MDDI 演讲稿 · 2026-01-24
杨莉明部长在新加坡 AI 研究周晚宴上的开幕致辞
Opening Remarks by Minister Josephine Teo at Singapore AI Research Week Gala Dinner
要点
- • 新加坡承诺 5 年（2025–2030）投入 10 亿+新元支持公共 AI 研究，分三块：基础 AI、应用 AI、人才培养。
- • 「资源高效 AI」是基础 AI 的一项重点——新加坡数据中心容量已为 50% 增长留出空间，但要在 2050 年净零的承诺下推进，所以从芯片架构到模型/应用设计全栈寻找效率。类比：水资源转化为「NEWater」的成功经验。
- • 在重点方向设立「AI 研究卓越中心」（AI Research Centres of Excellence），与本地及国际生态深度协作，研究成果开放共享。
- • 应用 AI 的标杆例子在「身边」——樟宜机场用 AI 优化中转、安检、行李、清洁机器人；与 SUTD、SMART、A*STAR 协作。
- • 本地资源已铺好：SEA-LION（AI Singapore）与 MERaLiON（A*STAR）开源语言模型代表东南亚 1000+ 语言；100 Experiments 计划支持上百家组织。
- • 人才方面：要培养「双语研究人才」（AI + 领域专长）；继续支持 IOAI 训练队（2024、2025 已夺金）；扩大奖学金与研究机会；继续 AI Visiting Professorship——目前已颁出 8 位。
完整译文（中文）
MDDI 英文原文译文 · 翻译日期：2026-05-02
尊敬的同事与朋友们：
晚上好，感谢大家的莅临。
我想感谢出席的内阁同事——尤其是负责交通事务的代理部长萧振祥（Jeffrey Siow）。
在这一重责之外，他还被总理委以「经济战略检讨委员会」联合主席之责。
在这一身份上，Jeff 也大力支持加速企业与劳动力 AI 采用的方案。
他稍后还有发言机会，今晚我先说一句——「Jeff，谢谢你」。
我也要感谢国家研究基金会（NRF）主席王瑞杰先生。
今晚我大量篇幅会谈到《国家 AI 研发计划》。
这份计划是 NRF 与我所在部门的真切合作；
建立在我们共同的信念之上——卓越的研发本身就具有内在价值，并惠及社会。
我们感谢瑞杰一直鼓励我们在梦想上更大胆。
在他的支持下，这份计划将以可触及的方式，把我们「AI 服务公共利益——为新加坡，也为世界」的愿景落地。
在更多介绍这份计划之前，请允许我先分享我在达沃斯参加世界经济论坛的一周。
并不令人意外，地缘政治成了主角。
幸运的是，对技术发展以及新加坡做法的兴趣依然强烈。
不到 48 小时前，尚达曼总统与我共同会见了 Yoshua Bengio。
我们讨论了「负责任的 AI」意味着什么，以及如何实现它。
也许有些人知道——Yoshua 与其他全球专家共同主持了 2025 年「新加坡 AI 大会」（SCAI）：「AI 安全国际科学交流」。
它与首届「新加坡 AI 研究周」同期举办；
并促成了《新加坡 AI 安全研究优先事项全球共识》。
我们在评估第二届交流的必要性，以及怎样才能让它有用。
再前一天，我和 Demis Hassabis 见了面。
一年多前，Demis 访问新加坡时见过黄循财总理。
去年，Google DeepMind 决定在新加坡扩展业务。
他们已经开始招人，团队正在迅速搭起来。
我的团队正与他们一起寻找有影响力的合作。
我也与 Yann LeCun 见了面——他离开 Meta 后已经创办了一家新公司。
Yann 解释了他的路线图——以及新加坡在其中的位置。
等他准备好了，他会自己讲——这件事让我非常兴奋。
对我来说最超现实的一刻——是与 2025 年诺贝尔物理学奖得主 John Martinis 同台。
作为台上唯一的政府代表，我尽力向满场观众解释——量子技术为什么对新加坡这样的国家重要，以及本地不断壮大的生态。
想象我意外地听到 John 宣告——他的初创公司将与新加坡合作。
那家公司是 Qolab——使命是开发大型超导量子计算机。
John 说他希望借助新加坡的半导体制造专长，来制造 Qolab 量子计算机的特定组件。
我把这件事讲出来，是为了让大家了解——新加坡深科技生态正在变得越来越有活力。
当然，我还能再分享许多镜头来呈现更完整的画面。
为节省时间，我邀请各位去那个房间后方的展区，了解我们部分正在推进的举措。
正是在这样的背景下，我们整合出了《国家 AI 研发计划》。
为了实施这份计划，新加坡将在 5 年（2025–2030）期间投入超过 10 亿新元，资助公共 AI 研究。
我们将分三大块：基础 AI、应用 AI、人才培养。
我用一点篇幅说明每一块的思考。
尽管 AI 取得了非凡突破，但其发展仍有根本性的局限。
比如，AI 训练与推理仍然极其耗资源——它对能与水的占用不能被忽视。
新加坡已是本地区数据中心容量密度最高的之一。
我们为最高 50% 的增长留出了空间。
考虑到 2050 年净零的承诺，这种扩张必须被审慎管理。
这就是「资源高效 AI」研究能与《绿色数据中心路线图》互补、并产生战略价值的地方。
我们要在整个技术栈上找到提升效率的新途径——从芯片架构，到模型与应用设计。
若能取得成功，不只惠及我们自己，也惠及面临类似约束的其他地方。
这种愿景从我们如何应对水资源约束（这是新加坡的存亡级风险）中获得启发。
几十年的研究帮助我们建立水的韧性——包括把废水回收为我们所谓的「NEWater」。
今天，新加坡被国际公认为「整合水资源管理」的样板城市，并是新兴的全球「水科技枢纽」。
本地一些公司的方案，已经走出新加坡——
比如，新加坡公司 Wateroam 开发的便携式滤水器，正在柬埔寨、印尼、尼泊尔被用于应急响应与人道救援。
除了资源效率，还有其他根本性挑战是我们希望应对的——比如「负责任的 AI」。
在我们划定的优先方向上，我们将在公共研究机构里设立「AI 研究卓越中心」。
这些中心将由研究团队组成——既包括成熟学者，也包括有潜力的新人——聚焦长期、困难的问题。
我们期望它们与本地以及国际同行积极合作。
我们也希望研究发现能开放共享，贡献于全球知识公共物。
事实上，这件事已在发生。
我们的大学在 AI 方面位列全球前茅。
在 NeurIPS、ICLR、AAAI（AAAI 本周正在新加坡举办）等主要 AI 会议上的论文份额可圈可点。
他们也与新加坡多家科技公司合作——包括 Google、IBM、英伟达、微软亚洲研究院。
我相信这些 AI 研究卓越中心，会让我们的生态在广度与深度上更上层楼。
与此同时，我们将以应用 AI 研究来与基础 AI 研究互补。
今晚我们所在的樟宜机场「Jewel」（星耀樟宜）就是一个相关的例子。
这座造型醒目的玻璃穹顶——内部容纳了热闹的生活方式枢纽与世界最高的室内瀑布——其实是「不得不」催生出来的。
T1（樟宜故事的起点）已经太繁忙了。
但与其在巨型停车场上方简单加层，Jewel 被想象为一个本身就是世界级的目的地，并融入了许多节能特性。
樟宜机场的创新还不止于此。它现在用 AI 来优化运营——
从飞机周转、到安检；
到自动化行李处理；
再到机器人巡查与清洁。
这意味着扩大内部产品与工程团队，并借助合作伙伴共同打造新方案——比如：
新加坡科技设计大学（SUTD）；
新加坡—麻省理工学院联盟（SMART）；以及
新加坡科技研究局（A*STAR）。
类似的努力，也存在于新加坡领先企业建立的 60 多个「卓越中心」里——把 AI 方案应用到日常问题。
与此同时，制造、医疗、可持续等其他领域的科学家也在增强自身能力，去开发并部署产业应用。
GovTech、内政科技局（HTX）这样的政府机构也在自我挑战——做 AI 使用上的领先者，而非落后者。
为进一步加强应用 AI 研究能力——
更新版的《国家 AI 研发计划》将培育「双语研究人才」（在 AI 与领域专长两者上都精熟）。
我们也希望构建核心 AI 工程能力——把理论转化为系统与应用。
我们有不错的基础。
过去十年里，国家级项目「AI Singapore」已通过其「100 Experiments」计划，帮助数百家机构使用 AI。
A*STAR 的「制造业行业级 AI 卓越中心」与跨国公司及本地企业合作，做了多个行业级用例——
如质量保证、预测性维护、产品设计、工业自动化。
AI Singapore 与 A*STAR 还分别开发了 SEA-LION 与 MERaLiON 两个开源语言模型家族。
这些模型代表了东南亚的语言与文化——东南亚有 1000 多种语言。
因此它们吸引了许多用户，也就毫不意外。
现在我谈谈我们的人才培养计划。
新加坡的 AI 科学家与研究者一直让我们引以为傲。
比如 Yi Tay 博士——他在南洋理工大学（NTU）取得本科与博士学位。今天他是 Google DeepMind 新加坡的研究科学家，共同领导 Gemini 模型的训练。
我们有一批前途看好的青年学者——比如助理教授 Pang Wei Koh、Zhi-Xuan Tan、Chun Kai Ling——他们在国内外都为新加坡 AI 生态注入活力。
我们的青少年也在 2024、2025 年「国际 AI 奥林匹克」（IOAI）上夺得金牌。
我们很高兴他们今晚都能到场。
我们将通过在各层级培育 AI 研究专长来强化人才基础。
我们将继续支持 IOAI 训练队。
我们将提升学生的奖学金与研究机会，让他们有更好的位置去争取顶尖机构的博士、博后与教职。
我们将继续推进「AI 访问教授」计划。
自该计划启动以来，已颁出 8 个教授席位给杰出的国际研究者，支持他们与本地研究者的合作。
这种紧密合作在攻克重要研究问题上正在结出果实。
其中 5 位今晚也在现场。
我前面提到的「AI 研究卓越中心」，也将是人才培养的重要平台。
并行地，我们将继续吸引顶尖 AI 初创与科技公司——
把研究与创新团队设在新加坡。
除了制度安排，导师在培养未来研究者方面扮演关键角色。
包括我们 IOAI 训练队的导师——南洋理工大学的 Alvin Chan、Liu Ziwei、Luu Anh Tuan、Wang Wenya 教授，以及 AI Singapore 的 Koo Seng Meng 先生。
我们大学的教员也提供重要指导——比如 Bryan Low、Ng See Kiong、Luke Ong 教授。他们花大量时间引导学生、为他们创造与一流 AI 学者学习的机会。
感谢各位不知疲倦地培育下一代 AI 领导者。
更新计划之后，自然的问题是——「成功是什么样子？」
我们相信好的成果，会在一个充满活力、丰富多样、彼此紧密的研究生态中涌现。
它体现在像「新加坡 AI 研究周」这样的活动里。
这才只是第二届——
但日程已经排满了 40 场活动，许多场几乎一开放就报满。
同样的能量，也能在 Lorong AI 这一协作 AI 枢纽的持续热度中感受到——去年它一共承载了 150 场活动、总到场人数超过 4000。
今晚——在所有齐聚于此的人中，我们也能感受到这种紧密的兄弟情谊。
学界、产业、政府的领导者；
我们的国际伙伴；
以及新加坡人——本地与海外的都有——他们正在 AI 领域留下印记。
似乎以一种安静且看不见的方式，我们都在共同塑造一种未来——AI 能够服务公共利益，为新加坡，也为世界。
让我们继续走下去。
英文原文
MDDI 官网原始记录 · 抓取日期：2026-05-02
Distinguished colleagues and friends,
Good evening and thank you all for being here.
I want to acknowledge the presence of my Cabinet colleagues, in particular, Acting Minister Jeffrey Siow who looks after Transport.
On top of this heavy responsibility, Jeff was tasked by our Prime Minister to co-chair one of the Economic Strategy Review Committees.
In this capacity, Jeff has strongly supported plans to accelerate AI adoption in enterprises and the workforce.
He will say more in due course, but I’d like to say tonight “thank you, Jeff.”
I’d also like to thank the Chairman of our National Research Foundation (NRF), Mr Heng Swee Keat.
Much of my speech tonight will be about the National AI R&D Plan.
This plan is truly a collaboration between the NRF and my Ministry;
Based on our shared belief that excellence in R&D has intrinsic value and benefits society.
We appreciate Swee Keat for always encouraging us to be bolder in our dreams.
With his support, the Plan will be a tangible way to realise our vision of “AI for the public good, for Singapore and the world.”
Before I say more about the Plan, allow me to share a bit about my week in Davos, attending the World Economic Forum.
Unsurprisingly, geopolitics took centre stage.
Fortunately, there was sustained interest in technology developments and Singapore’s approach.
Less than 48 hours ago, President Tharman and I met Yoshua Bengio.
We discussed what it would mean to have responsible AI, and how this could be achieved.
Some of you may know that Yoshua, alongside other global experts, led the 2025 Singapore Conference on AI (SCAI): International Scientific Exchange on AI Safety.
This was held in conjunction with the inaugural Singapore AI Research Week;
And resulted in the Singapore Consensus on Global AI Safety Research Priorities.
We are assessing the need for a second edition of the Exchange, and what would make it useful.
The day before that, I caught up with Demis Hassabis.
A little over a year ago, Demis met Prime Minister Lawrence Wong while visiting Singapore.
Last year, Google DeepMind decided to grow a presence here.
They have started hiring and the team is filling out well.
My team is working with them to identify impactful collaborations.
I also caught up with Yann LeCun, who as you know, has set up a new venture since leaving Meta.
Yann explained his roadmap and how Singapore features in it.
I will leave it to him to say more when he is ready, but it is very exciting to me.
The most surreal event for me was, however, sitting onstage next to John Martinis, a 2025 Nobel Prize Winner for Physics.
As the only government person on the panel, I tried my best to explain to the capacity crowd why quantum technology matters to a country like Singapore, and the growing ecosystem here.
Imagine my pleasant surprise when John exclaimed that his startup would be collaborating with Singapore.
That startup is Qolab, whose mission is to develop large superconducting quantum computers.
John said he hopes to leverage Singapore’s expertise in semiconductor manufacturing to build specific components of Qolab’s quantum computers.
I share this with you as context for the increasingly vibrant ecosystem for deep tech in Singapore.
There are of course many more snapshots I could give to provide a fuller picture.
In the interest of time, I invite you to visit the exhibits at the back of that room to find out more about some of our ongoing initiatives.
It is against such a backdrop that we have put together our National AI R&D Plan.
To implement this Plan, Singapore will commit more than S$1 billion to fund public research in AI over five years, from 2025 to 2030.
We will do so in three main areas: Fundamental AI, Applied AI, and Talent Development.
Let me outline our thinking in each area.
Despite extraordinary breakthroughs, there are fundamental limitations in AI development.
For example, AI training and inference remain extremely resource intensive. Their draw on energy and water cannot be ignored.
Singapore already has one of the most densest concentrations of data centre capacity in the region.
We are providing headroom for up to 50% growth.
Given our commitment to achieving net-zero emissions by 2050, this expansion must be carefully managed.
This is where research into resource-efficient AI can complement our Green Data Centre Roadmap, and yield strategic value.
We aim to find new ways to gain efficiency across the tech stack – from chip architectures to model and application design.
Achieving success will not only benefit ourselves, but others with similar constraints.
This vision is inspired by how we dealt with our water constraint – an existential risk for Singapore.
Research over many decades helped build our water resilience, including to reclaim wastewater – into what we call NEWater.
Today, Singapore is internationally recognised as a model city for integrated water management and an emerging global “hydrohub.”
Solutions by some of our local companies can now be found elsewhere.
For example, in Cambodia, Indonesia and Nepal, portable water filters developed by a Singapore company, Wateroam, are being used for emergency response and humanitarian relief.
Besides resource efficiency, there are other fundamental challenges that we hope to address, such as responsible AI.
In areas we have prioritised, we will establish AI Research Centres of Excellence, hosted in our public research institutions.
These Centres will comprise teams of researchers – established as well as upcoming individuals – focused on long-term, difficult questions.
We expect them to partner actively with others in our local ecosystem and internationally.
We also want their research discoveries to be shared openly, to contribute to the global knowledge commons.
In fact, this is already happening.
Our universities are ranked amongst the top in AI, globally.
They account for a commendable share of papers at major AI conferences, such as NeurIPS, ICLR, and AAAI, the last of which is happening this week in Singapore.
They also partner many tech companies in Singapore, including Google, IBM, Nvidia, and Microsoft Research Asia.
I am confident these AI Research Centres of Excellence will add greater breadth and depth to our ecosystem.
At the same time, we will complement research in fundamental AI challenges with applied AI research.
Jewel at Changi Airport, where we are tonight, is a relevant example.
This striking glass dome that houses a vibrant lifestyle hub and the world’s tallest indoor waterfall was in fact borne out of necessity.
Terminal One, where the Changi story first began, had gotten just too busy.
But rather than a simple extension on top of a giant car park, Jewel was imagined as a world-class destination in its own right, with many energy saving features.
Changi Airport’s innovations do not stop there. It now uses AI to optimise operations:
From aircraft turnarounds and security screening;
To automated baggage handling;
And the use of robots for inspections and cleaning.
That has meant growing in-house product and engineering teams, and creating new solutions with the help of partners, like:
The Singapore University of Technology and Design (SUTD);
The Singapore-MIT Alliance for Research and Technology (SMART); and
The Agency for Science, Technology and Research (A*STAR).
Similar efforts can be found in the more than 60 Centres of Excellence set up by leading corporations in Singapore, to apply AI-enabled solutions to everyday problems.
At the same time, scientists in other research domains – such as Manufacturing, Health, and Sustainability – are enhancing their capabilities to develop and deploy industry applications.
Government agencies, such as GovTech and the Home Team Science and Technology Agency (HTX), are equally challenging themselves to be leaders and not laggards in AI use.
To further strengthen our applied AI research capabilities:
The updated National AI R&D Plan aims to nurture “bilingual research talents” who are proficient in AI and have domain expertise.
We aim also to build core AI engineering capabilities for the translation of theory to systems and applications.
We have good foundations to work off.
Over the past decade, AI Singapore, a national programme, has helped hundreds of organisations use AI through its “100 Experiments” programme.
A*STAR’s sectoral AI Centre of Excellence for Manufacturing has worked with multinationals and local companies on industry-wide use-cases;
Such as in quality assurance, predictive maintenance, product design and industrial automation.
AI Singapore and A*STAR have also developed the SEA-LION and MERaLiON families of open-source language models, respectively.
These models are representatives of Southeast Asia’s languages and cultures – there are more than a thousand languages in Southeast Asia.
It is therefore no surprise that they have attracted many users.
Let me turn now to our plans for talent development.
Singaporean AI scientists and researchers have been doing us proud.
Take Dr Yi Tay, who obtained his Bachelor’s and PhD from NTU. Today, he is a Research Scientist at Google DeepMind Singapore, where he co-leads training of the Gemini models.
We have promising young academics like Assistant Professors Koh Pang Wei, Tan Zhi-Xuan, and Ling Chun Kai, who are adding vibrancy to our AI ecosystem, from home and abroad.
Our youths have also won gold medals at the 2024 and 2025 International Olympiad for AI.
We are delighted that they can all be here tonight.
We will strengthen our talent base through nurturing AI research expertise at all levels.
We will continue to support our International Olympiad training teams.
We will enhance scholarships and research opportunities for our students, so they will be well-placed for competitive PhD, post-docs, and faculty openings in top institutions.
We are continuing with the AI Visiting Professorship scheme.
Since the scheme was launched , eight Professorships have been awarded to distinguished international researchers, to support their collaborations with local researchers.
The close partnership is proving fruitful in tackling important research questions.
Five of these eight Professors are here tonight.
The AI Research Centres of Excellence, which I mentioned earlier, will also be significant platforms for talent development.
In parallel, we will continue to attract top-tier AI startups and tech companies;
To base their research and innovation teams in Singapore.
Beyond schemes, mentors play a critical role in nurturing future researchers.
They include the trainers of our Olympiad Teams - Professors Alvin Chan, Liu Ziwei, Luu Anh Tuan, and Wang Wenya from NTU, as well as Mr Koo Seng Meng from AI Singapore.
Our universities’ faculty too provide important mentorship, such as through Professors Bryan Low, Ng See Kiong, and Luke Ong. They spend much of their time guiding and creating opportunities for students to learn from leading AI minds.
Thank you all for tirelessly grooming our next generation of AI leaders.
Having updated our plans, naturally, the question becomes “what does success look like?”
We believe good outcomes will emerge out of a vibrant, diverse yet close-knit research ecosystem.
It is reflected in events like the ongoing Singapore AI Research Week.
This is only the second edition;
Yet the calendar is packed with 40 events, many of them fully subscribed almost instantly.
The same energy is felt through the constant buzz at Lorong AI, our collaborative AI hub, which recorded an attendance of over 4,000 participants across 150 events last year.
This evening too, we see the close fraternity in everyone gathered here.
Leaders from academia, industry, and government;
Our international partners;
And Singaporeans, based here and abroad, making their marks in the AI field.
It seems that in quiet and unseen ways, we are all working together to shape a future where AI can be for the public good, for Singapore and the world.
Let’s keep going.