SG-O-14: Jobs Versus AI in Singapore — The Labour-Market Reckoning (2023–2026)

Document Code: SG-O-14 Full Title: Jobs Versus AI in Singapore — The Labour-Market Reckoning (2023–2026) Coverage Period: 2023–2026 Level Designation: Level 1 Anchor (Block O: Mega Trends) Status: [COMPLETE]

Primary Sources Consulted:

1. Smart Nation and Digital Government Office (SNDGO) and Ministry of Communications and Information, National AI Strategy 2.0 — AI for the Public Good, For Singapore and the World, launched by Deputy Prime Minister Lawrence Wong, 4 December 2023, Singapore Conference on AI (smartnation.gov.sg/nais)
2. Forward Singapore Report — Building Our Shared Future, Government of Singapore, October 2023, with particular reference to the Empower and Equip pillars
3. Ministry of Finance, Singapore, Budget Statement 2025, delivered by Prime Minister and Minister for Finance Lawrence Wong, February 2025
4. Ministry of Finance, Singapore, Budget Statement 2026, delivered by Prime Minister and Minister for Finance Lawrence Wong, 12 February 2026, sections C (AI) and D (workforce); MOF Budget 2026 Highlights
5. Ministry of Manpower (MOM), Labour Market Report quarterly and annual issues, 2023–2026; Singapore Yearbook of Manpower Statistics (2024, 2025); Labour Force in Singapore annual reports
6. Monetary Authority of Singapore (MAS), Macroeconomic Review, semi-annual issues, 2024–2026; MAS Annual Report 2025
7. IMDA, Singapore Digital Economy Report (annual), 2023–2025; AI Talent and Workforce Reports, 2024–2025
8. International Monetary Fund, Selected Issues Paper: Impact of AI on Singapore's Labor Market (SIP/2024/040), August 2024; IMF Article IV Consultation Reports on Singapore, 2024 and 2025
9. World Economic Forum, Future of Jobs Report 2025, January 2025; Future of Jobs Report 2023, May 2023
10. SkillsFuture Singapore (SSG), Annual Reports 2023–2025; programme documentation on SkillsFuture Credit, Mid-Career Enhanced Subsidy, Career Conversion Programmes (CCPs), and SkillsFuture Jobseeker Support
11. National Trades Union Congress (NTUC) and Employment and Employability Institute (e2i), reports on PMET reskilling, 2024–2025; NTUC Company Training Committees (CTC) Grant data
12. Institute of Policy Studies (IPS), post-Budget commentaries (2025, 2026); IPS Commons articles on AI and labour-market dualism, 2024–2026, including Donald Low and Linda Lim contributions
13. DBS Group Holdings, Annual Report 2024; CEO Piyush Gupta public remarks at the Nasscom Technology and Leadership Forum, Mumbai, 24 February 2025, on AI-related workforce restructuring
14. OCBC Bank and United Overseas Bank (UOB), public disclosures on AI deployment in operations and customer service, 2024–2025
15. Ministry of Trade and Industry, Economic Survey of Singapore (quarterly and annual issues), 2024–2026
16. Singapore Parliamentary Debates (Hansard), Committee of Supply debates for MOM, MTI, and MCI/MDDI, 2024–2026; ministerial statements on AI and workforce, 2023–2026
17. Lawrence Wong, Budget 2026 Round-Up Speech, February 2026; Forward Singapore launch remarks, June 2022; National Day Rally 2024 and 2025 addresses on jobs and AI
18. Senior Minister Lee Hsien Loong, Microeconomics in Public Policy essay, March 2026 (cross-referenced in SG-L-32 and SG-O-12)
19. Minister Josephine Teo, Committee of Supply 2026 Speech, Building Singapore's Capability Advantage in a Digital Age, March 2026; Davos 2026 AI Safety remarks, 22 January 2026
20. Lim Sun Sun, public commentary on workers' adaptation to AI, 2024–2025; Donald Low (HKUST/LKYSPP), IPS commentary on AI and the Singapore model, 2024–2026 ⚠Requires verification
21. Channel NewsAsia, The Straits Times, Business Times, TODAY, contemporaneous reporting on AI, layoffs, and reskilling, 2023–2026
22. AI Singapore (AISG), AI Apprenticeship Programme (AIAP) and 100 Experiments (100E) programme documentation, 2017–2026; AISG Annual Reports 2023–2025

Related Documents:

• SG-O-01: The AI Mega Trend — Singapore's Strategy, Stakes, and Vulnerabilities
• SG-O-10: Future of Work and the Skills Economy
• SG-O-12: AI Governance Deep-Dive
• SG-K-24: Budget 2026 and the AI Transition
• SG-E-26: SkillsFuture
• SG-E-27: Committee on the Future Economy
• SG-C-20: Forward Singapore
• SG-B-09: Lawrence Wong Transition
• SG-L-17: PMO Speech Anthology — Economic Strategy, Productivity, and the Growth Compact

Version Date: 2026-10-09

---

1. Key Takeaways

• The labour-market reckoning moved from thought experiment to fiscal commitment in barely three years. When OpenAI released ChatGPT on 30 November 2022, Singapore's policy establishment was already drafting NAIS 2.0. The novelty of generative AI was not that it threatened jobs — automation had been a policy preoccupation since the Committee on the Future Economy of 2017 (cross-reference SG-E-27) — but that it threatened the professional-managerial-executive-technical (PMET) tier that was the prize of half a century of educational mobility. Within fourteen months, NAIS 2.0 (December 2023), the Forward Singapore Report (October 2023), Budget 2025 (February 2025), and Budget 2026 (12 February 2026) had moved the question to the centre of the fiscal architecture. Budget 2026 (cross-reference SG-K-24) did not announce a headline AI funding package. It set out national AI Missions, a National AI Council chaired by the Prime Minister, and worker-side AI-skilling measures built on the existing SkillsFuture architecture, and it framed them explicitly as a labour-market intervention as much as an industrial-policy one: Wong pledged that "we will not allow technological change to come at the expense of our workers".

• The IMF's August 2024 estimate that 77 per cent of Singapore's employed workforce holds jobs highly exposed to AI is the most-cited data point in the domestic debate. The IMF Selected Issues Paper (SIP/2024/040) decomposed that 77 per cent into a "complementarity" half (managers, engineers, doctors, lawyers, where AI augments) and a "substitution" half (clerical, administrative, ICT support, customer service, where displacement risk is meaningful). The figure — the highest such share among advanced economies — has been repeated in Hansard, in Budget speeches, in NTUC briefings, and in IPS commentary throughout 2024–2026. The asymmetry it captures — that Singapore's success in moving its workforce up the skills ladder has produced a labour force unusually exposed to the latest wave of automation — is the central irony of the period.

• The DBS Bank announcement of February 2025 was the first major public signal of corporate AI-driven workforce restructuring. At the Nasscom Technology and Leadership Forum in Mumbai on 24 February 2025, CEO Piyush Gupta confirmed that DBS would not renew approximately 4,000 contract and temporary positions over a three-year horizon as AI took over discrete tasks, while stating no permanent staff would be retrenched. His remark that "for the first time, I'm struggling to create jobs" became the defining phrase of the period. DBS's deployment of over 800 AI models across 350 use cases by end-2024 made the announcement credible. (Source: DBS investor relations; Business Standard and Fortune, 24–25 February 2025.) The pattern — quiet attrition rather than mass layoff, contract and temporary roles rather than permanent staff, productivity reframed as headcount discipline — has since recurred across Singapore banks and multinational operations centres.

• NAIS 2.0, launched by DPM Lawrence Wong on 4 December 2023, set the headline target of 15,000 AI practitioners — a tripling of the existing pool. The strategy organised itself around three systems and fifteen strategic actions, the most-quoted concerning talent. The 15,000 target functioned politically the way Smart Nation 2014's digital-readiness ambitions had functioned a decade earlier — as a planning benchmark giving coherence to dispersed programmes. The harder question NAIS 2.0 did not fully answer — what happens to the workers whose roles the 15,000 new practitioners will help automate — was left to the Forward Singapore architecture and subsequent budgets.

• The AI-skilling measures and the SSG-WSG merger are the labour-side response. Budget 2026's worker-side measures extend the existing SkillsFuture architecture rather than create a new named programme (cross-reference SG-K-24). They comprise practical AI capability-building starting with the accountancy and legal professions, delivered through an expanded TechSkills Accelerator; stronger AI literacy across the Institutes of Higher Learning; a redesigned SkillsFuture website with clearer AI learning pathways; six months of free access to premium AI tools for Singaporeans taking selected AI courses; and, from March 2026, extension of the SkillsFuture Level-Up Programme's Mid-Career Training Allowance to part-time trainees (cross-reference SG-E-26). The simultaneous merger of SSG and WSG into a single statutory board, jointly overseen by MOE and MOM, addressed the institutional fragmentation that had long separated training from placement. (No three-cohort "SkillsFuture for AI" programme and no "AI Career Conversion Programme" with a S$3,000 monthly allowance and guaranteed placement interviews appears in the Budget 2026 Statement or its highlights.)

• The Forward Singapore report (October 2023) supplied the social-compact frame within which the reckoning was politically legible. Forward Singapore (cross-reference SG-C-20) was not an AI document but the 4G leadership's broader social-compact exercise; its Empower and Equip pillars pre-committed the government to a more active role in cushioning workers against structural change. The SkillsFuture Jobseeker Support scheme — providing up to S$6,000 over six months for involuntarily unemployed Singaporeans actively seeking work and engaged in training — is the most consequential institutional novelty of the Forward Singapore era and the closest Singapore has come to an unemployment-insurance instrument (cross-reference SG-O-10 §7).

• The public-service AI deployment runs ahead of the private-sector one, with a different distributional logic. GovTech's Pair platform — launched September 2023 — was used by more than 50,000 civil servants by 2025, with AIBots piloted across ministries (cross-reference SG-O-12 §7). The public service has not announced AI-related headcount reductions; productivity gains have been absorbed as increased policy throughput. This augmentation-in-its-purest-form pattern has become a reference point in the broader debate about whether private employers will follow the same logic or whether displacement is the more honest description.

• The mid-career PMET squeeze is the politically salient anxiety of the period. MOM's quarterly Labour Market Reports through 2024 and 2025 showed elevated PMET retrenchment shares, with technology, professional services, and financial services contributing disproportionately: resident PMET retrenchment incidence rose from 8.6 per 1,000 resident employees in 2024 to 10.1 in 2025, above the 2015–2019 pre-recessionary norm of 8.0 (MOM, Labour Market Report 4Q 2025). The mid-career PMET narrative — workers in their forties and fifties with mortgages, school-fee obligations, and CPF trajectories built around a particular career arc — has structured the Budget 2026 mid-career measures and the rhetorical posture of Lawrence Wong's 2024 and 2025 National Day Rally addresses. Entry-level disruption is real but less politically activating; senior-worker disruption is treated through Workfare and senior employment support.

• The foreign-PMET question is inseparable from the reckoning. The COMPASS regime, evaluating Employment Pass applications against salary, qualifications, employer diversity, and skills criteria, was tightened over 2023–2025 precisely as AI-jobs anxiety intensified (cross-reference SG-O-10 §6). The narrative that AI displaces local PMETs while foreign-PMET inflows continue is politically combustible. The government's position — that selective foreign-talent inflow remains necessary to close AI-skills gaps domestic training cannot fill in the available time — sits uneasily alongside the promise that displaced locals will be retrained for AI-complementary roles.

• The Singapore wager is recognisable as a continuation of older Singapore wagers, with three new elements. The continuation: the state identifies structural transitions in advance, deploys fiscal and institutional firepower, and persuades workers and employers to follow. The new elements are (i) the explicit acknowledgement, in Forward Singapore, that not every worker will land softly and that the state must provide income support during transitions — a quiet but real departure from the "no welfare state" doctrine of 1965 (cross-reference SG-L-19); (ii) the recognition that PMETs, the class the educational ladder was built to produce, are now the displacement-exposed group; and (iii) the open question of whether voluntary AI governance frameworks (cross-reference SG-O-12) will hold if algorithmic harms in hiring, performance management, and termination accumulate without redress.

2. The Record in Brief: From Smart Nation to GenAI Shock

The arc that produced Singapore's labour-market reckoning of 2026 is short by Singapore-policy standards — barely three years from the public release of ChatGPT on 30 November 2022 to the delivery of Budget 2026 on 12 February 2026 — but it is dense. It rests on a much longer inheritance. By the time generative AI arrived in late 2022, Singapore had already run through two decades of digital-economy strategising. The Civil Service Computerisation Programme of 1981, the five sequential national IT plans from 1986 through 2010, the Smart Nation initiative launched by Prime Minister Lee Hsien Loong on 24 November 2014 at the Science Centre, and the National AI Strategy 1.0 launched by Deputy Prime Minister Heng Swee Keat in November 2019 had all produced an institutional architecture — IMDA, GovTech, SNDGO, AI Singapore, SkillsFuture Singapore, Workforce Singapore — that was unusually well-prepared for an AI shock (cross-reference SG-O-12 §2). What had not been prepared was the political and social architecture for a shock that landed on the white-collar workforce.

The generative-AI moment did not produce an immediate Singapore policy response. Through most of 2023, the public reaction was muted: ChatGPT was treated in government communications as a productivity tool to be evaluated, not a labour-market threat to be governed. The first significant institutional move was internal: GovTech's Pair platform, an internal LLM service for civil servants, was launched in September 2023, several months after similar internal deployments at the Monetary Authority of Singapore, the Inland Revenue Authority, and the Ministry of Home Affairs. The first significant external move was the launch of NAIS 2.0 on 4 December 2023, at the inaugural Singapore Conference on AI, by Deputy Prime Minister Lawrence Wong. NAIS 2.0's framing — "AI for the Public Good, For Singapore and the World" — was as much diplomatic as domestic, but its workforce target (tripling AI practitioners to 15,000) and its commitment to embed AI literacy across the wider labour force constituted the first formal acknowledgement that the labour-market implications of generative AI required a strategic response.

Three documents published in October–December 2023 set the conceptual frame for everything that followed. The Forward Singapore report (October 2023), particularly its Empower and Equip pillars, committed the government to a more redistributive social compact and a more interventionist workforce-transition architecture (cross-reference SG-C-20). NAIS 2.0 (December 2023) committed the government to an aggressive AI build-out. And the Model AI Governance Framework for Generative AI consultation, which would be published in final form on 30 May 2024, committed the government to a voluntary-framework approach to AI governance broadly (cross-reference SG-O-12 §6). These three commitments — to social redistribution, to AI capability, and to voluntary governance — were not obviously consistent, and the politics of 2024–2026 can largely be read as the working-out of their tensions.

The 2024 calendar was dominated by political transition. Lawrence Wong was sworn in as Prime Minister on 15 May 2024, succeeding Lee Hsien Loong (cross-reference SG-B-09). The first Wong budget as Prime Minister was Budget 2025, delivered in February 2025; its workforce measures were significant but incremental — additional SkillsFuture top-ups, expanded Career Conversion Programmes, and the formal implementation of the SkillsFuture Jobseeker Support scheme from April 2025 — and they were calibrated to the political reality that a general election would be called within fifteen months. The election, held on 3 May 2025, returned the People's Action Party with 65.57 per cent of the vote and 87 of 97 parliamentary seats, providing the political capital for the more ambitious commitments of Budget 2026 (cross-reference SG-K-24 §3).

By late 2025 the empirical signals were unmistakable. DBS's February 2025 announcement of 4,000 contract and temporary positions to be eliminated over three years had been followed by quieter but cumulatively significant patterns at OCBC, UOB, and the foreign banks' Singapore operations. The Ministry of Manpower's quarterly Labour Market Reports through 2024 and 2025 showed elevated PMET retrenchment shares, particularly in financial services, information and communications, and professional services. The IMF Selected Issues Paper of August 2024 had supplied the most-cited single statistic of the period — 77 per cent of Singapore's employed workers in AI-exposed occupations — and it had been absorbed into Hansard, into Budget speeches, into IPS commentary, and into NTUC briefings throughout 2025. The intellectual debate — Donald Low's IPS Commons commentaries on the limits of the Singapore reskilling model, Linda Lim on labour-market dualism, Tharman Shanmugaratnam's lifelong-learning architecture, Lim Sun Sun's NTU work on workers' adaptation — had also matured.

Budget 2026, delivered on 12 February 2026, was the response. It did not announce a headline AI funding package. Its AI measures were the national AI Missions in four sectors, the National AI Council chaired by Prime Minister Wong personally, the Champions of AI programme, and the expansion of the Enterprise Innovation Scheme so that AI expenditure qualifies for its 400 per cent tax deduction (YA2027–2028, capped at S$50,000 per Year of Assessment). Its worker-side measures were the merger of SkillsFuture Singapore and Workforce Singapore into a single statutory board, the expanded TechSkills Accelerator starting with accountancy and law, and six months of free access to premium AI tools for Singaporeans taking selected AI courses. It projected an FY2026 surplus of S$8.5 billion (1 per cent of GDP). The opposition response came from the Workers' Party, which holds all opposition seats after GE2025, and focused on the adequacy of the workforce-transition support, the distributional impact, and the absence of binding AI legislation ⚠Requires verification. Senior Minister Lee Hsien Loong's Microeconomics in Public Policy essay, published in March 2026, provided the intellectual scaffolding (cross-reference SG-L-32 and SG-O-12 §1). By May 2026 the architecture was in place; the test would be whether the labour-market data over 2026–2028 vindicated the wager or exposed its limits.

---

3. Timeline of Key Events (2023–2026)

30 November 2022: OpenAI releases ChatGPT, the public-facing interface to GPT-3.5. Within five days, the service has one million users; by January 2023, one hundred million. Singapore's policy establishment treats the moment initially as a productivity question, not a labour-market one.

March 2023: OpenAI releases GPT-4. Enterprise adoption of generative AI accelerates globally. DBS, OCBC, UOB, and Singapore's professional-services firms begin internal pilots.

June 2023: AI Verify Foundation is incorporated on 7 June 2023 as the institutional home for Singapore's voluntary AI-testing toolkit (cross-reference SG-O-12 §5).

September 2023: GovTech launches Pair, the Singapore government's internal large-language-model platform. Initial pilot covers approximately 5,000 civil servants; the platform scales to over 50,000 users by 2025.

October 2023: The Forward Singapore Report — Building Our Shared Future — is published, articulating the 4G leadership's social compact across six pillars (Empower, Equip, Care, Build, Steward, Unite). The Empower and Equip pillars commit the government to a more interventionist workforce-transition architecture (cross-reference SG-C-20).

31 October 2023: The Generative AI Evaluation Sandbox is launched by IMDA and the AI Verify Foundation, marking the first formal extension of Singapore's AI governance instruments into the LLM era (cross-reference SG-O-12 §5).

4 December 2023: Deputy Prime Minister Lawrence Wong launches the National AI Strategy 2.0 — AI for the Public Good, For Singapore and the World — at the inaugural Singapore Conference on AI. The headline workforce target is the tripling of AI practitioners to 15,000.

February 2024: Budget 2024 is delivered. SkillsFuture Credit receives a one-time S$4,000 top-up for Singaporeans aged 40 and above. SkillsFuture Jobseeker Support is announced for implementation from 2025 — Singapore's first formal unemployment-support instrument (cross-reference SG-E-26 §1 and SG-O-10 §7).

15 May 2024: Lawrence Wong is sworn in as Singapore's fourth Prime Minister, succeeding Lee Hsien Loong. Wong retains the finance portfolio (cross-reference SG-B-09).

30 May 2024: IMDA and the AI Verify Foundation publish the Model AI Governance Framework for Generative AI — Fostering a Trusted Ecosystem, extending the 2019/2020 voluntary framework to generative-AI systems (cross-reference SG-O-12 §6).

August 2024: The International Monetary Fund publishes Selected Issues Paper SIP/2024/040, Impact of AI on Singapore's Labor Market, estimating that 77 per cent of Singapore's employed workforce holds jobs highly exposed to AI — the highest such share among advanced economies. The figure rapidly becomes the single most-cited data point in the domestic debate.

22 August 2024: Minister Josephine Teo addresses the Singapore Computer Society Tech3 Forum, outlining the government's "innovation-friendly" approach to AI governance and the talent-pipeline commitments under NAIS 2.0.

May–October 2024: Project Moonshot — an open-source LLM evaluation toolkit — is released into open beta on 31 May 2024 by IMDA and the AI Verify Foundation, operationalising the Generative AI Evaluation Sandbox into a usable red-teaming tool. (Source: IMDA press release, 30–31 May 2024.) Prime Minister Wong launches Smart Nation 2.0 on 1 October 2024.

November 2024: Ministry of Manpower's Q3 2024 Labour Market Report documents elevated PMET retrenchment share relative to the historical baseline, with information and communications, professional services, and financial services contributing disproportionately ⚠Requires verification.

February 2025: Budget 2025 is delivered. Workforce measures include enhanced Career Conversion Programmes, expanded SkillsFuture training subsidies for AI-related courses, and formal commencement of SkillsFuture Jobseeker Support from April 2025.

24 February 2025: At the Nasscom Technology and Leadership Forum in Mumbai, outgoing CEO Piyush Gupta confirms that DBS will not renew approximately 4,000 contract and temporary positions over a three-year horizon as AI takes over discrete tasks. His remark that "for the first time, I'm struggling to create jobs" becomes the defining phrase of the period. (DBS's Q4 2024 financial results were separately released on 10 February 2025.) DBS has deployed over 800 AI models across 350 use cases by end-2024. (Source: Business Standard, 24 February 2025; Fortune, 25 February 2025; DBS investor relations.)

February 2025: IMDA and the AI Verify Foundation expand the Generative AI Evaluation Sandbox into the Global AI Assurance Pilot, with international partners including the UK AI Safety Institute, US NIST, and the Japan AI Safety Institute.

3 May 2025: The People's Action Party wins the 2025 General Election with 65.57 per cent of the popular vote, securing 87 of 97 parliamentary seats. The result is broadly interpreted as a mandate for the Forward Singapore agenda and the AI-centric economic strategy (cross-reference SG-K-24 §3).

Throughout 2025: OCBC, UOB, and Standard Chartered's Singapore operations expand internal generative-AI use in customer service, credit decisioning, and back-office operations. Singapore-based regional shared-services centres of multinational firms (Procter & Gamble, Unilever, Visa, others) report internal headcount discipline against productivity gains from AI deployment ⚠Requires verification.

June–November 2025: Ministry of Manpower quarterly Labour Market Reports continue to show elevated PMET retrenchment share. Resident unemployment averages 2.8 per cent for 2025 (citizen 3.0 per cent; overall 2.0 per cent), against 2.7 per cent in 2024 (MOM, Unemployment Summary Table); long-term unemployment among PMETs aged 40+ remains a focal point of NTUC and IPS commentary.

January 2026: Minister Josephine Teo delivers Davos 2026 remarks on AI Safety, restating Singapore's voluntary-framework posture ⚠Requires verification.

12 February 2026: Prime Minister and Minister for Finance Lawrence Wong delivers Budget 2026 in Parliament. Headline measures: national AI Missions in advanced manufacturing, connectivity, finance and healthcare; the National AI Council chaired by the Prime Minister; the Champions of AI programme; the expansion of the Enterprise Innovation Scheme's 400 per cent deduction to AI expenditure (YA2027–2028, capped at S$50,000 per Year of Assessment); the expanded TechSkills Accelerator and six months' free premium AI tools for selected courses; the merger of SkillsFuture Singapore and Workforce Singapore; and a projected FY2026 surplus of S$8.5 billion (cross-reference SG-K-24 §1 and SG-O-12 §10).

February–March 2026: Budget debate in Parliament. Opposition responses from Workers' Party MPs focus on the adequacy of workforce-transition support and the absence of binding AI legislation ⚠Requires verification. The PSP won no seats at GE2025 and took no part in the debate.

March 2026: Minister Josephine Teo's Committee of Supply speech, Building Singapore's Capability Advantage in a Digital Age, operationalises the Budget 2026 AI talent commitments. Senior Minister Lee Hsien Loong publishes Microeconomics in Public Policy, providing the intellectual scaffolding for the AI investment frame (cross-reference SG-L-32 and SG-O-12 §1).

April–May 2026: The Budget 2026 AI-skilling measures move into delivery; concerns about post-training placement rates for mid-career displaced PMETs persist ⚠Requires verification.

---

4. The Pre-GenAI Baseline — Industry 4.0, ITMs, and the SkillsFuture Bet

To understand the 2026 reckoning, one must first understand what Singapore had already built. By late 2022 — when ChatGPT was released — Singapore had spent nearly a decade constructing an institutional and fiscal architecture for technological transitions in the labour market. The 2023–2026 response built on this baseline rather than constructing one from scratch.

The first stratum was the Committee on the Future Economy (CFE), convened January 2016 under then-Minister for Finance Heng Swee Keat and reporting in February 2017 (cross-reference SG-E-27). The CFE's framing centred on automation, ageing, globalisation, and platform-mediated services. Its seven recommended strategies — deepening international connections, acquiring deep skills, strengthening enterprise capability, building digital capabilities, developing a connected city of opportunity, implementing Industry Transformation Maps, and partnering for innovation — produced the operational framework for the next half-decade. The CFE did not anticipate the generative-AI shock; no exercise of the period did. But its thesis — that prosperity depended on continuous skills upgrading at the level of workers, firms, and sectors — was the foundation on which SkillsFuture and ITMs were elaborated.

The second stratum was Industry Transformation Maps (ITMs). The first tranche launched in 2016, with 23 ITMs covering sectors from food manufacturing to financial services rolled out over 2016–2018 and refreshed periodically (cross-reference SG-O-10 §2). The ITMs were the developmental-state model in the labour-market domain: each identified technologies, business models, and skills defining the sector's future, and laid out transition pathways. Some — financial services, professional services, electronics — were tightly integrated with sectoral regulators and showed measurable transformation; others were more aspirational. Collectively they established that workforce transitions would be sectorally specific, technocratically planned, and tripartite in governance.

The third stratum was SkillsFuture. Launched in 2015 with the initial S$500 SkillsFuture Credit, SkillsFuture had grown by 2023 into a comprehensive infrastructure encompassing the Credit (with Budget 2024's S$4,000 enhancement for those aged 40+), the Mid-Career Enhanced Subsidy, Work-Study Programmes, Career Conversion Programmes, and the SkillsFuture Series of curated tracks in digital, green, and care skills (cross-reference SG-E-26 §3). By 2024, cumulative SkillsFuture Credit claims exceeded S$1.5 billion. But utilisation was skewed toward younger, better-educated workers, and the structural problem of mid-career reskilling at scale had not been solved. The infrastructure was the most ambitious lifelong-learning system in the developed world; it had not yet been tested against a labour-market shock of the breadth generative AI threatened.

The fourth stratum was the AI-specific apparatus. AI Singapore (AISG), established 2017 and hosted at NUS, had by 2023 trained over 500 AI practitioners through its AI Apprenticeship Programme (AIAP) and completed more than 100 industry projects through 100 Experiments (100E). The AIAP was a nine-month full-time pathway combining coursework, industry projects, and mentorship; its scaling was among NAIS 2.0's talent commitments. IMDA's TechSkills Accelerator (TeSA), launched 2016, had by early 2024 placed and trained more than 17,000 locals in in-demand tech areas through its conversion initiatives, and upskilled some 231,000 individuals (IMDA factsheet, 1 March 2024); IMDA later reported more than 24,300 TeSA placements by May 2026.

The fifth stratum was the broader manpower-policy architecture. The Progressive Wage Model, introduced 2012 for cleaners and progressively extended through 2024 to security, landscape, retail, food services, and waste management, established a wage-and-skill ladder at the bottom of the labour market structurally distinct from the SkillsFuture-and-PMET architecture above it (cross-reference SG-O-10 §5). The Workfare Income Supplement supplemented lower-wage incomes (cross-reference SG-L-19). The EP, S Pass, and Work Permit framework — calibrated over 2020–2024 through COMPASS, progressive wage thresholds, and sectoral dependency ratios — managed foreign-workforce composition (cross-reference SG-O-10 §6). The Platform Workers Act 2024 extended CPF contributions, workplace injury insurance, and union-representation rights to platform workers.

The sixth stratum was the public-service AI capability. By 2023 GovTech's Data Science and AI Division (DSAID), formed 2017, had grown to over 200 staff and was operating algorithmic decision-support tools across the Inland Revenue Authority, the Immigration and Checkpoints Authority, the Housing Development Board, the Ministry of Education, and most major statutory boards (cross-reference SG-O-12 §2). The Pair LLM platform, launched September 2023, extended this capability into general-purpose generative AI for the civil service. The infrastructure positioned the state to act as a sophisticated AI-deploying employer and to design training programmes informed by direct experience of what AI could and could not do.

This is the baseline against which the 2023–2026 response must be read. Singapore did not begin from scratch in late 2022. It began with a CFE conceptual frame, 23 sectoral ITMs, a SkillsFuture infrastructure with over S$1.5 billion in cumulative claims, an AI Singapore programme with nearly 500 trained practitioners, a manpower-policy architecture spanning PWM and WIS through COMPASS, and a sophisticated public-service AI capability. The generative-AI shock did not invalidate this baseline; it tested whether it could be scaled, retargeted, and recalibrated quickly enough to absorb a shock the IMF would soon estimate at 77 per cent of the workforce.

5. The 2023–2024 GenAI Shock — Local Cases

The generative-AI shock did not announce itself in Singapore through mass layoff. It announced itself through a quieter signal: enterprise pilots, internal productivity claims, contract-and-temporary-position attrition, and a small number of high-profile corporate disclosures that gave the underlying pattern a public face. The most important of these was DBS Bank.

DBS Bank, under CEO Piyush Gupta, had been an early and aggressive AI adopter throughout the 2010s and had positioned itself as "the most digital bank in the world." When generative AI arrived in late 2022, DBS already had the infrastructure, the data, and the in-house engineering capability to move quickly. By end-2023 the bank had identified hundreds of internal use cases for generative AI; by end-2024 it had deployed more than 800, spanning customer-service chatbots, software-engineering productivity tooling, contact-centre summarisation, marketing content generation, credit-decisioning support, fraud-pattern surfacing, and a wide range of internal knowledge-management applications.

The public moment came at the Nasscom Technology and Leadership Forum in Mumbai on 24 February 2025. Piyush Gupta confirmed that DBS would not renew approximately 4,000 contract and temporary positions over a three-year horizon as AI took over discrete tasks. (DBS's Q4 2024 financial results had been released separately on 10 February 2025.) The bank emphasised that no permanent staff would be retrenched. Gupta's now-famous remark — "for the first time, I'm struggling to create jobs" — landed precisely because it inverted the productivity-growth-jobs nexus that had underwritten Singapore's economic model for half a century. The phrase travelled through Hansard, opposition speeches, IPS commentary, and NTUC briefings. It was the moment at which the abstract IMF estimate of 77 per cent AI-exposed workforce became a concrete corporate disclosure with a number attached.

The other Singapore banks did not match DBS's disclosure cadence but followed similar internal trajectories. OCBC Bank rolled out its generative-AI chatbot — OCBC GPT, powered by Azure OpenAI — to all 30,000 staff globally in November 2023, following a six-month trial with 1,000 staff. (Source: OCBC press release, 24 October 2023, announcing November 2023 bank-wide rollout.) UOB pursued internal AI deployment more quietly. Standard Chartered's Singapore operations deployed internal LLM tooling to regional staff. The pattern — quiet attrition of contract and temporary positions, augmentation of permanent staff productivity, hiring slowdown rather than retrenchment, and a separation between the optics of "no layoffs" and the reality of workforce composition shifting — became the defining shape in Singapore's most AI-ready sector.

The public service rolled out its own infrastructure in parallel. GovTech's Pair platform launched September 2023 as an internal LLM service for civil servants, building on a closed environment that addressed data-residency and security concerns. By end-2024 Pair had over 30,000 users; by 2025 more than 50,000 across ministries and statutory boards (cross-reference SG-O-12 §7). Alongside Pair, GovTech and ministries piloted AIBots — internally trained agents calibrated to specific policy domains. The productivity claims were made in Budget 2025 and 2026 speeches; the government did not disclose AI-related headcount adjustments, and productivity gains were absorbed as increased policy throughput rather than reduced staffing.

Professional services moved more variably. The international law firms in Singapore — Allen & Gledhill, WongPartnership, Rajah & Tann, Drew & Napier, and Singapore offices of magic-circle and US firms — deployed generative-AI tools for due diligence, contract review, document summarisation, and legal research over 2023–2025. Junior-associate work was disproportionately affected: tasks that had occupied first- and second-year lawyers were compressed by AI tools to a fraction of prior hours. Whether this would manifest in slower junior-associate hiring or faster ramp-up to complex work was, by mid-2025, an open question that the Law Society and major firms had begun to discuss publicly ⚠Requires verification. The audit, tax, and consulting firms (the Big Four and Singapore competitors) deployed internal platforms with similar effect: compression of routine audit-procedure documentation, faster tax-return drafting, accelerated consulting-report production, with hiring at the junior end visibly slower than the historical baseline ⚠Requires verification.

The media sector provided different evidence. SPH Media — publisher of The Straits Times, Business Times, and a portfolio of vernacular and digital titles — was already in a structural crisis predating generative AI, having received public-funding support of up to S$900 million over five years from 2022. By 2024 SPH Media was actively integrating generative AI into editorial workflows; Mediacorp followed a similar trajectory. The combination of existing structural decline with generative-AI tooling produced visible workforce contraction in Singapore's media sector through 2024–2025, though the precise sequencing of structural and AI-driven effects was difficult to disentangle.

The early evidence did not produce a clean displacement narrative. It produced a more granular picture: large enterprises with prior digital investment moved fastest; contract and temporary positions absorbed the first wave of attrition; permanent staff productivity rose, with gains absorbed as throughput or headcount discipline rather than wage increases; and the entry-level end of the white-collar pipeline was visibly squeezed before the mid-career tier was. The mid-career tier would arrive at the centre of the political debate in 2025 not because it had been hit first but because it was the politically loudest and the policy-architecturally hardest part of the labour market to retrain.

---

6. The Government Response — NAIS 2.0, Budget 2025/2026, AI Apprenticeship

The Singapore government's response to the generative-AI shock proceeded across three intertwined tracks: the strategic-framework track (NAIS 2.0), the fiscal track (Budgets 2024–2026), and the workforce-instrument track (SkillsFuture, the Budget 2026 AI-skilling measures, and the SSG-WSG merger).

NAIS 2.0 was the strategic anchor. Launched on 4 December 2023 by DPM Lawrence Wong at the inaugural Singapore Conference on AI, the strategy structured itself around three systems and fifteen strategic actions (cross-reference SG-O-12 §4). The headline workforce commitment was the tripling of AI practitioners to 15,000 — defined broadly to encompass AI engineers, data scientists, AI translators, AI ethicists, and machine-learning operations engineers. The strategy also committed to embedding AI literacy across the broader workforce, recruiting senior AI specialists internationally, and scaling AI Singapore's apprenticeship programmes. What it did not specify was what would happen to workers displaced by AI; that was left to the Forward Singapore architecture and subsequent budgets.

The fiscal track moved in three steps. Budget 2024 committed "more than $1 billion over the next five years into AI compute, talent, and industry development" (MDDI factsheet, 1 March 2024), and supplied the SkillsFuture Credit enhancement of S$4,000 for Singaporeans aged 40 and above and announced SkillsFuture Jobseeker Support for implementation from April 2025 — not AI-targeted measures, but instruments that sharpened mid-career and unemployment-transition pathways as the AI-displacement risk crystallised. Budget 2025, Wong's first as Prime Minister, expanded Career Conversion Programmes targeted at AI-related roles and operationalised SkillsFuture Jobseeker Support. Budget 2026, delivered on 12 February 2026, was the decisive step — not through a headline AI package, which it did not announce, but through national coordination (the AI Missions and the National AI Council) and AI-skilling measures layered onto the SkillsFuture base (cross-reference SG-K-24).

Budget 2026 did not allocate a single AI investment package. The FY2026 estimates carry large top-ups to statutory and trust funds — S$5 billion to the National Research Fund and S$6 billion to the National Productivity Fund among them — but MOF does not present these as AI funding. The AI measures were instead institutional and programmatic: national AI Missions in advanced manufacturing, connectivity and logistics, finance, and healthcare, with a commitment to review regulations and create sandboxes; a Champions of AI programme for firms seeking comprehensive AI transformation; a wider Productivity Solutions Grant; and a larger AI park at one-north building on the Lorong AI pilot. Wong framed the strategic choice explicitly: "Our advantage does not lie in building the largest frontier models. It lies in deploying AI effectively, responsibly, and at speed." (No "S$5 billion AI investment package" with a four-way breakdown appears in the Budget Statement, its highlights or the FY2026 tables; see SG-K-24.)

The labour-side response was a set of AI-skilling measures rather than a new named programme (cross-reference SG-K-24). Because "AI is already reshaping many forms of white-collar and cognitive work", practical AI capability-building would "start with the accountancy and legal professions, and progressively extend them to other fields", delivered by expanding the TechSkills Accelerator. AI literacy would be strengthened across all Institutes of Higher Learning; the SkillsFuture website would be redesigned with clearer AI learning pathways; and Singaporeans taking selected AI training courses would receive six months of free access to premium AI tools. For mid-career workers, the SkillsFuture Level-Up Programme's Mid-Career Training Allowance was extended to part-time trainees from March 2026. For lower-wage workers, the Local Qualifying Salary rose to S$1,800 from July 2026, Progressive Wage Credit Scheme co-funding rose to 30 per cent for 2026, and the basic tier of Workfare Skills Support was enhanced. For older workers, the Senior Employment Credit was extended to end-2027.

The design is notable for what it omits: Budget 2026 did not create a large-scale, income-supported AI career-conversion programme. (No "AI Career Conversion Programme" with twelve months of full-time retraining, 90 per cent subsidies, a S$3,000 monthly allowance and guaranteed placement interviews, and no three-cohort "SkillsFuture for AI" programme, appears in the Budget 2026 Statement or its highlights.) Instead the government relied on existing pathways — WSG's Career Conversion Programmes, AI Singapore's AI Apprenticeship Programme, TeSA, the Level-Up Programme and SkillsFuture Jobseeker Support — with the merged SSG–WSG agency intended to join them up.

The merger of SSG and WSG into a single statutory board addressed an institutional fragmentation that had complicated the worker's journey from training to employment for nearly a decade. Under the pre-merger architecture, SSG administered SkillsFuture Credits and training-provider quality regulation; WSG administered Career Conversion Programmes and job-matching. Workers had to traverse two agency boundaries, two sets of administrative processes, and two programme catalogues. The merger created a "one-stop agency" compressing time-to-placement and reducing administrative friction (cross-reference SG-E-26 §1).

The demand-side complement was the expansion of the Enterprise Innovation Scheme: AI expenditure now counts as a qualifying activity for the scheme's existing 400 per cent tax deduction for YA2027 and YA2028, capped at S$50,000 per Year of Assessment — a targeted adoption incentive rather than a large-scale R&D subsidy. The National AI Council, chaired by Prime Minister Wong personally "to provide strategic direction and to drive Singapore's AI agenda", was the institutional capstone, elevating AI from a minister-level coordination function to a head-of-government priority (cross-reference SG-O-12 §10 and SG-K-24).

The cumulative architecture by mid-2026 — the Budget 2026 AI Missions and AI-skilling measures, the prior SkillsFuture base, the Career Conversion Programme architecture, SkillsFuture Jobseeker Support, and the Workfare and Progressive Wage scaffolding — represented the most extensive workforce-transition architecture in Singapore's history. Whether it would prove adequate to the labour-market shock it was designed to absorb was the question that 2026–2028 data would answer.

---

7. Sectoral Impact — Banking, Legal, Public Service, Media, Creative

The labour-market reckoning has not landed evenly. Some sectors moved fast and visibly; others quietly; a few have barely moved at all. The pattern matters because it determines where the policy instruments must work hardest.

Banking and financial services. This is the sector that has moved furthest and earliest. DBS's 800-plus generative-AI use cases and 4,000-position contract-and-temporary attrition disclosure are the most visible expressions; OCBC's 30,000-staff platform and UOB's quieter trajectory follow the same logic. MAS, as sectoral regulator, has anchored the response through the FEAT principles (2018), the Veritas consortium (2019–2022), and the Guidelines for AI Risk Management (Consultation Paper, 2025), supplying governance scaffolding that allowed Singapore banks to deploy AI faster than peers in less-mature regulatory environments (cross-reference SG-O-12 §6). Displacement has concentrated in contact centres, operations centres, and middle-office processing; augmentation in relationship management, software engineering, and credit analysis. The composition of the workforce has shifted faster than its size: end-2024 banking-sector employment was approximately flat year-on-year, but the role mix had measurably shifted toward AI-complementary functions ⚠Requires verification.

Legal services. The Singapore legal sector employs several thousand practising lawyers ⚠Requires verification. Generative AI's impact has concentrated at the junior-associate level — discovery, contract review, document analysis, initial drafting of routine clauses. By 2025 the major Singapore firms had deployed internal generative-AI platforms; the Law Society and Singapore Academy of Law had issued practice guidance. The displacement pattern has manifested as slower junior-associate hiring and faster progression to complex work for retained associates. International firms — whose Singapore operations function partly as regional service centres — have been visibly more cautious in associate intake ⚠Requires verification.

Public service. The deployment of Pair and AIBots is the most positively-framed AI deployment story in the Singapore economy. The government has emphasised productivity gains — drafting time, summarisation, policy-analysis throughput — without disclosing AI-related headcount adjustments. The civil service has not announced AI-driven retrenchments and is unlikely to; what it has done is constrain new hiring in some categories and reallocate freed officer time toward priorities (national security, healthcare transformation, AI policy itself) that would otherwise have required additional headcount. The civil service is also the largest single domestic deployer of generative AI and a source of accumulated experience for AI-skilling curriculum design (cross-reference SG-O-12 §7).

Media and publishing. SPH Media's pre-existing structural crisis has been overlaid on the generative-AI deployment. Editorial workflows have incorporated AI for translation, summarisation, headline generation, and back-office content production. Mediacorp's Channel NewsAsia operation has followed a parallel trajectory. The combined effect has been visible workforce contraction in Singapore's media sector through 2024–2025, with the precise sequencing of structural and AI-driven effects difficult to disentangle. The vernacular newspapers — Berita Harian, Lianhe Zaobao, Tamil Murasu — have absorbed AI translation tooling with measurable productivity gains ⚠Requires verification.

Creative, design, manufacturing, logistics, healthcare, education. Singapore's creative-economy workforce — design, advertising, marketing, content — has seen visible agency-level workforce effects from generative-AI image, video, and copywriting tooling. Manufacturing has been less directly exposed to generative AI than to the broader Industry 4.0 programme that preceded it; ITM trajectories from 2016 onward have absorbed generative AI rather than been supplanted by it. Logistics — PSA, Changi Airport, the hub-economy infrastructure — has integrated AI for routing optimisation, demand forecasting, and customs documentation, with gradual but cumulative workforce effects. Healthcare has deployed AI primarily for diagnostic support, clinical-decision support, and administrative-workload reduction; the workforce effect has been overwhelmingly augmentation given structural clinician shortages (cross-reference SG-O-05). Education has wrestled with generative AI both as a workforce question and as a curriculum question; MOE issued guidance in 2024 and updated it in 2025, with the deeper structural question concerning the redesign of what students learn.

The sectoral pattern collectively confirms what the IMF's 77 per cent estimate implied: breadth of exposure is wide, depth varies. Sectors with prior digital investment moved first; sectors with regulated workforce structures (public service, healthcare) moved through augmentation rather than displacement; sectors with structural pre-existing pressures (media) experienced compounded effects. The policy architecture — the sector-first AI-skilling measures (starting with accountancy and law), the AI Missions, the merged SSG-WSG — is calibrated to a labour market that varies sectorally in precisely these ways. Whether the calibration is fine enough is the question the next several years will answer.

8. The PMET Squeeze and the Mid-Career Reskilling Problem

The political and policy weight of the AI-jobs reckoning rests disproportionately on a single demographic: the professionals, managers, executives, and technicians (PMETs) in their forties and fifties. By 2024, PMETs constituted more than 60 per cent of Singapore's resident workforce — the prize of half a century of educational mobility and the SkillsFuture-era investment in skills upgrading. It is now, in the IMF's framing, the most AI-exposed PMET workforce in the developed world.

The vulnerability has three structural sources. The first is occupational composition. The IMF's "substitution" half — clerical, administrative, ICT support, customer service — is disproportionately occupied by mid-career workers without the highly specialised credentials that protect senior managerial roles and without the entry-level absorption-into-AI-complementary-roles that newer workers can more easily accept. The second source is wage-trajectory rigidity. A mid-career PMET typically holds an HDB or private-property mortgage, school-fee obligations, and CPF contributions calibrated to a particular wage band and career arc. Career disruption — even with training support — can produce a wage trajectory that, if reset substantially downward, destabilises a household-financial architecture built over two decades. The SkillsFuture Level-Up Programme's Mid-Career Training Allowance addresses income loss during training but not the longer-term wage-reset risk if post-training placement lands at a lower band. The third source is credential bias. Singapore's labour market has long privileged paper qualifications. A mid-career PMET re-entering the labour market after AI conversion training arrives with a hybrid credential — original degree plus conversion certificate plus project experience — whose employer-side reception is not yet established.

The SkillsFuture utilisation pattern has long flagged the challenge. By 2024 cumulative SkillsFuture Credit claims exceeded S$1.5 billion, but utilisation was skewed toward younger, better-educated workers (cross-reference SG-E-26 §1). The S$4,000 mid-career credit enhancement in Budget 2024 addressed the skew, but 2025 utilisation data suggested that structural barriers — time, employer release, training-provider credibility, placement uncertainty — were not solved by a larger credit alone. Budget 2026's response — the part-time extension of the Mid-Career Training Allowance, sector-specific AI training through TeSA, and the merged training-and-placement agency — addresses some of these barriers but not placement uncertainty directly.

NTUC's role has expanded substantially. The Employment and Employability Institute (e2i) has scaled services for displaced PMETs; the Company Training Committees (CTC) Grant programme has channelled funding to firm-level transformation initiatives. The Labour Movement's 2025 NTUC Strategy on AI, announced through Secretary-General Ng Chee Meng, committed the union to a particular framing: AI as an instrument of worker empowerment, no worker left behind, and union advocacy for binding employer retraining-rather-than-retrenchment commitments in collective agreements.

The foreign-PMET friction is inseparable from the mid-career problem. Singapore's economy depends on selective inflow of foreign professionals to fill skills gaps domestic training cannot close in the available time. The Employment Pass, S Pass, and professional-services-specific frameworks together produced an EP-and-S-Pass population of approximately 380,300 by December 2024 (202,100 EP holders and 178,200 S Pass holders). (Source: MOM, Foreign Workforce Numbers, December 2024.) The COMPASS regime, introduced for EP applications from September 2023, was the most consequential foreign-PMET policy tightening of the period. The narrative that AI displaces local PMETs while foreign-PMET inflows continue is electorally combustible; the Workers' Party and, in the 2020–2025 Parliament, the Progress Singapore Party have both raised the tension. The government has responded with COMPASS tightening, higher EP and S Pass qualifying salaries (S$6,000 and S$3,600 from January 2027, announced in Budget 2026), and the AI-skilling build-out, attempting to reserve mid-tier AI roles for retrained locals while continuing to admit senior-tier specialists. Whether the management works will be visible in 2026–2028 EP-issuance and placement data.

SkillsFuture Jobseeker Support, operational from April 2025, provides the financial cushion. Involuntarily unemployed residents with a past average monthly income of up to S$5,000, who have worked at least six of the previous twelve months and are engaged in job search and training, receive up to S$6,000 over six months. The scheme is modest by Nordic standards — Denmark's flexicurity unemployment insurance provides up to roughly 90 per cent of prior wages for up to two years (cross-reference SG-N-06) — but it represents a structural departure from the long-standing Singapore doctrine that the state should not provide unemployment income support (cross-reference SG-L-19). Early uptake data were modest and showed tight eligibility: by 31 August 2025 Workforce Singapore had received over 7,200 applications, approving almost 2,900 (40 per cent) and rejecting slightly more than 4,300 (60 per cent). Most applicants, approved and rejected, were aged 26 to 40, and the main reasons for rejection were not being assessed as involuntarily unemployed, exceeding the S$5,000 income threshold, and not meeting the six-months-worked rule (MOM written answer to Parliamentary Questions, 22 September 2025). Absolute numbers remained small against the government's projection of about 60,000 beneficiaries a year.

The squeeze is, finally, a question of identity as much as economics. The PMET tier was the prize of Singapore's developmental model — the visible evidence that the meritocratic ladder worked. To watch that tier be the first major group displaced by an exogenous technological shock is psychologically destabilising in ways the policy architecture cannot fully address. The AI-skilling measures, the SSG-WSG merger, Jobseeker Support, and COMPASS together address the financial and institutional dimensions; the identity dimension is harder, and partly what Forward Singapore and Lawrence Wong's repeated "every worker matters" emphasis attempt to address at the rhetorical level.

---

9. Intellectual Debate — Critiques and Defences

The Singapore policy establishment's response has not gone uncontested. Through 2023–2026 a distinctive intellectual debate has emerged in IPS Commons commentary, NTU and NUS academic outputs, Hansard exchanges, and the broader op-ed ecosystem. The debate has not divided cleanly along party lines; it has structured itself around conceptual disagreements about what the reckoning actually requires.

Donald Low (HKUST and former LKYSPP) has been the most consistent critic of the Singapore reskilling model's adequacy. Across IPS Commons commentary 2024–2026, Low has argued that the speed and breadth of AI displacement exceed the capacity of the SkillsFuture architecture to absorb. His argument is not that the SkillsFuture architecture is poorly designed but that the underlying mismatch between AI-driven labour-market change and human reskilling timescales is too wide to be bridged by training alone. Low has called for a more substantial unemployment-income-support architecture, a more redistributive tax-and-transfer system to absorb wage compression, and honest political acknowledgement that some displaced workers will not be re-employable at prior wage levels.

Linda Lim (University of Michigan emerita, frequent IPS contributor) has emphasised the labour-market dualism that AI risks intensifying. Her argument is that Singapore's labour market is already structured around a dual architecture — a higher tier of citizens and Employment-Pass professionals enjoying competitive wages and protection, and a lower tier of Work-Permit foreign workers and lower-income locals subject to a different regulatory and social architecture (cross-reference SG-O-08 and SG-O-10 §6). AI threatens to deepen this dualism by hollowing out the middle, forcing displaced mid-tier workers either down into the lower tier or out of the labour market entirely. Lim has emphasised institutional reform of the foreign-worker regime and a more inclusive social compact rather than additional training subsidies.

Tharman Shanmugaratnam (President of Singapore from September 2023, formerly Senior Minister and Coordinating Minister for Social Policies) has been the most prominent defender of the lifelong-learning architecture. His framing has emphasised that the question is not whether AI will displace workers but how rapidly the institutional architecture for skills upgrading, career transition, and income support can be scaled. Tharman has supported the SkillsFuture build-out and the broader Forward Singapore architecture, while acknowledging that the speed of AI-driven change tests the system's capacity.

Lim Sun Sun (communication and technology scholar, NMP 2018–2020) ⚠Requires verification has contributed the most sustained empirical work on workers' adaptation to AI in Singapore. Her commentaries 2024–2026, drawing on field research across banking, professional services, and media, have emphasised the heterogeneity of the displacement experience — that the same AI tool can be a productivity multiplier for one worker and a displacement instrument for another, depending on role, skill, age, and institutional context. Her contribution has been to push back against both the optimistic-augmentation and pessimistic-displacement narratives and to insist on granular empirical engagement.

Senior Minister Lee Hsien Loong's Microeconomics in Public Policy essay, published March 2026, provided the most influential intellectual scaffolding for the government's position (cross-reference SG-L-32 and SG-O-12 §1). The essay framed AI as a "general-purpose technology" requiring not regulatory restraint but proactive state shaping of incentives, talent, and infrastructure. Its argument was historical and analogical: electricity, the internal combustion engine, and the integrated circuit each produced labour-market disruptions that markets alone did not absorb efficiently, and state-coordinated investment in skills and institutional architecture had been the difference between successful and unsuccessful transitions. The framing supplied the conceptual scaffolding within which the Budget 2026 AI Missions and AI-skilling measures could be defended as complementary rather than competing claims.

Minister Josephine Teo has been the most consistent public defender of the AI-friendly governance posture. Across speeches at the Singapore Computer Society (August 2024), Davos (January 2026), and Committee of Supply 2026, Teo has argued that voluntary AI governance frameworks combined with heavy state investment in workforce capability produce better outcomes than binding legislation that risks suppressing investment.

The opposition has engaged through specific policy critiques rather than systemic alternative frameworks. Workers' Party leader Pritam Singh has questioned the adequacy of SkillsFuture Jobseeker Support and the absence of binding employer retraining commitments ⚠Requires verification. In the 2020–2025 Parliament, Progress Singapore Party NCMP Leong Mun Wai emphasised the foreign-PMET tension, arguing that COMPASS tightening was insufficient relative to EP issuance scale; the PSP won no seats at GE2025.

The international and comparative literature — WEF's Future of Jobs Reports 2023 and 2025, the OECD's Employment Outlook series, the IMF's August 2024 Selected Issues Paper — has informed the Singapore debate substantially. The Nordic flexicurity comparison has been the most-discussed institutional reference point (cross-reference SG-N-06). Singapore's policy choices have selectively borrowed from the flexicurity model — particularly in the Jobseeker Support design — while remaining structurally distinct from it.

The debate's centre of gravity, as of May 2026, has shifted modestly toward acknowledging that the reckoning is more substantial than a routine skills-upgrading challenge and that the policy architecture requires elements — income support during transitions, more interventionist career-conversion programmes, more sophisticated employer commitments — that depart from the historical Singapore model. Whether the shift has been sufficient is the question the empirical evidence over 2026–2028 will answer.

---

10. Outcomes and Evidence as of May 2026

The data through Q1 2026 supplies a partial but informative picture. The headline indicators — resident unemployment, retrenchment, employment composition, sectoral wage growth, AI adoption — together suggest a labour market absorbing the AI shock through slow attrition, role-mix shifts, and selective wage adjustment rather than dramatic mass displacement.

Resident unemployment. Singapore's resident unemployment rate through 2024 and 2025 averaged 2.7 per cent in 2024 and 2.8 per cent in 2025 (citizen: 2.9 and 3.0 per cent), and stood at 2.9 per cent (seasonally adjusted) in March 2026 — slightly elevated from the pre-COVID baseline but within the historical norm for a tight labour market (MOM, Unemployment Summary Table). The PMET unemployment subset has been more volatile, with long-term unemployment among PMETs aged 40 and above persisting at levels NTUC and IPS have flagged as warranting policy attention.

Retrenchment. MOM's quarterly Labour Market Reports through 2024 and 2025 documented elevated PMET retrenchment shares relative to the historical baseline, with information and communications, professional services, and financial services contributing disproportionately. Total retrenchments reached 14,490 in 2025, with incidence rising from 5.9 per 1,000 employees in 2024 to 6.3, with Financial Services (2,240) and Professional Services (1,900) recording the highest sectoral numbers in 2025; resident PMET retrenchment incidence reached 10.1 per 1,000, above the 2015–2019 norm of 8.0. MOM's own reading, in a 4Q 2025 annex examining AI-driven displacement, was that the evidence did not point conclusively to broad-based displacement: the number of residents in PMET jobs rose 3.5 per cent to 1,525,600 in 2025, and the same sectors also posted relatively high PMET vacancies (MOM, Labour Market Report 4Q 2025). Through 2025, banking-sector contract-and-temporary attrition (the DBS pattern, repeated more quietly elsewhere) supplemented conventional retrenchment categories.

Employment composition and wage growth. Overall employment grew by 44,500 in 2024 (8,800 residents) and 55,500 in 2025 (11,600 residents); in 2025 resident employment declined in Information & Communications and Professional Services after a marked increase in 2024 (MOM, Labour Market Report 4Q 2025). AI-practitioner wages grew faster than the broader PMET wage band, with senior AI engineers, AI translators, and AI ethicists commanding wage premiums of 20–40 per cent over comparable non-AI roles ⚠Requires verification. Wage growth in disrupted roles — clerical, administrative, routine analytical — has been flat or negative in real terms.

AI adoption and AI-training uptake. IMDA's Singapore Digital Economy Report 2025 documented substantial growth in enterprise AI adoption across 2023–2025, with the strongest growth in financial services, professional services, and manufacturing. SME adoption lagged the large-enterprise curve — the gap that Budget 2026's wider Productivity Solutions Grant and Enterprise Innovation Scheme AI expansion were intended to address (cross-reference SG-K-24). Uptake of the Budget 2026 AI-skilling measures through May 2026 cannot yet be assessed ⚠Requires verification, reflecting programme newness, the twelve-month full-time commitment requirement, and uncertainty about placement outcomes.

Productivity, foreign-PMET data, income support. MAS's Macroeconomic Review through 2024 and 2025 documented measurable productivity gains in AI-adopting sectors, particularly financial services and information and communications ⚠Requires verification. EP issuances reflected COMPASS tightening, with the population stable or modestly declining and the composition shifting toward higher wage thresholds and strategic-priority sectors ⚠Requires verification. SkillsFuture Jobseeker Support had approved almost 2,900 applications by 31 August 2025 (MOM written answer, 22 September 2025) ⚠Requires verification.

Worker sentiment and comparative position. Survey data from IPS, NTUC, and MOM documented elevated worker anxiety about AI-driven displacement, particularly among mid-career PMETs in disrupted sectors ⚠Requires verification. Singapore's labour-market outcomes have been broadly favourable relative to comparable advanced economies: unemployment lower than in the United States, the United Kingdom, or most European peers; wage growth positive in real terms across most sectors; productivity growth comparable or better. The IMF's Article IV consultations of 2024 and 2025 praised the Singapore policy architecture while flagging the AI-displacement risk as the most significant medium-term challenge.

The aggregate picture as of May 2026: substantial policy architecture in place, early empirical signals consistent with the augmentation-and-attrition pattern the architecture is designed for, persistent concerns about the mid-career PMET segment, and an open question about whether the architecture will hold under a more substantial labour-market shock if AI capabilities continue to advance at the rate observed over 2022–2025. The Singapore wager is in place; the test is ongoing.

11. Conclusion — The Singapore Wager on AI-Era Labour

The Singapore wager on AI-era labour, as it stands in May 2026, can be summarised in a single sentence: that voluntary AI governance frameworks combined with heavy state investment in compute and capability, paired with a segmented workforce-transition architecture and a modest income-support cushion, will absorb the labour-market consequences of a technological transformation that the IMF estimates will expose 77 per cent of the country's workforce. The wager has three legs and one assumption.

The first leg is the AI build-out. Budget 2026's national AI Missions, Champions of AI programme, expanded Enterprise Innovation Scheme and Productivity Solutions Grant, and larger AI park at one-north are intended to position Singapore as a regional AI hub generating domestic demand for AI practitioners, attracting foreign AI investment that further generates such demand, and producing productivity growth that funds the workforce-side response (cross-reference SG-K-24 and SG-O-12 §10). The National AI Council under the Prime Minister and the Smart Nation 2.0 architecture supply the supporting scaffolding. The wager is that AI investment at scale produces AI jobs at scale, and that policy can channel local workers into those jobs.

The second leg is the workforce-transition architecture. The Budget 2026 AI-skilling measures, the SSG-WSG merger, the Jobseeker Support scheme, the COMPASS tightening, the enhanced Workfare and Progressive Wage scaffolding — together the most extensive workforce-transition architecture in Singapore's history. The architecture is segmented, institutionally consolidated, and outcome-oriented. The wager is that it can retrain and redeploy displaced workers fast enough to prevent displacement from compounding into long-term unemployment, wage compression, and political destabilisation.

The third leg is the social-compact framing. Forward Singapore committed the government to a more redistributive social compact than the historical model implied (cross-reference SG-C-20). The Jobseeker Support scheme, enhanced Workfare, expanded preschool subsidies, the ComLink+ Progress Packages — each modest in isolation, collectively significant — represent a quiet shift toward more active state cushioning. The wager is that this shift is sufficient to maintain political legitimacy through the AI transition without producing the fiscal-and-cultural disruption that a more substantial welfare-state turn would imply.

The underlying assumption is that AI capabilities, while continuing to advance, will not advance so rapidly or so broadly that the architecture is overwhelmed. If displacement accelerates beyond the rate at which the SkillsFuture system can retrain workers, or lands on roles the current architecture cannot effectively reskill workers into, the wager will be tested in ways the present architecture may not absorb. The SM Lee essay positions AI as a general-purpose technology requiring sustained state shaping; this framing both motivates the substantial investment and acknowledges that the transition will be longer and more uncertain than a discrete shock would imply.

Success, by 2030, would look like: resident unemployment in the 2–3 per cent band; PMET retrenchment moderating from 2024–2025 elevated levels; AI-practitioner population reaching the 15,000 NAIS 2.0 target; high placement rates for mid-career AI conversion trainees; Jobseeker Support recipient numbers stable or declining; productivity growth accelerating across AI-adopting sectors; foreign-PMET inflow moderating as domestic AI capacity grows; political legitimacy of the architecture intact through subsequent elections. Failure would look like the inverse: elevated long-term PMET unemployment; PMET wage compression; low placement rates for conversion trainees; sectoral hollowing-out; rising Jobseeker Support recipient numbers beyond design assumptions; growing local-versus-foreign-PMET tension; and increasing calls for binding AI regulation. The realistic case lies between: uneven outcomes across sectors and worker categories, with aggregate figures remaining favourable while specific subpopulations experience substantial dislocation.

The deeper question is whether the Singapore developmental-state model itself — the government identifies structural transitions in advance, deploys fiscal and institutional firepower, and persuades workers and employers to follow — remains fit for purpose in an era of technological change that exceeds the speed at which institutional architecture can be built and recalibrated. The 2026 architecture is the most sophisticated test of this model that Singapore has yet conducted. Its outcome will tell us not only about the AI transition but about the model's capacity to absorb future structural shocks that follow it.

---

12. Spiral Index — What the Archive Has Not Yet Revealed

1. The actual displacement-versus-augmentation balance. Through May 2026, public data has been sufficient to identify the broad outlines of the AI-driven labour-market shift but insufficient to determine the precise share of displaced workers re-employed at comparable wage levels versus the share landing in lower-wage roles or exiting the labour force. The 2026–2028 MOM Labour Market Reports, the 2027 Census of Population (or its closest equivalent), and the longitudinal-tracking studies that IPS and the academic community are likely to undertake will supply the data that the present archive does not yet contain. Without this data, the policy-effectiveness debate remains partly speculative.

2. The post-training placement outcomes for mid-career AI conversion trainees. Budget 2026 relied on existing conversion pathways (WSG's Career Conversion Programmes, TeSA, the Level-Up Programme) rather than a new placement-guaranteed programme, but the placement rates, post-placement wage levels, and retention rates for AI-related conversions will not be available until late 2026 at earliest and more likely 2027. These outcomes will determine whether the conversion pathways are treated as credible by displaced workers (driving uptake) or as an institutional veneer over an unsuccessful retraining model (eroding uptake). The early empirical evidence — which the archive does not yet contain — will shape the political legitimacy of the AI-skilling architecture.

3. The precise wage and composition effects in AI-exposed sectors. Sectoral wage growth data through 2024–2025 has been suggestive but incomplete. The harder analytical question — how the within-sector wage distribution has shifted, whether the median has moved differently from the mean, and how the wage premium for AI-complementary skills compares to the wage compression in AI-substituted roles — requires more granular data than has been publicly released. The Department of Statistics and MOM data over 2026–2028 should fill this gap.

4. The political durability of the foreign-PMET architecture. COMPASS tightening through 2024–2025 has been a calibrated response to the political tension between AI-driven local PMET displacement and continued foreign-PMET inflow. Whether the tightening proves sufficient to maintain political legitimacy through the 2026–2030 period, or whether opposition pressure (from Workers' Party, Progress Singapore Party, and broader public sentiment) forces further restriction, is uncertain. The 2030 general election will likely be a substantial test.

5. The behaviour of AI capabilities themselves. The policy architecture described in this document is calibrated to AI capabilities as observed through 2025 and projected modestly forward. If AI capabilities advance more rapidly than this projection — if, for example, agentic AI systems by 2027 or 2028 can autonomously execute multi-step professional tasks at substantially higher quality and lower cost — the displacement risk would broaden beyond what the present architecture is designed to absorb. The technical trajectory of AI itself is the most consequential external variable, and it is the one over which Singapore policy has the least direct control.

---

13. Sources and References

1. Smart Nation and Digital Government Office (SNDGO) and Ministry of Communications and Information, National AI Strategy 2.0 — AI for the Public Good, For Singapore and the World, launched by Deputy Prime Minister Lawrence Wong, 4 December 2023, at the inaugural Singapore Conference on AI (smartnation.gov.sg/nais).

2. Forward Singapore Report — Building Our Shared Future, Government of Singapore, October 2023, with particular reference to the Empower and Equip pillars on workforce resilience and lifelong learning.

3. Ministry of Finance, Singapore, Budget Statement 2025, delivered by Prime Minister and Minister for Finance Lawrence Wong, February 2025, including the formal implementation of SkillsFuture Jobseeker Support from April 2025.

4. Ministry of Finance, Singapore, Budget Statement 2026, delivered by Prime Minister and Minister for Finance Lawrence Wong, 12 February 2026; Budget Debate: Round-Up Speech, February 2026; Revenue and Expenditure Estimates for FY2026; Budget Highlights 2026 and supporting annexes. Key measures include the national AI Missions, the National AI Council, the Champions of AI programme, the Enterprise Innovation Scheme AI expansion (400 per cent deduction, YA2027–2028, capped at S$50,000 per Year of Assessment), the expanded TechSkills Accelerator and premium-AI-tool access, the merger of SSG and WSG, and a projected FY2026 surplus of S$8.5 billion.

5. Ministry of Manpower (MOM), Labour Market Report (quarterly), 2023–2026; Singapore Yearbook of Manpower Statistics (2024, 2025); Labour Force in Singapore (annual). PMET retrenchment shares, resident unemployment, employment composition, and sectoral employment data are drawn from these series.

6. Monetary Authority of Singapore (MAS), Macroeconomic Review (semi-annual), 2024–2026; Annual Report 2025. AI-adoption productivity-growth estimates and sectoral macroeconomic assessments.

7. IMDA, Singapore Digital Economy Report (annual), 2023–2025; AI talent and workforce-related publications, 2024–2025; Model AI Governance Framework for Generative AI, 30 May 2024; Project Moonshot, October 2024.

8. International Monetary Fund, Selected Issues Paper: Impact of AI on Singapore's Labor Market (SIP/2024/040), August 2024 — the source of the 77 per cent AI-exposed workforce estimate; Article IV Consultation Reports on Singapore, 2024 and 2025.

9. World Economic Forum, Future of Jobs Report 2023, May 2023; Future of Jobs Report 2025, January 2025.

10. SkillsFuture Singapore (SSG), Annual Reports 2023–2025; programme documentation on SkillsFuture Credit (including the Budget 2024 S$4,000 enhancement for citizens aged 40 and above), Mid-Career Enhanced Subsidy, Career Conversion Programmes, and SkillsFuture Jobseeker Support; Workforce Singapore (WSG), Annual Reports 2023–2025.

11. National Trades Union Congress (NTUC), Employment and Employability Institute (e2i) reports, 2024–2025; NTUC Strategy on AI (2025); NTUC Company Training Committees (CTC) Grant programme documentation.

12. Institute of Policy Studies (IPS), post-Budget commentaries (2025, 2026); IPS Commons articles on AI and the Singapore labour market, 2024–2026, including contributions by Donald Low, Linda Lim, and other commentators; IPS post-election analyses (May 2025).

13. DBS Group Holdings, Annual Report 2024; CEO Piyush Gupta's public remarks at the Nasscom Technology and Leadership Forum, Mumbai, 24 February 2025, on AI-related workforce restructuring and the 4,000 contract-and-temporary-position attrition disclosure (Q4 FY2024 financial results released separately, 10 February 2025); DBS public communications on internal generative-AI deployment.

14. OCBC Bank and United Overseas Bank (UOB), public disclosures and corporate communications on AI deployment in operations, customer service, and credit decisioning, 2024–2025.

15. Ministry of Trade and Industry, Economic Survey of Singapore (quarterly and annual), 2024–2026.

16. Singapore Parliamentary Debates (Hansard): Committee of Supply debates for MOM, MTI, and MCI/MDDI, 2024–2026; ministerial statements on AI and workforce, 2023–2026; Budget 2025 and Budget 2026 debates, including opposition responses by Workers' Party MPs; foreign-PMET debates of the 2020–2025 Parliament, including Progress Singapore Party NCMP Leong Mun Wai.

17. Lawrence Wong: Budget Speeches 2024, 2025, 2026; Budget 2026 Round-Up Speech, February 2026; Forward Singapore launch remarks, June 2022; National Day Rally addresses 2024 and 2025 on jobs and AI; remarks at the Singapore Conference on AI, 4 December 2023.

18. Senior Minister Lee Hsien Loong, Microeconomics in Public Policy essay, March 2026 (cross-referenced in SG-L-32 and SG-O-12); selected earlier speeches on technology and the future of work.

19. Minister Josephine Teo, Speech at Singapore Computer Society Tech3 Forum, 22 August 2024; Committee of Supply 2026 Speech, Building Singapore's Capability Advantage in a Digital Age, March 2026; Remarks at World Economic Forum, Davos, 22 January 2026.

20. Lim Sun Sun, public commentary on workers' adaptation to AI, 2024–2025; Donald Low (HKUST and former LKYSPP), IPS commentary on AI and the Singapore model, 2024–2026 ⚠Requires verification; Linda Lim, IPS commentary on labour-market dualism, 2024–2025.

21. Channel NewsAsia, The Straits Times, Business Times, TODAY, contemporaneous reporting on AI, layoffs, reskilling, COMPASS, and Budget 2026, 2023–2026.

22. AI Singapore (AISG), AI Apprenticeship Programme (AIAP) documentation, 2017–2026; 100 Experiments (100E) programme documentation, 2017–2026; AISG Annual Reports 2023–2025.