MDDI 演讲稿 · 2023-06-07
杨莉明部长在亚洲科技 x 人工智能大会上的演讲
Speech by Minister Josephine Teo at the Asia Tech x Artificial Intelligence Conference
要点
- • Josephine 用「AI 与电力」做类比：电本身不直接给我们好处——是「驱动器械」时才出现便利与生产力。AI 给的是另一种「电」——以远低成本的「类人智能」。
- • 新加坡公共部门已有 AI 用例：自 2015 年用 ML 处理市民反馈；用 AI 检测货柜异常（去年近 1000 万集装箱 + 5000 万航空件）；钓鱼侦测工具每天扫 12 万个网站。
- • 更大挑战的应用：医疗（2030 年新加坡 1/4 人口将 65 岁+，AI 助力诊断与精准医疗，10 万新加坡人 DNA 测序）、可持续（建筑占全球能耗 40%）、教育（让老师从批改作业等中解放，给弱势学生定制化学习计划）。
- • AI 治理：开源 AI Verify、设立 AI Verify Foundation——成员含 IBM、Google、微软、Red Hat、Salesforce、Aicadium；今天 IMDA 与 Aicadium 联合发布生成式 AI 关切的讨论论文。
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。
各位阁下、
产业伙伴、
朋友们：
早安。
对 AI 的兴奋
自我们去年见面以来——AI 已经取得显著突破。如果一年前问人们「ChatGPT 是什么」——他们可能猜成一款约会 App！生成式 AI 与它的潜能——乘以变化的规模与速度——令人惊叹。
上个月——我在旧金山参观了 Alamo Square 的「Hacker House」——那个区域如今被称为「Cerebral Valley」（大脑谷）——反映 AI 活动的密度。那里有初创创办人、研究者与投资者。周五发表的学术论文——周末就被做出原型。
让我高兴的是——湾区也有满怀热情的新加坡人——他们身处这场科技革命的最前线。每个人都被生成式 AI 的变革潜能唤醒——所幸——并非所有人都在追炒作。
深思的领导者——仍在探索替代性的 AI 架构与方法。他们也提醒我——AI 的危险。
这一切对新加坡意味着什么？我们从这里去往哪里？在与非科技社群的讨论中——我发现把「AI 与电」做类比有用。电本身给我们的好处不大——但当用来驱动器械与设备时——我们就获得便利、生产力与能力。
AI 交付的——是另一种「力」——类人智能的力——可能是非常高级的——而成本大幅降低。这对人力资本「事关重大」的新加坡——尤其有价值。
正如副总理王瑞杰昨晚所说——若我们能把这种力驾驭为「Augmented Intelligence」（增强智能）——支持人而不是替代人——我们的公民将大受其惠。但与此同时——就像不当使用器械会触电——不当使用 AI 也可能造成巨大伤害。
因此——「护栏」是必要的——指引人们负责任地使用——并让 AI 产品「设计上对所有人都安全」。
新加坡为「AI 服务公共利益」所做的努力
这就是为什么——在新加坡——我们相信必须竭尽所能——把 AI 用在「公共利益」上。所幸——我们不是从零开始。我们公共部门里的创新者——已经在构建 AI 产品——以改善治理与服务交付。
请允许我分享几个例子——
自 2015 年——我们用机器学习处理市民关于「邻里所需修缮」的反馈——比如有人看到游乐场坏了的秋千、坏了的路灯。市民因此能更快得到「合适的人」给出的回应。
去年——近 1000 万个集装箱与货柜跨过我们国境。另有 5000 万件航空货物。我们需要清关既快又安全。
用 AI 检测货柜扫描图像中的异常——这「双重目标」更易达成。
更近一些——我们也开始用「AI 加持」的图像与文本对比——侦测诈骗。我们的钓鱼侦测工具——每日扫 12 万个网站——下架用于欺诈的仿冒站点。
若没有 AI 这件「武器」——执法人员几乎无法把精力聚焦在「诈骗预防」与「为受害者追回资产」上。
这只是几个「AI 服务公共利益」的明显用法。
应对当代挑战的其他用例
AI 也能帮助应对当代更大的挑战。
到 2030 年——新加坡每 4 人中将有 1 人超过 65 岁。对一个老龄化、慢性病负担上升的人口——AI 是新加坡改进临床诊断与患者福祉的关键工具。它也能为家庭与医院降低成本。毕竟——我们花在住院上的费用——一半发生在生命的最后 3 个月。
还不止。在癌症预防性照护中——我们的 AI 平台已经能基于患者状况数据——开出最优药物剂量。我们正在为 10 万名新加坡人测绘 DNA、做全基因组测序。AI 能用来更好地理解「基因—临床数据」的关联——以推行精准医疗。
另一项重大挑战是可持续。新加坡承诺到本世纪中叶达成净零。新加坡也是「花园中的城市」。但跟所有城市一样——建筑是问题。全球而言——建筑占能耗的 40%。
AI 能在多方面提升能效——从建筑设计与仿真，到能源监测与优化。在更广层面——已有 AI 方案与应用——能实时平衡电力供需——优化能源负载部署与储能。当然——讽刺的是——AI 自己也必须变得更节能——这是我们的集体挑战！
在医疗与可持续之外——我们看到 AI 在教育领域的机会。如果老师能从批改作业、管理学生记录中解放——会怎样？给来自弱势背景的孩子——个性化学习计划与辅导支持——又会怎样？想象一下——这能如何缓和不平等——把下一代的所有人都拉起来！
增强新加坡 AI 生态
这些机会——是新加坡向所有人发出号召——让大家「为公共利益借力 AI」的原因。请允许我分享一些「我们如何前进」的想法。
第一——我们相信——政府能在「广泛的 AI 实验、采用与规模化」上领跑。甚至可以说——政府独特地占据着——既能审视问题规模、又拥有让这种实验产生影响的资源。
第二——我们相信——AI 熟练度可以通过「深度技能发展 + 自下而上的学习」组合来构建。这些是重要的根基——好比为「百花齐放」丰沃土壤。
第三——我们相信——为公共利益负责任地部署 AI。我们鼓励实验与采用——但我们也将努力——把社会从最严重的 AI 风险中保护出来。
要在我提到的所有方面取得进展——政府无法独自完成。私营部门与研究生态有丰富的专长——他们能够也必须被鼓励——有意义地参与「为公共利益推进 AI」。
生态中的创新与机会
这件事正在发生。在这里的公司——正在积极发挥作用——成长我们的生态。SAP、CISCO、Sea、Grab 已在新加坡扎根 AI 实验室——创造了数千个优质岗位——并让新加坡人在各自领域成为科技领导者。
甚至有这样一项合作——新加坡国家级 AI 项目「AI Singapore」与「世界自然基金会新加坡」（WWF Singapore）合作——用 AI 加持的工具——对抗非法野生动物贸易。
全面提升熟练度
一个挑战是——人才的广度与深度跟不上需求。我们将继续扩大培训容量。但即便培训课程在更新——也可能很快过时。
因此——AI 的技能习得「不必非通过正规教育」——这是一个加分项。事实上——许多人通过自主实验与「做中学」获得了成功。
因此——我们鼓励雇主——以更宽的视野识别人才。在 AI 上——年龄与学历都不是障碍。如果我们拥抱这一点——我们就更接近把 AI 实现为「Augmented Intelligence」（增强智能）——拓宽通往成功路径的 AI——一定会更具包容、并是「为善之力」。
新加坡的 AI 治理路线图
现在我谈一谈「负责任 AI 使用」的话题。对 AI 安全的强烈渴望——并不意味着「关上创新与采用的吊桥」。正如交通运输领域——刹车、限速、安全带、安全气囊——能在道路使用者中培育信心。
在 AI 领域——安全机制与共享标准——同样会培育对 AI 使用的信心。但有效的安全防护——需要时间与好的研究去发现。
与此同时——我们已开发了多个框架——以推动问责与信任。你们当中可能有人记得——新加坡 2019 年发布了《Model AI Governance Framework》——它当时并仍是亚洲首个。
去年——我们引入了 AI Verify——一个治理测试框架与工具包。这个「最小可行产品」（MVP）至今已吸引 50 多家公司的兴趣。你们可以参观由 IBM、UBS、新加坡航空（Singapore Airlines）等公司参与的展示。
通过伙伴关系强化 AI 治理、推动安全与负责任使用
为了支持「AI 服务公共利益」——新加坡决定把 AI Verify 开源——并启动「AI Verify Foundation」。我们相信——系统开发者、方案提供者、研究社群——都能使用 AI Verify、并为它做贡献。「众包」他们的专长——也将促进新的、更好的测试工具的成长。
「AI Verify Foundation」将设定 AI Verify 的战略方向与发展路线图。现有成员——包括 IBM、Google、微软、Red Hat、Salesforce 与 Aicadium。他们将作为我们的「大使」——把社群聚到一起——开发更好的框架、标准与最佳实践。我们也欢迎更多有兴趣的伙伴加入。
我们还将增强我们的治理工具集。今天——IMDA 与 Aicadium 联合发布一份讨论论文——突出生成式 AI 中的关键关切。我们希望它能激发许多对话——并构建对所需护栏的认知。
第四部分：构建一个 AI 就绪的新加坡
总结——我们相信——AI 是自互联网与移动以来的下一个「大转移」。在围绕 AI 发展非常真实的恐惧与关切之中——我们必须主动把 AI 引向有益的用法、远离不良的用法。这是新加坡看待 AI 的核心。
这样做——我们希望让新加坡成为「人才、想法、实验」的一个出色地点。我们也希望成为一个全球网络中的有活力节点——在这个网络中——力量被引向「可信的 AI 系统与负责任使用」。在那里——「AI 服务公共利益」会真正鲜活起来。
致今天在场的每一位——我们的故事才刚开始——我们邀请你成为其中一员！
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.
Excellencies,
Industry Partners,
Friends.
A very good morning to you
Excitement about AI
Since last year when we met, there have been significant breakthroughs in AI. If asked a year ago what ChatGPT was, people may have guessed it was a dating app! Generative AI and its potential, multiplied by the scale and speed of change, have been nothing short of stunning.
Last month, in San Francisco, I visited a Hacker House at Alamo Square. This area is now referred to as “Cerebral Valley”, reflecting the concentration of AI activity. There were startup founders, along with researchers and investors. Academic papers published on a Friday were being prototyped over the weekend.
Happily for me, there were also passionate Singaporeans in the Bay Area, who are themselves at the forefront of this tech revolution. Everyone has come alive to the transformative potential of generative AI. Thankfully, not all were chasing the hype.
Thoughtful leaders are still exploring alternative AI architectures and approaches. They cautioned me about AI’s dangers.
What does all of this mean for Singapore and where do we go from here? In discussions with non-tech communities, I have found it useful to draw the comparison of AI and electricity . Electricity itself brings us little benefit. But when used to power appliances and equipment, we get so much more in convenience, productivity and capabilities.
What AI delivers is a different kind of power. It is the power of human-like intelligence, potentially a very high form of it, at far reduced cost. This is especially valuable for Singapore where human capital makes all the difference.
If as Deputy Prime Minister Heng Swee Keat put it last night, we can harness this power to make it “Augmented Intelligence”, to support rather than replace people, our citizens will have a lot to gain. At the same time, just as the improper use of appliances and equipment can cause electrocution, inappropriate use of AI can also do great harm.
Guardrails are therefore necessary to guide people to use it responsibly, and for AI products to be “safe for all of us” by design.
Singapore’s Efforts in Establishing AI for Public Good
This is why in Singapore, we believe we must do all we can to harness AI for the Public Good. Fortunately, we are not starting from scratch. Innovators in our public service are already building AI products to improve governance and service delivery.
Allow me to share a few examples:
Since 2015, we have used machine learning to process citizen feedback on repairs needed in their neighbourhoods, such as when someone sees a broken swing at the playground, or a lamppost that’s not working. Citizens benefit from faster responses, by the right people.
Last year, close to 10 million containers and consignments crossed our shores. Another 50 million parcels came through air shipment. We need cargo clearance to be fast but also safe.
These twin objects are better achieved by the use of AI to detect anomalies in scanned images of these shipments.
More recently, we have started using AI-enabled image and text comparisons to detect scams. Our phishing detection tool combs through 120,000 websites daily to take down spoof sites used for fraudulent purposes.
Without such AI in its arsenal, law enforcement agents will hardly have the capacity to focus on scam prevention or to recover the assets of victims.
These are a few of the obvious uses of AI for the Public Good.
Other use cases to tackle challenges of our time
AI can also help address the bigger challenges of our time.
By 2030, 1 in 4 Singaporeans are expected to be over 65 years old. For an ageing population with growing chronic disease burden, AI will be a vital tool for Singapore to improve clinical diagnosis and patient well-being. It can also reduce costs for families as well as hospitals. Afterall, half of what we spend on hospitalisation is incurred in the last three months of our lives.
There’s more. In cancer preventive care, our AI platform can already prescribe the optimal drug doses based on data about a patient’s condition. We are mapping the DNA of 100,000 Singaporeans and sequencing whole genomes. AI can be used to better understand genomic-clinical data linkages, for the practice of precision medicine.
Another big challenge is sustainability. Singapore has committed to becoming net-zero by the mid-century. Singapore is also a city in a garden. But as with all cities, buildings are an issue. Globally, they contribute 40% to energy consumption.
There are many ways AI can improve their energy efficiency, from building design and simulation to energy monitoring and optimisation. At the broader level, there are already AI solutions and applications that balance electricity supply and demand in real time. This can optimise energy load deployment and storage. The irony of course, is that AI itself must become more energy efficient. This is a collective challenge for us all!
Beyond healthcare and sustainability, we see opportunities for AI in education. What if teachers’ time can be freed up from tasks like grading assignments and managing student records? How about individualised learning plans and tutoring support for children from disadvantaged backgrounds? Imagine how this can moderate inequality and uplift everyone in our next generation!
Enhancing Singapore’s AI ecosystem
These opportunities are why Singapore is making a rallying call for all of us to harness AI for the Public Good. Allow me to share some thoughts about how we may proceed.
First, we believe the government can lead the way in widespread AI experimentation and adoption, as well as scaling. One could even argue that the government is uniquely positioned to have oversight on both the scale of the problems, and the resources to make such experimentation impactful.
Second, we believe AI proficiency can be built through a combination of deep skills development as well as ground-up learning. These are important foundations, akin to enriching the soil conditions for a thousand flowers to bloom.
Third, we believe in responsible AI deployment for Public Good. We will encourage experimentation and adoption. But we will also strive to shield society from the most serious AI risks.
To make progress on all that I have just mentioned, the government cannot do it alone. The private sector and the research ecosystem have rich expertise. They can and must be encouraged to participate meaningfully to advance AI for the Public Good.
Innovations and opportunities within the ecosystem
This is already happening. Companies here are playing an active role in growing our ecosystem. SAP, CISCO, Sea and Grab have anchored AI labs in Singapore. They have created thousands of good jobs and enabled Singaporeans to be tech leaders in their fields.
There is even a partnership between AI Singapore, our national AI programme, and the World Wide Fund for Nature Singapore, to use AI-enabled tools to combat illegal wildlife trade.
Uplifting proficiencies across the board
One challenge is that the breadth and depth of talent cannot keep up with demand. We will continue to grow our training capacity. But even as training curriculum is updated, it risks becoming outdated all too soon.
It is a plus therefore that the skills acquisition in AI does not always need to come through formal education. In fact, many people have succeeded through independent experimentation and learning-by-doing.
We therefore encourage employers to take this wider aperture for spotting talent. With AI, neither age nor academic qualifications are barriers. If we embrace this, we come closer to realising AI as “Augmented Intelligence”. AI that broadens the pathways for success must surely be more inclusive and a force for good.
Singapore’s AI Governance Roadmap
Let me now turn to the topic of responsible AI use. A strong desire for AI safety need not mean pulling the drawbridge to innovation and adoption. As we have seen in transportation, brakes, speed limits, seatbelts and airbags can promote confidence among road users.
In AI, safety mechanisms and shared standards will equally instil confidence in its use. But effective safeguards will take time and good research to discover.
Meanwhile, we have developed several frameworks to promote accountability and trust. Some of you may recall that Singapore published our Model AI Governance Framework in 2019. It was and remains the first of its kind in Asia.
Last year, we introduced AI Verify, a Governance Testing Framework and Toolkit. This minimum viable product has since attracted interest from over 50 companies. You can check out the showcase featuring companies like IBM, UBS and Singapore Airlines.
Partnerships to strengthen AI governance and promote safe and responsible use
In support of AI for the Public Good, Singapore has decided to open source AI Verify and launch the AI Verify Foundation. We believe that system developers, solution providers and the research community can all use and contribute to AI Verify. Crowding-in their expertise will also promote the growth of new and better testing tools.
The Foundation will set the strategic directions and development roadmap of AI Verify. Current members include IBM, Google, Microsoft, Red Hat, Salesforce and Aicadium. They will be our ambassadors to gather the community to develop better frameworks, standards and best practices. And we welcome more interested partners to join in.
We will also enhance our suite of governance tools. IMDA and Aicadium are releasing a joint discussion paper today, highlighting key areas of concerns in generative AI. We hope it will spark many conversations and build awareness on the guardrails needed.
Part IV: Building an AI-ready Singapore
To conclude, we believe AI is the next big shift since the internet and mobile. Amidst very real fears and concerns about its development, we will need to actively steer AI towards beneficial uses and away from bad ones. This is core to how Singapore thinks about AI.
In doing so, we hope to make Singapore an outstanding place for talent, ideas and experimentation. We also aim to be a vibrant node within a global network where efforts are directed towards trusted AI systems and responsible use. This will be where AI for the Public Good will truly come alive.
So to everyone here today, our story is just beginning and we invite you to be a part of it!
Thank you!
PDF version of the speech