MDDI 演講稿 · 2024-08-22
部長 Josephine Teo 在新加坡電腦學會 Tech3 論壇上的演講
要點
- • 新加坡於2023年12月釋出《國家人工智慧戰略2.0》(NAIS 2.0),確立「為公眾利益服務的AI」願景,並於2024年1月在達沃斯啟動《生成式AI模型治理框架》國際諮詢。
- • DBS銀行已部署逾600個AI/ML模型、300項用例,實現約1.8億新元的經濟價值,是本地AI商業應用的標杆案例。
- • 新加坡資訊通訊媒體發展局(IMDA)與SkillsFuture將更新《資訊通訊技術技能框架》,納入生成式AI技能,為各領域專業人士提供技能升級路徑。
- • 為擴大合資格培訓師來源,新加坡電腦學會將聯合成人學習學院推出加速認證途徑,培訓時長從88.5小時壓縮至40小時以內。
- • 16個行業已制定《職位轉型地圖》(JTMs),協助數百種崗位的從業者瞭解AI對所需技能的具體影響,並指引轉型方向。
- • 新加坡電腦學會將與廣告營銷協會及電影專業人士協會合作,為創意行業專業人士策劃AI工具應用研討會及工作坊,IMDA將通過TechSkills Accelerator計劃提供支援。
完整譯文(繁體中文)
MDDI 英文原文譯文 · 翻譯日期: 2026-06-21
部長Josephine Teo在2024年8月22日新加坡電腦學會(SCS)Tech3論壇上的演講
新加坡電腦學會會長Sam Liew先生、各位貴賓、同事與朋友們
引言
早上好,感謝再次邀請我出席SCS Tech3論壇。
我記得去年在這同一個宴會廳,當天正是總統選舉的投票日。鑑於貴會在選擇吉日方面的記錄,我在接受今年邀請時不禁想到,8月22日會不會又是另一個具有重要政治意義的日子。
幸好,總理黃循財已於週日晚間在國慶群眾大會上向我們發表了講話。大家都知道,他強調了我們在規劃新加坡前進道路時,既要大膽進取,也要審慎穩健。
確實,在致力發展新加坡人工智慧生態系統的過程中,我們採取了同樣大膽進取、審慎穩健的方式。去年12月,總理釋出了更新後的國家人工智慧戰略,即NAIS 2.0,闡明瞭我們認為有助於實現"AI為公共利益服務"這一願景的關鍵推動因素,造福新加坡與世界。
次月,在達沃斯世界經濟論壇上,我啟動了針對生成式人工智慧《模型治理框架》的國際諮詢工作,彰顯了我們在推動人工智慧應用的同時,對確保人工智慧安全的高度重視。
過去幾個月,我們推出了更多舉措——從研究專案、獎學金和人才引進計劃,到通過產業主導的卓越中心打造人工智慧新高峰。在擴大算力的同時,我們正投資於新的綠色基礎設施。通過與美國和中國的合作,我們旨在加強獲取人工智慧專業知識的渠道,並推動互操作標準的建立。
我們正穩步構建新加坡的人工智慧生態系統,並在國際上贏得廣泛關注。更重要的是,我們看到了有意義的人工智慧應用不斷湧現,有時還出現在我們最意想不到的領域!
例如,上週末我出席了"Build for Good"加速器專案的頒獎典禮。該專案是政府科技局(GovTech)旗下開放政府產品(OGP)團隊發起的倡議。許多人都知道OGP是我們的"特種部隊",是當今驅動數字政府運轉的諸多最成功數字產品和服務的幕後推手;"Build for Good"則是OGP的"特別專案",旨在讓公民參與共同創造。其中一個專案將機器學習應用於從武吉知馬自然保護區採集的音訊錄音,以識別特定鳥類。這看似不起眼,但無論從技術難度還是對自然保護人士的價值而言,都意義重大。若無法準確辨別鳥類種類並估算其數量,任何環境影響評估都無從完成。那麼,開發這一最小可行產品(MVP)的是誰呢?實際上是兩名來自新加坡國立大學(NUS)的學生,他們在數月前的一場駭客馬拉松中結識了三名陸路交通管理局(LTA)工程師,發現彼此對自然有著共同的熱情。
還有更多像這樣善用人工智慧和機器學習(ML)解決現實問題、改善市民生活的人和專案。與此同時,商業應用也在不斷增長。以DBS為例,該行已開發超過600個AI/ML模型和300個應用場景,創造了1.8億新元的經濟價值。
隨著人工智慧應用的不斷深化,我們在AI安全方面也取得了進展。為幫助各組織評估AI模型和應用程式的人工智慧風險,我們開發了名為AI Verify的測試框架和軟體工具包;通過Project Moonshot,我們將AI Verify的工作延伸至生成式人工智慧領域。
上個月,我們釋出了一套面向生成式AI模型開發者和應用部署者的《安全準則》。我們還大力推動隱私增強技術的使用,這對於確保持續獲取優質資料以用於人工智慧開發至關重要。
我感謝SCS積極為AI安全作出貢獻,並對Sam剛剛宣佈的《人工智慧倫理與治理知識體系2.0》的編制工作表示讚賞。這無疑是業界主導的有益補充,與政府推動負責任AI應用的努力相得益彰。
賦能廣大群體
今天上午,呼應黃循財總理在國慶群眾大會上關於政府支援終身學習的講話,我想重點談談NAIS 2.0的一個方面——通過技能提升來賦能我們的人民。
許多有見地的觀察人士指出,與其說是AI取代工人,不如說是具備AI能力的工人取代了缺乏AI能力的工人。換言之,與其說是"人類對抗AI",不如說將是"人類與AI協作"對抗"人類不用AI"。
這也是我和我的團隊看待人工智慧的方式。我希望你們也這樣看待AI——不是將其視為裁員的機會,而是賦能員工的契機。
這也是我們希望支援整個勞動力做好AI準備的原因,而不僅僅是一小撮技術人員。換句話說,不是每個人都能像Max Maeder那樣成為奧運獎牌得主,但人人都能、也應該學會欣賞某項運動。
據一些分析人士指出,新加坡人在人工智慧方面的適應能力已相當出色:近日,來自初級學院和理工學院的年輕學生在首屆國際人工智慧奧林匹克競賽中贏得了兩枚金牌。
國際管理發展學院的世界數字競爭力指數,在對數字技術尤其是人工智慧的適應態度方面,將我國列為全球第三。
根據畢馬威(KPMG)和昆士蘭大學的一項研究,新加坡人是全球職場中使用人工智慧比例最高的群體之一。
一家前沿AI公司近日與我們分享,新加坡人是其生成式AI應用人均使用量最高的群體,領英(LinkedIn)的《工作的未來》報告也顯示,我們仍是習得新AI技能速度最快的國家之一。
這些排名聽起來很好,但請不要對此沾沾自喜!它們本身無害,除非讓我們產生自滿情緒。請記住:以人工智慧發展的速度,我們在AI準備方面永遠不可能做得太充分。這些排名至多告訴我們,我們的人民對擁抱人工智慧充滿熱情。我們應在此基礎上加以鞏固,持續推動勢頭,凝聚各方努力,讓這份熱情轉化為切實有用的成果。
其中一個辦法是更新《資訊通訊技術(ICT)技能框架》。該框架由IMDA和新加坡技能創前程局(SkillsFuture Singapore)於2016年首次推出。該框架如同一張地圖,提供資訊通訊技術領域現有及新興技能、相關職位和職業發展路徑等關鍵資訊。我們將更新這一框架,納入生成式AI技能。這將為各領域專業人員提供寶貴資源,幫助他們瞭解充分運用生成式AI所需掌握的新技能。我們已著手與主要行業參與者、行業協會及高等學府展開諮詢,以確保更新後的框架與其需求相匹配。
我們還將擴大合資格培訓師的隊伍。他們至關重要,因為他們擔任導航者,幫助學員識別技能差距,規劃下一步發展方向。
目前,在提供經SkillsFuture批准的課程之前,培訓師需取得勞動力技能資歷(WSQ)學習與績效高階證書。這一資格需要耗時88.5小時、歷時三個月方可取得,對於全職培訓師來說尚可接受,但對於兼職提供培訓的行業從業者而言則不切實際。
我們將縮短具備相關行業知識的從業者取得培訓師資格所需的時間。正如Sam此前提到的,通過與IMDA和SkillsFuture Singapore的合作,新加坡電腦學會將與成人學習學院合作,推出一條加速路徑,耗時40小時,不到原來時長的一半。這條加速路徑將向行業從業者開放,使更多從業者具備教授與我國資訊通訊技術技能框架對應課程的資格。IMDA和新加坡電腦學會將盡快與這些從業者取得聯絡。
針對性技能提升,發揮最大效益
除了賦能廣大基礎群體,我們還將幫助我們的人才在各自專業領域掌握與AI相關的技能,從而取得成功。
我們已在16個行業推出了職位轉型地圖(Jobs Transformation Maps,簡稱JTM),涵蓋數百種職位,幫助各崗位人員瞭解技術對其所需技能的影響。
例如,會計業的JTM將AI定位為革命性技術,同時也視其為審計人員檢測異常、生成深度洞見的"得力助手"。該地圖建議金融取證領域的專業人員提升技能,以應對更復雜的案件、協助客戶預防欺詐,而不僅僅是在事後進行調查。它甚至指出,其中已具備數字取證知識的專業人員可以轉移技能,轉型至"道德駭客"等新興職位!
每個領域的專業人員都需要藉助幫助來掌握AI等數字技術。這包括創意行業的專業人員,如平面設計師、市場營銷人員、電影製作者、藝術家和作家。他們多年來精心磨練自身技藝。技術應當增強而非削弱他們的才能與貢獻。
我們已開始看到越來越多的創意專業人員將AI轉化為自身優勢。近期,一家廣告公司向MDDI提案時,使用了由Midjourney生成的影片,以及由Runway公司開發的另一款工具Gen-3 Alpha。這段AI生成的影片彙集了該公司創意專業人員的過往作品,生動呈現了故事板內容。這段影片讓我們的討論更具成效——我們可以就細節展開探討,例如背景音樂如何與各場景配合、是否與所提議的光線層次相協調,以及影片中所描繪人物的面部表情等。若沒有生成式AI,製作這樣一段"草稿"影片的成本將高得難以承受。作為甲方,我們認為,這家公司所製作的影片能夠如此有效地傳達其創意概念,使其佔據了明顯優勢——這正是"有AI加持的人"與"沒有AI加持的人"之間差距的體現。
創意專業人員還可以在日常工作中運用AI,無論是逐字檢查指令碼中的錯誤,還是加速繁瑣的照片和影片內容校正流程。事實上,NTUC Learning Hub報告指出,市場營銷專業人員是利用AI分析資料和最佳化內容的最積極使用者之一。
因此,我非常高興地注意到,新加坡電腦學會正與新加坡廣告與營銷協會(Association of Advertising and Marketing Singapore)及新加坡影視專業協會(Singapore Association of Motion Picture Professionals)攜手合作。他們將為創意專業人員策劃研討會和工作坊,幫助其充分利用AI驅動工具等數字技術。這進一步拓展了該學會已與會計、銀行與金融及物流等其他行業相關協會建立的現有合作關係。與此前一樣,IMDA將通過TechSkills Accelerator計劃提供支援,推動各行業專項數字技能提升。
結語
我希望已傳達出我們在新加坡數字發展程序中如何以大膽而審慎的態度發展AI生態系統的理念。我們正穩步超越炒作階段。隨著時間推移,我們越來越多地看到AI在多個行業和機構中提升工作方式的有力證據。我們也看到許多致力於將AI用於公共利益的熱切努力。隨著我們在這些應用場景中積累經驗,也將不斷加深對風險的認識。這將有助於我們為AI安全設計更強有力的防護機制,維護對創新與實驗至關重要的信任環境。
在所有這些努力中,我們衷心感謝新加坡電腦學會一貫堅定的支援,並期待雙方持續深化合作。
感謝你們始終積極進取的精神。祝各位大會圓滿成功。
英文原文
MDDI 官網原始記錄 · 抓取日期: 2026-06-21
SPEECH BY MINISTER JOSEPHINE TEO AT THE SINGAPORE COMPUTER SOCIETY (SCS) TECH3 FORUM ON 22 AUGUST 2024
Mr Sam Liew, President of the Singapore Computer Society, Distinguished Guests, Colleagues and Friends
INTRODUCTION
Good morning and thank you for inviting me once again to join you for the SCS Tech3 Forum.
I remember being here last year at this very ballroom, and it was polling day for the Presidential Election. Given your track record at picking an auspicious date, I wondered when I accepted your invitation this year whether 22 August would turn out to be another politically significant day.
Fortunately, Prime Minister Lawrence Wong has already addressed us on Sunday night at the National Day Rally. You will be aware that he emphasised the need for us to both bold and prudent as we chart our way forward for Singapore.
And indeed, in seeking to develop the AI ecosystem in Singapore, we have taken a similar approach of being bold and prudent. In December last year, PM launched the refreshed National AI Strategy, or NAIS 2.0, setting out the key enablers that we believe will help us to realise the vision of AI for the Public Good, for Singapore and the world.
The following month, at the World Economic Forum in Davos, I launched international consultations for our Model Governance Framework for Generative AI which underscores the importance that we place on ensuring AI safety, even as we promote AI adoption.
In the past few months, we have introduced more initiatives – from research programmes, scholarships and talent attraction schemes to growing new peaks of excellence in AI through industry-led centres of excellence. As we expand compute capacity, we are investing in new and greener infrastructure. Through our partnerships with the US and China, we aim to strengthen access to AI expertise and promote inter-operable standards.
We are steadily building up the AI ecosystem in Singapore and gaining mindshare internationally. More importantly, we are seeing meaningful AI applications appear, sometimes in areas we least expect!
For example, last weekend, I was at the finale of the Build for Good accelerator programme, an initiative by the Open Government Products (OGP) team within GovTech. Many of you know OGP as our “special forces” behind some of the most successful digital products and services that power our digital government today; “Build for Good” is OGP’s “special project” to involve citizens in co-creation. One project applied machine learning to audio recordings collected from the Bukit Timah Nature Reserve to detect specific bird species. This may not seem like a big deal, but it is, both in terms of a technical difficulty, as well as its value to conservationists. Without the ability to correctly identify bird species and estimate their numbers, no environmental impact assessment is complete. Now, who were the people who developed the MVP? They were in fact, two students from NUS who met three LTA engineers during a Hackathon only a few months ago and discovered that they shared a common passion for nature.
There are many more people and projects like this that make good use of AI and Machine Learning (ML) to solve real-world problems and improve citizens’ lives. At the same time, commercial applications are growing. DBS, for example, has already developed more than 600 AI/ML models and 300 use cases, delivering economic value of S$180 million.
Along with greater AI adoption, we are making progress in AI safety. To help organisations evaluate AI models and applications for AI risks, we developed a testing framework and software toolkit called AI Verify; through Project Moonshot, we extended our work on AI Verify into Generative AI.
Last month, we introduced a set of Safety Guidelines for Generative AI Model Developers and App Deployers. We also gave a bigger push to the use of privacy-enhancing technologies, which are critical to ensuring continued access to good quality data for AI development.
I thank SCS for your active contributions to AI safety and commend you for putting together the AI Ethics and Governance Body of Knowledge 2.0 that Sam has just announced. It is certainly a useful industry-led complement to the Government’s efforts to promote responsible AI use.
EMPOWERING THE BROAD BASE
This morning, in line with what PM Wong had said at the National Day Rally about Government’s support for lifelong learning, I want to focus on one area of NAIS2.0 – that of empowering our people through skills upgrading.
Many thoughtful observers have pointed out that it is not so much AI displacing workers, but AI-proficient workers displacing AI-deficient workers. In other words, rather than “Humans vs AI”, it is going to be “Humans with AI” vs “Humans without AI”.
This is the way my team and I think of AI. And I hope it is how you think of AI too – not as a chance to get rid of people, but as an opportunity to empower people.
This is also why we want to support the entire workforce to be more AI-ready, and not just a narrow group of technologists. Put another way, not everyone can be an Olympic medallist like Max Maeder, but everyone can and should learn to enjoy some sports.
According to some analysts, Singaporeans are already quite AI-fit: Recently, young students from junior colleges as well as polytechnics won two gold medals at the first International Olympiad in AI.
The World Digital Competitive Index by the International Institute for Management Development ranks us third in our adaptive attitudes towards digital technology, particularly towards AI.
According to a study by KPMG and the University of Queensland, Singaporeans are amongst the highest users of AI in the workplace globally.
A frontier AI company recently shared with us that Singaporeans are the highest per capita users of their generative AI application, and LinkedIn’s Future of Work report says we are still one of the fastest in acquiring new AI skills.
These rankings sound great but please take them with a pinch of salt! They are not harmful, except if they make us complacent. Remember this: at the rate that AI is advancing, we can never be too AI-ready. If anything at all, they tell us that there is enthusiasm among our people to embrace AI. We should build on it, grow the momentum and align our efforts so that the enthusiasm translates into something useful.
One way of doing so is by updating the Skills Framework for Infocomm Technology (ICT), which IMDA and SkillsFuture Singapore first launched in 2016. The Framework works like a map, providing key information on existing and emerging skills in ICT, as well as the relevant job roles and career pathways. We will refresh this framework to incorporate generative AI skills. This will provide professionals across all fields with a valuable resource on the new skills needed to get the most out of generative AI. We have started consultations with key industry players, industry associations and the Institutes of Higher Learning to ensure that the updated Framework is aligned with their needs.
We will also expand the pool of qualified trainers. They are important because they serve as navigators, helping learners to identify their skills gaps and map out their next steps.
Today, before they can deliver SkillsFuture-approved programmes, trainers need to attain the Workforce Skills Qualifications (WSQ) Advanced Certificate in Learning and Performance. This takes 88.5 hours over three months, which may be alright for a full-time trainer but is impractical for industry practitioners providing training on a part-time basis.
We will shorten the time it takes for practitioners with relevant industry knowledge to qualify as trainers. As earlier mentioned by Sam, in partnership with IMDA and SkillsFuture Singapore, the Singapore Computer Society will collaborate with the Institute for Adult Learning on an accelerated pathway that will take 40 hours, or less than half the previous duration. This accelerated pathway will be open to industry practitioners and qualify more of them to teach courses mapped to our Infocomm Technology Skills Framework. IMDA and the Singapore Computer Society will reach out to these practitioners soon.
TARGETED UPSKILLING FOR MAXIMUM IMPACT
Besides empowering the broad base, we will equip our people to succeed with AI-related skills in their respective fields of expertise.
We already have Jobs Transformation Maps, or JTMs, in 16 sectors that help individuals in hundreds of job roles understand the impact of technologies on the skills required of them.
For example, the JTM for Accountancy identifies AI as a revolutionary technology but also considers it a “valuable ally” for auditors to detect anomalies and generate deeper insights. It advises professionals in Financial Forensics to upskill to solve more complex cases and help clients prevent fraud, rather than just investigate after something bad has happened. It even suggests that those amongst these professionals who already have knowledge of digital forensics can transfer their skills and pivot to new job roles like “Ethical Hacking”!
Professionals in every domain will need help to master digital technologies such as AI. This includes professionals in the creative industries, such as graphic designers, marketers, film makers, artists, and authors. They have spent years honing their craft. Their talents and contributions should be enhanced, not diminished, by technology.
We have started to see more of our creative professionals use AI to their advantage. Recently, an advertising firm made a pitch to MDDI using a video generated by Midjourney, and another tool called Gen-3 Alpha developed by the company Runway. This AI-generated video drew on the past works done by the firm’s creative professionals and vividly conveyed the storyboard. The video made our discussions much more productive – we could talk about details like the music, how it should go with various scenes, and whether they work with the lighting levels that were proposed. And also, what were the facial expressions of the characters depicted in this video. Such a “draft” video would have been too costly to produce without Generative AI. From our perspective as the client, the fact that this particular firm had produced a video that so much more effectively conveys their concept put them at an edge, and that is what is meant by “Humans with AI” vs “Humans without AI”.
Creative professionals can also use AI in their everyday work, whether to scrub through scripts for errors, or to speed up the laborious process for correcting photo and video content. In fact, NTUC Learning Hub reports that marketing professionals are among the most avid users of AI to analyse data and optimise content.
This is why I’m very pleased to note that the Singapore Computer Society is partnering the Association of Advertising and Marketing Singapore, and the Singapore Association of Motion Picture Professionals. They will curate seminars and workshops to help creative professionals make the most of digital technologies such as AI-powered tools. This expands the existing partnerships the Society already has with relevant associations in other sectors such as Accounting, Banking & Finance, and Logistics. As in the past, IMDA will provide support through the TechSkills Accelerator programme to promote sector-specific digital upskilling.
CONCLUSION
I hope I have conveyed some sense of how we intend to be bold yet prudent in growing the AI ecosystem in Singapore, as part of our digital developments. We are steadily moving beyond the hype. With each passing day, we see more and more evidence of AI enhancing the way we work across multiple industries and organisations. We also see many enthusiastic efforts to use AI for the public good. As we gain experiences in these use cases, we will also improve our understanding of the risks. This will help us design stronger guardrails for AI safety, to uphold an environment of trust that is essential for innovation and experimentation.
In all of these efforts, we deeply appreciate the Singapore Computer Society for your steadfast support, and look forward to our continued partnership.
Thank you for always leaning forward. I wish you all a fruitful conference.