AI innovation should be measured by impact on people, Josephine Teo says at UN platform

AI is also being used to route public feedback to relevant agencies more quickly, identify anomalies among millions of cargo shipments at Singapore’s borders, and detect signs of phishing and impersonation scams.

Published Sep 22, 2026, 03:05 PM

Updated Sep 22, 2026, 03:05 PM

SINGAPORE – The real measure of artificial intelligence innovation is not how powerful its models become, but whether the technology improves people’s lives, said Minister for Digital Development and Information Josephine Teo on Sept 22.

Laying out Singapore’s approach to AI development and adoption at a United Nations General Assembly platform, she said the country focuses on how AI can make a difference to people, including by deploying the technology across the public service to help officers work more effectively and improve services for citizens.

Almost 70 per cent of public officers now regularly use AI tools for tasks such as analysing information and preparing drafts.

The Government is also exploring the use of AI agents to redesign entire workflows, while keeping public officers accountable for important decisions, she said at the headline segment of the Digital@UNGA , which profiles how governments and industry are deploying AI innovatively.

Minister of Digital Development and Information Josephine Teo laying out Singapore’s approach to AI development and adoption at Digital@UNGA on Sept 21. PHOTO: MDDI

Minister of Digital Development and Information Josephine Teo laying out Singapore’s approach to AI development and adoption at Digital@UNGA on Sept 21.

But for people to continue relying on AI as its use becomes more widespread, they must also be able to trust the technology, said Teo.

AI could amplify existing threats, such as making scams more convincing and personalised and cyberattacks easier to execute. Putting safeguards in place to manage these risks would therefore allow AI to be adopted with greater confidence, she added.

“From that perspective, trust does not constrain innovation. In fact, trust allows innovation to scale,” said Teo.

She is on a working visit in the United States from Sept 21 to 24, where she will be participating in events held alongside the United Nations General Assembly, and engaging with leaders in industry and academia in New York and Boston. Apart from improving government services and implementing safeguards, a third priority is broadening participation in AI, so that more countries and communities can benefit from the technology, she added.

South-east Asia, for instance, is linguistically and culturally diverse, but many AI models have a weaker understanding of the region’s languages, accents and cultural contexts.

Singapore has sought to address this through the development of SEA-LION, a family of open-source AI models designed around South-east Asian languages and cultures, including the multilingual speech and code-switching commonly used in the region.

Research institute AI Singapore is also working with partners including the Gates Foundation, Google and the United Nations Development Programme on Project ATLAS, which aims to build AI-ready datasets for under-represented languages around the world.

Teo said countries do not all need to develop their own frontier AI models, which require vast amounts of computing resources and capital. Instead, they should be supported to adapt existing AI technologies to their own languages, cultures and needs.

“For us, this is how AI inclusion can be meaningfully expanded,” she said

Singapore is also sharing its experience with other countries through the Digital Pillar of the Forum of Small States, a grouping of 108 countries, as well as providing open-source digital tools and sharing ways to build digital capabilities.

The Republic has also set up the Singapore Digital Gateway, an online platform where other countries can access digital toolkits and resources developed or used by Singapore.

“Ultimately, the success of AI should not be measured by how powerful our machines become. It should be measured by how many people are empowered by them,” she said.

AI/artificial intelligence

Artificial Intelligence

人工智能还被用于更快地将公众反馈传递给相关机构，识别新加坡边境数百万货物运输中的异常情况，以及检测网络钓鱼和身份冒用诈骗的迹象。

发布于 2026 年 9 月 22 日下午 3:05

更新于2026年9月22日下午3:05

新加坡——数码发展及新闻部长杨莉明于9月22日表示，衡量人工智能创新的真正标准不是其模型变得多么强大，而是这项技术是否改善了人们的生活。

她在联合国大会上阐述新加坡在人工智能发展和应用方面的做法时表示，新加坡关注人工智能如何为人们带来改变，包括将这项技术部署到公共服务领域，以帮助公务员更有效地工作，并改善为公民提供的服务。

近70%的公务员现在经常使用人工智能工具来完成诸如分析信息和起草文件等任务。

她在联合国大会数字峰会的主题环节中表示，政府也在探索利用人工智能代理来重新设计整个工作流程，同时确保公职人员对重要决策负责。该环节重点介绍了各国政府和行业如何创新地部署人工智能。

新加坡数码发展及新闻部长杨莉明于9月21日在联合国大会数字峰会上阐述新加坡在人工智能发展和应用方面的策略。（图片：新加坡数码发展及新闻部长办公室）

新加坡数字发展及新闻部长杨莉明于9月21日在联合国大会数字会议上阐述了新加坡在人工智能发展和应用方面的方法。

Teo表示，随着人工智能的应用越来越广泛，人们要想继续依赖人工智能，就必须能够信任这项技术。

她补充说，人工智能可能会加剧现有威胁，例如使诈骗更具说服力、更个性化，以及使网络攻击更容易实施。因此，采取保障措施来管理这些风险，将使人们更有信心采用人工智能。

“从这个角度来看，信任并不会限制创新。事实上，信任能够促进创新规模化发展，”Teo说道。

她将于9月21日至24日对美国进行工作访问，期间将出席联合国大会相关活动，并与纽约和波士顿的产业界和学术界领袖进行交流。她补充说，除了改善政府服务和落实保障措施外，第三项优先事项是扩大人工智能的参与范围，使更多国家和社区能够从中受益。

例如，东南亚在语言和文化上具有多样性，但许多人工智能模型对该地区的语言、口音和文化背景的理解较弱。

新加坡试图通过开发 SEA-LION 来解决这个问题。SEA-LION 是一系列开源人工智能模型，其设计围绕东南亚语言和文化展开，包括该地区常用的多语种语音和语码转换。

新加坡人工智能研究所 (AI Singapore) 也正与盖茨基金会、谷歌和联合国开发计划署等合作伙伴共同开展 ATLAS 项目，该项目旨在为世界各地代表性不足的语言构建人工智能就绪的数据集。

张先生表示，各国并非都需要开发自己的前沿人工智能模型，因为这需要大量的计算资源和资金。相反，应该支持各国将现有的人工智能技术应用于本国的语言、文化和需求。

她表示：“对我们而言，这才是真正意义上扩大人工智能包容性的方式。”

新加坡还通过由 108 个国家组成的小国论坛的数字支柱与其他国家分享经验，并提供开源数字工具和分享构建数字能力的方法。

新加坡还设立了新加坡数字门户网站，这是一个在线平台，其他国家可以通过该平台访问新加坡开发或使用的数字工具包和资源。

“归根结底，人工智能的成功不应该以机器变得多么强大来衡量，而应该以有多少人因此受益来衡量，”她说道。

人工智能

人工智能