교육부 장관 데스몬드 리의 스트레이츠 타임즈(ST) 교육 포럼 연설 「고등교육의 인공지능 - 과장인가, 희망인가」

Ms Karamjit Kaur, Associate Editor, The Straits Times My fellow panellists

1. Good afternoon. I'm very glad to be here with you and our panellists to discuss the impact of AI on our education landscape.

2. In fact, a couple of weeks ago I was in Tallinn, Estonia, for the International Summit on the Teaching Profession (ISTP). The challenges and issues that we are grappling with here, both in general education and higher education in Singapore, are not unique to us. They are issues that are being grappled with all around the world.

3. AI is advancing at a remarkable speed, and Generative AI systems are now widely accessible to students, educators, and workers. So, we can expect big changes in how we learn, work, and live. Some of it will just explode at the seams, and some of it will be incremental creep.

4. This raises fundamental questions for us: If thinking and doing can increasingly be done by machines, what would our role be as humans, as people?

5. And what is the role of our Institutes of Higher Learning (IHLs), which have long been central to nurturing our people with the knowledge, skills and values to thrive?

6. To answer this, it might be useful to step back and consider how education has evolved. Education systems have never been static. They have continually adapted alongside changes in society, economy and technology.

7. The earliest universities in medieval Europe, such as Oxford, were small scholarly communities centred on dialogue and debate. Learning took place through lectures and oral defences. Knowledge then was scarce and carefully transmitted through handwritten manuscripts.

8. Centuries later, the rise of research universities across Europe and America transformed higher education as we know it.

9. Laboratories, scientific inquiry, and disciplinary specialisation flourished – particularly as science and engineering advanced during the Industrial Age and the 20 th century.

10. Universities became not just places to transmit knowledge, but to create it.

11. And in Singapore, our IHLs have similarly evolved over time.

12. Our universities began by focusing on foundational disciplines, training professionals needed for a fledging nation.

13. Over time, we have expanded and diversified by strengthening interdisciplinary learning, building research capabilities, and developing applied pathways through universities like SIT and SUSS.

14. Likewise, our Polytechnics and ITE were first established to train skilled technicians to support Singapore's industrialisation.

15. These institutions have since evolved from purely providing technical training to preparing work-ready graduates in a wide range of disciplines including the Arts, Design and Media, Business, Health Sciences and so on.

16. We have also strengthened Continuing Education and Training (CET), with our IHLs playing a major role in our SkillsFuture movement.

17. Lifelong learning is increasingly important in an era where technological cycles are getting shorter, and we need to constantly refresh our knowledge and skills.

18. Today, we are witnessing yet another major technological shift.

19. In just a few years, we have seen how powerful and widespread AI systems have become. ChatGPT has become a household name, and each new generation of Generative AI models is becoming more capable than the previous.

20. These developments present both opportunities and risks.

21. For our students, AI can provide more personalised and tailored educational experiences. However, there is also a risk that AI may become a shortcut that bypasses thinking and learning, and causes cognitive offloading, and that students become overly reliant or trusting of AI output.

22. For the workforce, AI can augment human capabilities, enabling workers to be far more productive.

23. This is particularly valuable for small countries like Singapore.

24. But AI can also create economic disruption, substantially reshaping job roles and the skills in demand.

25. These opportunities and risks compel all of us to respond. Whether we are ready or not, AI is here to stay, and it is advancing rapidly – our students are already using it, and companies are racing to incorporate AI into their workflows.

26. But the path ahead is not clear, with even industry leaders uncertain about how AI will fully transform their sectors.

27. So, we must feel the stones as we cross the river, and cannot take a wait-and-see approach. Across our schools and IHLs, we need to learn, adapt, and grow.

28. Our schools must help our students emerge into the workforce ready to add value in an AI age.

29. We must also help our existing workers upskill through CET to augment their experience and skills with new skills needed for tomorrow's jobs.

30. And so the question is – how will we do this?

31. Increasingly, our work will move away from routine application of knowledge, towards judgement, creativity and invention, relationship-building, and collaboration, with both people and intelligent systems.

32. To better understand how AI is reshaping work, it is useful to distinguish between what some have described as "horizontal" and "vertical" capabilities.

33. Horizontal capabilities are broad, general tasks that cut across many jobs – such as writing, summarising information, or preparing presentations. These are the areas where AI is already highly effective, and improving very rapidly.

34. Vertical capabilities, on the other hand, are domain-specific and built on deep expertise. They involve applying knowledge in complex, real-world contexts, where judgement, experience, and responsibility matter. Such vertical expertise remains harder to replicate.

35. This is why education must evolve in a deliberate way.

36. We must equip students not just with the abilities to use "horizontal" AI tools, but with the depth of knowledge, experience and judgment to use them well in "vertical" applications that will be true game-changers.

37. To guide how we develop these capabilities, MOE has introduced a common framework that we call the four 'Learns', to enable our students to be AI-ready. What are they?

38. First, learn about AI . Students need to know what AI is, how it works, what its impact is, and what its limitations are. And this continues to be dynamic as technology evolves.

39. Second, learn how to use AI . Students will learn how to harness AI tools effectively and responsibly.

40. Third, learn with AI . Educators will incorporate AI in teaching and learning to enhance our student outcomes and learning.

41. Fourth, and most importantly, learn beyond AI so we remain masters of technology. When AI can produce outputs instantly, the ability to evaluate, challenge, and take responsibility for those outputs become indispensable.

42. Our schools and IHLs must continue to nurture precisely these human capabilities in our students. These capabilities must be built progressively across different stages of education.

43. At every stage, we must try to build strong foundations. Students must first develop a firm grounding in core concepts and deep disciplinary knowledge.

44. Even in an age where information can be easily accessed or generated by AI, these fundamentals remain essential.

45. In fact, we think that those who are best able to use AI are those with the deepest expertise, because they can ask better questions, interpret output more critically, apply them meaningfully, and use AI to push the boundaries.

46. Beyond deep knowledge, we must also hone deep instinct, strengthen critical thinking, and ground students in strong moral and ethical judgement.

47. Hence, we have sought to carefully calibrate our approach:

48. For younger students, we will prioritise learning the fundamentals first. And so a lot of that has got to do with tactile learning, real world physical learning in the classroom and in the outdoor classroom.

49. So, we introduce AI only at Primary 4, under close supervision and with low exposure.

50. Thereafter, we increase AI usage progressively, grounded in pedagogical practice. At this stage, AI should seek to function like a teacher – prompting students to think and asking them questions, guiding them to derive learning – rather than spoon feeding answers or giving lazy shortcuts.

51. At the IHL level, students must both leverage AI on one hand and deepen independent thinking on the other – learning to question, analyse, and create and communicate with confidence.

52. Taken together, these shifts will fundamentally reshape the classroom and the campus. The classroom and campus of the future will need to be anchored in four key shifts: Inquiry, Application, Collaboration, and Adaptability.

53. To bring this vision to life, and this is a work in progress, we must rethink how educators teach their students and do their work, across three main areas.

54. First, AI for productivity . Our educators and institutions must find ways to harness AI to improve their administration in the IHLs, to reduce repetitive tasks, or to set up so that they then can go in to teach, so that educators can free up bandwidth to focus on meaningful interactions with students in classrooms and laboratories.

55. For example, educators in NUS, NTU, SIT, and SUTD have begun using AI tools to support the grading of student work.

56. NUS's AI tool to grade English Competency Tests has been used to grade over 3,000 students per year with an assured accuracy of beyond 95%, and the time saving is more than 100 man-days per year.

57. For students, the AI marking also provides higher consistency and fairer outcomes, since AI does not show inter-rater variability.

58. Notwithstanding this, all AI-assisted grading will and must still be reviewed and supervised by educators, who ultimately retain full responsibility for the accuracy of the final grades. And this is important for our students.

59. Our colleagues at ITE are also provided with AI tools such as D2L Lumi Pro, which help educators generate the first cut of their teaching materials, which they can then adjust and customise based on the learning needs of their students.

60. Second, AI for pedagogy . We must reimagine how teaching and learning take place. For example, by incorporating AI tools directly into teaching and learning, with appropriate guardrails in place.

61. Lecturers in our five Polytechnics – Nanyang Polytechnic, Ngee Ann Polytechnic, Republic Polytechnic, Singapore Polytechnic and Temasek Polytechnic – embarked on what they call the Joint Polytechnic Analytics in Education (AiE) project, where they use AI to better and more systematically identify and support the learning needs of at-risk and underperforming students early, and support the design of data-informed teaching strategies for these students.

62. At SUSS, they designed the Adaptive Learning System to diagnose learning gaps and help our learners level up their competencies by providing individualised support such as hints and feedback.

63. We may also see a rise of AI-enabled assignments, where students are assessed on how effectively they collaborate with AI to solve problems or create products.

64. Third, AI for empowerment . Our classrooms must equip students with the capability to succeed in an AI-transformed world, where jobs and industries will evolve rapidly.

65. This includes building up students' ability to apply AI meaningfully within their own disciplines.

66. At SUTD, its Urban Science, Policy & Planning Masters students designed and facilitated a master-planning workshop using their own custom-built AI agents. The students used Generative AI to translate participants' ideas into visuals instantly, enhancing the quality of discussions as participants can move beyond mere dialogue to seeing and visualising how their ideas might look like.

67. The AI agents can also provide feedback on whether the ideas are feasible from the technical and infrastructural angle. And imagine the capabilities this gives the teachers and lecturers.

68. At SMU, its students in the "Storytelling with AI" course use AI to generate high-quality multimedia content without the need for extensive budgets or filming resources. And this allows them to learn more frequently and learn more intensively.

69. To give you a better sense of what this looks like, and because SMU is our host, we made a request, and SMU has prepared a short for your viewing.

70. This is an example of how our IHL faculties are experimenting, trying and weaving AI to enable learning.

71. Beyond teaching students how to use AI in their respective domains, we also need to strengthen students' 21 st century competencies and uniquely human abilities that AI cannot easily replicate today, such as critical thinking, confident communication, and cross-cultural understanding.

72. Our IHLs recognise the importance of this.

73. For example, graduates from Temasek Polytechnic receive an official skills transcript alongside their academic transcript, which showcases both technical skills as well as soft skills and co-curricular achievements.

74. SMU's Co-Curricular Transcript similarly showcases soft skills that graduates have developed through their co-curricular participation.

75. We must also empower our students long after they have graduated. As skills demands continue to evolve, workers will need to upskill all through their careers.

76. Last month, we announced moves to support Singaporeans in the workplace to do this, including a simple self-diagnostic tool developed by SkillsFuture Singapore and the Singapore Institute of Technology to help individuals understand their AI readiness levels, so that they can be directed to courses pegged at their level of understanding.

77. And from the second half of this year, all our IHLs will offer selected AI-related courses at significant discounts for their alumni, for a period of one year.

78. These moves will help our workers refresh their skills in an AI-pervasive economy.

79. To position Singapore at the forefront of AI in Higher Education, we will establish a new Committee for AI in Higher Education to spearhead the next phase of transformation across our IHLs.

80. Today, our Autonomous Universities, Polytechnics, and ITE already have various workgroups that bring faculty members together – not just within their institution, but across institutions – to share best practices in AI adoption, and we have seen how they helped to accelerate learning and innovation across institutions.

81. However, as AI continues to evolve rapidly and reshape the education landscape, more needs to be done at the system-level, to identify opportunities and address challenges collectively across the whole higher learning landscape.

82. Hence, we are doubling down our efforts by establishing this new Committee for AI in Higher Education, which I will chair.

83. This Committee will also include Senior Minister of State for Education Dr Janil Puthucheary, as well as the Presidents of our Autonomous Universities and the Principals and CEOs of all our Polytechnics and ITE,

84. To drive a more strategic and coordinated approach towards AI in Higher Education, this Committee will serve two key purposes.

85. First, to provide strategic direction, steer and oversight, so as to strengthen coordination and collaboration across our IHLs on key AI priorities, even as each IHL continues to experiment and innovate with AI.

86. And second, to deepen sharing of best practices, experiences, and emerging developments in the use of AI in higher education among our IHLs.

87. By strengthening sharing and coordination at the leadership level, and supporting the architecture of workgroups at the technical and working level, we can build on existing efforts and move with greater purpose and ambition, not only adapting to change but seeking to shape it proactively as the future of higher learning in our campuses evolve.

88. Beyond establishing the Committee for AI in Higher Education, my colleagues at MOE will also strengthen research on how AI can advance teaching and learning at the tertiary level, by supporting inter-IHL AI projects through our Tertiary Education Research Fund.

89. This will bring together educators and researchers to explore innovative approaches to learning, generate evidence on what works, and translate these insights into teaching practice across our institutions.

90. This work within the higher education sector also supports our broader AI national efforts, which will be spearheaded by the new National AI Council, chaired by PM Lawrence Wong.

91. To sum up, Singapore's approach to AI in education must neither be alarmist nor complacent. We will continue to experiment carefully, learn quickly as a whole system, and adjust where necessary.

92. So, is "AI in higher education hype or hope" as the theme of today's conference asks? I would say that it depends on how we respond.

93. If we treat AI as a shortcut, to simply bypass thinking, we will diminish the very purpose of education. But if we treat AI as a catalyst, a tool to sharpen what truly matters in education, it can strengthen our IHLs and strengthen our people.

94. Navigating this transformation requires collective effort. Government agencies, our IHLs, industry partners, educators, and students themselves must work together to share perspectives, experiment responsibly, and share learning points from global developments. This will help ensure that Singapore remains competitive in an AI-impacted world.

95. So in that regard, I value your views and ideas during the dialogue later. Thank you.