to new technologies in a more structured way. 

52. Second, in determining the content of a rule or regulation, we must remember that we are aiming at a moving target.⁹⁹ We should be prepared for our projections of technological change to be wrong in their direction or degree.¹⁰⁰ We should also avoid the extremes of being overly permissive on the one hand, or unduly precautionary, on the other hand.¹⁰¹ When confronted with new technologies, policymakers face what is known as the “Collingridge Dilemma”. Early in the life of a technology, its social consequences may be unpredictable and difficult to regulate; yet, once undesirable consequences are discovered, the technology may be too entrenched to easily control.¹⁰² The goal, then, must be to keep pace with technological developments, and to avoid either outrunning them or falling behind them. We can achieve this through various techniques, such as relying on the incremental development of the common law,¹⁰³ or designing “adaptive legislation” that combines general standards, periodic reviews and carve-outs such as safe harbours.¹⁰⁴

VI. Conclusion

53. Over the course of this address, I have spoken about reimagining the justice system, the legal services sector and the law in response to technological change. Allow me to conclude by identifying some common threads that tie these areas together. These take the form of two attributes that I suggest are critical for judges, lawyers and regulators alike.

54. The first attribute is a commitment to multi-disciplinarity. We live in times when interdisciplinary knowledge and skills are not merely a distinguishing quality on a resume, but a defining attribute for the modern professional. I have spoken at length about what this means for lawyers, and the same holds true for the Judiciary. We have invested significantly in developing our judges as “learning judges”, with a mindset of lifelong and self-directed learning, including in adjacent fields such as technology, statistics and forensic sciences.¹⁰⁵ And this should also apply to our law students. Following the recommendations of the Working Group for the Reform of Legal Education, our law schools are working on integrating technology and data literacy skills into the mandatory curricula and enhancing the related electives.¹⁰⁶

55. Second, we must adopt a transnational outlook and carefully examine how foreign jurisdictions have approached technological change. Issues and opportunities that appear novel to us will often have received consideration elsewhere, and we would do well to draw upon the combined wealth of experience and learning that can be derived from a comparative approach. The significance of a transnational focus is sharpened by the reality that we very rarely operate in legal or geographical silos, and that coherence across legal regimes is instrumental to supporting transnational commercial activity.¹⁰⁷ This is especially important given the problem of regulatory arbitrage that I mentioned at the outset of my address.

56. The Singapore Judiciary has taken an active role in promoting dialogue between judiciaries on law and technology. We regularly convene the Meeting of Chief Justices and Judges in Charge of Technology with several like-minded judiciaries, establishing a standing forum for the discussion of emerging issues and best practices concerning the use of technology in our courts. This spirit of cooperation can be found in many parts of the world. Just this year, LawtechUK, a digital transformation initiative backed by the UK Ministry of Justice, established an International Jurisdiction Taskforce that is chaired by the Master of the Rolls, Sir Geoffrey Vos. The taskforce seeks to address legal barriers to the adoption of digital assets, international digital finance and digital trade, and it has brought together an independent panel of legal experts from some of the most widely-used private law systems, including the UK, the US, Japan and Singapore.¹⁰⁸

57. Returning closer to home, what makes TechLaw.Fest such an invaluable and indeed, exceptional, event is how it combines the value of multidisciplinary perspectives with a transnational outlook. Each year, it brings together legal professionals, technologists, leaders and regulators from across the region and beyond, to engage in thought-provoking discussions that advance the frontiers of law and technology. That we now mark its tenth anniversary is a testament to its enduring relevance to the legal services sector in Singapore and in the region. I commend the conference’s organisers, past and present, for their remarkable vision and tireless dedication in developing TechLaw.Fest into the flagship conference that it is today, and I look forward to its success continuing in the decades to come.

58. Thank you very much, and I wish all of you a most engaging and fruitful conference.