Public data does not show AI wiping out whole occupations in Singapore. It shows AI reaching deep into office work: an estimated 44% of mapped resident workers are in occupations scoring 7 or more out of 10 for AI exposure. Among those highly exposed fields, some grew fast between 2021 and 2025 (ICT professionals, +37.9%) while others shrank (general and keyboard clerks, −15.5%).
How much of Singapore's workforce is exposed to AI
The AI exposure map scores 481 occupations from Singapore's official SSOC 2020 classification. Each score, from 0 to 10, estimates how much of the job's day-to-day work is digital and within reach of current AI systems. The occupations cover about 2.25 million resident workers (an estimate; see the notes below).
- About 44% of those workers are in occupations scoring 7 or higher.
- About 34% are in the middle band, scoring 4 to 6.
- About 22% are in occupations scoring below 4, mostly hands-on and face-to-face work.
- About 54% of the annual wages of these workers are paid in the high-exposure occupations, because those jobs tend to pay more.
The most exposed jobs are routine desk roles. Data entry clerks and telemarketers score 10. Ledger clerks, payroll clerks, bookkeepers, bank operations clerks and software developers score 9. The least exposed are physical jobs: kitchen assistants, cleaners, dishwashers, hotel housekeepers and construction labourers all score 1.
Exposed is not the same as replaced
An exposure score describes the work, not the future of the job. A score of 9 says that much of what a software developer does each day is something current AI can help with or do. It does not say whether employers will hire fewer developers, the same number, or more.
What happens next depends on things the score does not measure: whether demand for the work grows, whether AI makes each worker more productive or replaces tasks outright, and how employers redesign roles. That is why the map pairs every score with employment trends.
Which clusters grew and which shrank, 2021 to 2025
MOM reports resident employment for 38 occupation clusters. The table shows the ten clusters with the highest average AI exposure, weighted by estimated headcount, and how their employment changed over four years.
| Occupation cluster | Avg. AI exposure | Employment change, 2021–2025 |
|---|---|---|
| General & keyboard clerks | 8.7 | −15.5% |
| Numerical & material recording clerks | 8.3 | −7.9% |
| ICT professionals | 7.5 | +37.9% |
| Business & administration professionals | 7.3 | +1.9% |
| Customer service clerks | 7.1 | −0.2% |
| Other clerical support workers | 6.9 | +9.4% |
| Business & administration associates | 6.7 | +7.1% |
| Administrative & commercial managers | 6.6 | +18.6% |
| Science & engineering professionals | 6.6 | +2.7% |
| Production & specialised services managers | 6.5 | −0.2% |
Six of the ten most exposed clusters grew. The two that shrank most clearly are both clerical. Exposure alone does not sort winners from losers.
The steepest falls on the map are somewhere else entirely. Assemblers (−50.7%), personal care workers (−28.3%) and stationary plant and machine operators (−25.2%) all shrank sharply, and all score 4.1 or lower for AI exposure. Generative AI cannot explain those declines. Some of these clusters are small in the resident data, so treat the percentages with care.
The double squeeze on clerical work
Clerical work is where high exposure and falling employment meet. General and keyboard clerks average 8.7 and shrank 15.5%. Numerical and material recording clerks, such as bookkeepers and stock records clerks, average 8.3 and shrank 7.9%.
The timing matters. ChatGPT launched in November 2022, and the trend window starts in 2021. Office software was taking over filing, data entry and record keeping long before that, and the data cannot separate the two effects. A fair reading is that AI is speeding up a decline that was already under way, rather than starting a new one.
How Singapore is responding
The government's position is that AI should augment jobs, not replace workers. According to a gov.sg explainer published in May 2026:
- a new Tripartite Jobs Council brings the government, employers and unions together on responsible AI adoption, job redesign and skills training;
- SkillsFuture Singapore and Workforce Singapore merged on 1 July 2026 into the Skills and Workforce Development Agency, a single place for career guidance, training and job matching;
- from 16 September 2026, Singaporeans on selected SkillsFuture AI courses get six months of free access to premium AI tools;
- MOM says it is monitoring AI's impact on jobs.
If your job scores high, the useful question is which of your tasks AI can take on, and which skills make you the person who uses it well. The map's rationale for each occupation is a starting point.
See where your own job sits: its AI exposure score, the reasoning behind it, its median pay and how its cluster changed from 2021 to 2025.
Look up your jobFrequently asked questions
Which jobs in Singapore are most exposed to AI?
Routine clerical and desk roles. Data entry clerks and telemarketers score 10 out of 10. Ledger and accounts clerks, payroll clerks, bookkeepers, billing clerks, bank operations clerks and software developers score 9.
Which jobs are least exposed to AI?
Hands-on, physical work. Kitchen assistants, cleaners, dishwashers, hotel housekeepers, landscape workers, waste collection workers and construction labourers score 1 out of 10.
Does a high score mean I will lose my job?
No. The score estimates how much of the daily work AI can already do, not whether demand for the job will rise or fall. ICT professionals score 7.5 on average and their resident employment grew 37.9% from 2021 to 2025.
How much do exposed jobs pay?
More than most. The typical occupation scoring 7 pays a median of S$6,717 a month. See Singapore median salary by occupation for pay and exposure side by side.
Notes on the data
- Headcounts are estimates. MOM does not publish employment for each detailed occupation. The map spreads published group totals evenly across the occupations in each industry, so treat worker counts as order-of-magnitude.
- Residents only. The data covers citizens and permanent residents. The roughly 1.5 million work-pass holders are excluded, so construction and domestic work are under-represented.
- Trends are for clusters. Employment change is measured for 38 clusters, not for individual occupations.
- Common occupations only. Occupations without published wage data do not appear.
- Scores read the job title, not the person. Your actual exposure depends on your tasks, employer and tools.
Sources
- Ministry of Manpower, Labour Force in Singapore 2025, Table D8 (employed residents by occupation, 2021 to 2025). Checked 2 November 2026.
- Department of Statistics Singapore, table M182081 (employment by industry and occupation group). Checked 2 November 2026.
- Ministry of Manpower, Occupational Wages 2024, Table 4 (common occupations and wages, June 2024).
- gov.sg, How Singaporean workers are supported through the AI transition, 8 May 2026, updated 22 September 2026. Checked 2 November 2026.
- OpenAI, Introducing ChatGPT, 30 November 2022.
- AI exposure scores and rationales: AI Exposure of the Singapore Job Market, Huper Technology Pte Ltd, CC-BY 4.0.