MDDI 演讲稿 · 2024-01-29
杨莉明部长在 Explore AI 活动上的演讲
Speech by Minister Josephine Teo at Explore AI
要点
- • 新加坡 AI 治理观：「治理 ≠ 监管」——治理还包括基础设施（算力 + 工具 + 治理框架）+ 能力（产业、公司、人）+ 伙伴关系（包容、共学、跨政府）。
- • Trailblazers 100/100 目标（100 个项目 100 天）已轻松完成；6 个月内 46 个项目已造出 MVP。
- • 案例：南洋理工学院（NYP）用生成式 AI 帮助讲师快速跟上 IT 领域变化、设计课程；Doctor Anywhere 把转诊到 24×7「Doctor Anytime」。
- • Trailblazers 2.0 即将启动——规模至少是当前的 1.5 倍。当前项目仅时间节省一项——预估每年价值 1000 万新元以上。
完整译文（中文）
MDDI 英文原文译文 · 翻译日期：2026-05-03
本文已从早期版本的网站迁移过来——格式可能有不一致之处。
1. 感谢 Google 再次接待我。我看到台下许多同仁——很高兴回来。两周前我在瑞士达沃斯——见到许多 Google 同仁。政府与 Google 的所有活动都指向一个好目的——我们都关心 AI 能做什么。不只是为公司或行业——也许听起来宏大或过分雄心——我们也关心——我们能做什么帮助这个世界。
2. 在达沃斯——AI 不出意外地高居议程。围绕 AI 的一个问题——是「我们到底需要什么才能把它治理好」。与新加坡生态打交道的人会注意到——在治理上——我们不会只把它当作监管。监管当然是良好治理的一部分——但在数字、特别是 AI 上——我们必须确保有支撑活动的良好基础设施。
3. 因此——当我们提出更新版《国家 AI 战略》时——我们认为——产业、研究、政府这些「活动驱动力」——只能在「基础设施」语境的支撑下——才能走得很远。这种基础设施包括——算力与工具的可获得性、各项治理框架的就位。所以——基础设施（以及与之配套的「公用事业」）——是绝对必要的。
4. 良好治理的另一个非常重要的面向——是构建能力。这里说的是——产业、公司、人三层面的能力。如果没有人帮你把它变为现实——几乎不可能让 AI 在多个行业中被部署。
5. 伙伴关系——又一个良好治理的重要面向。我们「为包容而合作」。包容意味着——人们不仅能接触到工具——还被给予机会发展「能用好这些工具」的技能。也意味着——「为彼此学习而合作」。AI Verify 就是一个很好的例子。要把实操级测试工具放就位——我们需要搞清楚——当企业部署 AI 系统并使用这些测试工具时——他们学到了什么？我们如何改进这些工具的设计与部署？当然——伙伴关系也延伸到与其他政府的合作。这就是我们本周晚些时候要尝试的事——与 ASEAN 大家庭的同行、并在全球层面。
6. 我今天分享所有这些——是因为今天「Trailblazers」的重大里程碑——确实勾选了我所描述的所有方框——良好治理的力量、把基础设施/公用事业/工具汇集起来、构建企业与人的能力、尝试理解如何在落地 AI 时既包容又负责任、以及「彼此学习」的理念。所以——我真的想感谢 Google——让这件事发生——并如此热情地与我们合作。
7. 与所有政府努力一样——我们总会问自己——这些项目里最大的回报是什么？我也对 Trailblazers 这样问过。一开始——我们希望看到「100 天 100 个项目」——这件事很容易达成——拿到 100 个项目没问题。下一个问题——这些项目能不能成为「最小可行产品」（MVP）？答案是——能。事实上——仅 6 个月里——其中 46 个就已经做出了 MVP。
8. 我走访的几支团队中——遇到一支非常有趣的——南洋理工学院（NYP）。我曾在 NYP 董事会任职——所以这种「生成式 AI 可能帮助讲师设计回应产业需求的课程，以学生为受益者」的想法让我很感兴趣。在 IHL 工作过的人都知道——设计一门课程、整合一份课纲——通常要花几个月。但在 IT 领域——6 个月里很多事就变了。因此——这种新工具有可能让讲师——快速感知所授领域里发生的事——并把课程设计与所需材料整合起来——以达成学习成果。
9. 另一个引起我注意的项目——是 Doctor Anywhere。作为一名病人——你可能需要看专科医生——但首先要找到合适的专科医生——这通常依赖于「代理人」——他会基于你的病史与保险计划给建议。因此——Doctor Anywhere 在 Trailblazers 项目里开始尝试——能否克服「病人在专科转诊上的困难」。借助生成式 AI 工具——Doctor Anywhere 现在能成为「Doctor Anytime」——24×7 在病人需要时提供。
10. 借助生成式 AI——NYP 能从「节省时间 + 提升课程时效」中受益；Doctor Anywhere 能改善客户体验——并对感到焦虑的病人有所帮助。这些事——我们如何量化收益？如何衡量它们的价值与影响？说实话并不容易——病人满意度的改善、毕业一届届学生带着更相关技能进入职场所带来的好处——并不易测量。但我们仍必须尝试。我们仅基于 Trailblazers 中毕业的产业项目——做了一个保守评估——发现仅在「时间节省与效率提升」上——估值每年超过 1000 万新元。
11. 现在——我很高兴地说——在新加坡经济发展局（EDB）同仁加快评估的支持下——我们能推进「Trailblazers 2.0」——它至少是当前项目的 1.5 倍规模。基于此——我们能保守地说——潜在的时间节省一定会超过当前的运行。但「美元价值」之外——更重要的是——我们正在新加坡内部——快速构建起这种「想象力与兴趣」——去用生成式 AI 工具——显著改进与改造企业的工作方式——并通过「让员工做得更好」来提升客户满意度。我把这一切——视为对我们「在这里成长 AI 生态」非常有益。
12. 在此——再次感谢邀请。我祝贺所有 Trailblazers 毕业生——以及「Transformation Award」与「Innovation Award」的获奖者。
13. 谢谢。
演讲 PDF 版本
英文原文
MDDI 官网原始记录 · 抓取日期：2026-05-02
This article has been migrated from an earlier version of the site and may display formatting inconsistencies.
1. Thank you Google for hosting me again. I see many colleagues in the audience, and I am happy to be back. Two weeks ago, I was in Davos, Switzerland where I met many colleagues from Google. The activities that the Government has with Google have are all towards a good cause - we are all interested in seeing what AI can do. Not just for companies or industries - but it may sound lofty and overly ambitious - we are interested to see what we can do to help the world.
2. In Davos, unsurprisingly, AI was very high on the agenda. One of the questions around AI is about what we really need to govern it well. For those who have interacted with the Singaporean ecosystem, they notice that in terms of governance, we tend not to just think of it as just regulations. Regulations are certainly part of good governance, but in digital, and certainly in AI, we have to make sure there is good infrastructure to support the activities.
3. So, when we put forward our refreshed National AI Strategy, we felt that the activity drivers of industry, research, and government can only push far ahead if supported within the context of that infrastructure. Such infrastructure would include having the compute and tools available, and the various governance frameworks in place. So, the infrastructure and utilities, which go along with that infrastructure, are absolutely essential.
4. Another very important aspect of good governance is in building capabilities. By this, we are talking about capabilities within industries, companies, and people. It is next to impossible to have AI deployed across many different sectors if you do not have the people to help make that a reality.
5. Partnerships is yet another important aspect of good governance. We partner for inclusion. Inclusion means making sure that people not only have access to the tools, but they are provided with opportunities to grow the skills that will enable them to use these tools well. It also means partnering to learn together. AI Verify is a very good example. If we want to be able to put in place practical testing tools, we need to figure out - when companies deploy the AI systems and use these testing tools, what are they learning and how can we improve the way in which these tools are designed and deployed? Of course, partnership also extends to partnering with other governments. And that is what we are attempting to do later this week with our colleagues in the ASEAN family, but also at the global level.
6. The reason I am sharing all this today, is that today's major milestone for Trailblazers actually ticks all of the boxes that I described - the power of good governance; bringing together the infrastructure, utilities, and tools; building enterprise and people capabilities; trying to understand how you can be inclusive and yet, at the same time, responsible in the way we implement AI; the idea of us all learning together. So, I really want to thank Google for making this happen and for partnering with us so enthusiastically.
7. As with all government efforts, we always ask ourselves - what has been the biggest payoff in these projects? And I attempted to do the same for Trailblazers. When we started, we wanted to be able to see 100 projects within 100 days. That was achieved easily as we had no problem getting 100 projects. The next question you ask – are the projects able to become Minimum Viable Products (MVP)? And the answer is, yes. In fact, in the short period of just about six months, 46 of them have in fact built an MVPs.
8. Among the several teams that I visited, I came across this very interesting one by Nanyang Polytechnic (NYP). I used to serve on the Board of NYP, so I was quite intrigued by the idea that generative AI potentially can help instructors design curriculum courses that respond to the industry needs for their students’ benefit. For those of us who have been involved with institutes of higher learning, we know that designing a course and pulling together a curriculum usually takes months; but in the field of IT, many changes can take place in six months. Hence, this new tool can potentially enable the instructors to quickly get a sense of what is happening in the field that they are required to instruct in, and then to put together a course design together with the course materials that are needed to achieve the learning outcomes.
9. The other project that caught my attention was by Doctor Anywhere. As a patient, you may need to see a specialist, but first you have to identify the right specialist; for that to happen, it would usually often depend on an agent who would advise you depending on your own medical history and insurance programme. Hence, Doctor Anywhere started working on the Trailblazers project to see whether they could overcome difficulties that patients have in accessing specialist referral services. With the use of a generative AI tool, Doctor Anywhere can now be Doctor Anytime, that is to be offered anytime, 24/7, when the patient needs it.
10. With Generative AI, NYP can potentially benefit from time savings and enhanced currency in their curriculum to the students. With Doctor Anywhere, they experienced enhancements to customer experience and being helpful to patients who feel a sense of anxiety. For these things, how do we quantify the benefits? How do we measure their value and their impact? The truth is, it is not so easy. It is not so easy to measure the value of improved patient satisfaction, and it is not so easy to quantify how much more beneficial it is for every graduating cohort of students to come out into the workforce with more relevant skills than they did before. But still, we must try. We did a modest assessment based on just the graduating industry projects from Trailblazers, and found that in terms of time savings and greater efficiency, the estimate is in excess of $10 million savings annually.
11. Now, I am very happy to say that with the help and support of our colleagues in the EDB that have fast tracked their assessment, we can move ahead with Trailblazer 2.0, which is at least 1.5 times bigger than the current programme. With that, we could very modestly say that the potential time savings will definitely exceed the current run. But beyond the dollar value is the idea that we are building up very quickly this capacity within Singapore, and the imagination and interest to use the tools of Generative AI to significantly improve and transform the way business is done and enhance customer satisfaction by helping employees do a better job. So, I see all of this as really being very beneficial for how we are trying to grow the AI ecosystem here.
12. On that note, thank you once again for inviting me. I want to congratulate all of the Trailblazers graduates, and the ‘Transformation Award’ winners and ‘Innovation Award’ winners.
13. Thank you.
PDF Version of the Speech