About This Project
562 occupations · 88 roles · No LLM in scoring · MIT licensed · global-first methodology
Relative AI exposure rankings with separate demand and adoption context. They are not predictions of job losses or probabilities that a role will disappear.
Structural Score
Core model. Displacement pressure × (1 − demand resilience). Published as the primary dataset.
Labour Monitor
Quarterly live-market data (Singapore). Vacancy rates, hiring, retrenchment. Cluster-level, not per-occupation.
Offset & Support
Separate support layers. Offset potential, transition pathways, official skills programmes (e.g. SkillsFuture for Singapore), and scenario guidance. Useful context, not a forecast.
This model measures one side of the equation
In the Acemoglu & Restrepo (2019) framework, AI's net impact = displacement - reinstatement. We measure displacement only. Scores miss important augmentation and new-work effects, which V8 reports as a separate pathway.
State of the science (early 2026)
- Single exposure scores are poor unemployment predictors — ensembles do better (Frank et al., 2025)
- No consensus on measurement — "still in the first inning" (Brookings/PIIE, 2026)
- Entry-level workers face earliest pressure (Stanford DEL, 2025; Anthropic, 2026)
Model Card
Direct / Reproducible
- Reliability-weighted 4-source exposure ensemble when matched (AIOE + Anthropic + Eloundou + ILO)
- Theta complementarity scores (O*NET survey data)
- Within-market percentile rules (fully reproducible)
- Official demand signals (SOL 2026, Jobs in Demand)
Estimated / Group-Level
- Market resilience (group-level employment trends + occupation wage structure)
- Crosswalk quality (national occupations mapped to ISCO-08)
- Labour monitor (cluster-level, not occupation-level)
- Observed-usage calibration (Anthropic usage, not universal AI adoption)
- BLS cross-country check: ρ = 0.01 across 243 unique mapping signatures; no rank association
Synthetic / Illustrative
- Modern role estimates (weighted SSOC priors + workflow/context adjustment)
- Transition support (deterministic feasibility estimates + official programme infrastructure)
- Offset potential (heuristic demand, redesign, and friction layer)
- Outlook/scenario modelling (rule-based guidance, not prediction)
- Seniority modifiers (research-grounded, not independently validated)
Still Limited
- Occupation-level backtesting is still limited; current public validation remains cluster- and family-level, not occupation-level
- Company-size modifiers (not part of the current structural model)
- Causal displacement claims are out of scope (current evidence is correlational)
- Occupation-level employment counts (not publicly released; requested from agencies)
Data Vintage
Wages
2024 MOM data (Singapore)
Demand Signals
SOL 2026 + Jobs in Demand 2025
Labour Market
Q4 2025 full
Model Version
V7 · 221 checks
Inspiration & How We Differ
Inspired by Andrej Karpathy's AI Job Exposure Map (2026) and Josh Kale's extended visualization, which score 342 US occupations using LLM-generated ratings (Gemini Flash, 0–10 scale).
What we do differently:
- No LLM in scoring — we use deterministic transforms of published research and official data, not live model-generated ratings
- One live scored market. Singapore is the V8 reference market; U.S. and global occupation scores are withdrawn pending local validation
- Separate mechanisms — change, substitution, augmentation and labour-market context are visible rather than collapsed into one opaque probability
- Negative results retained — the deduplicated BLS comparison shows essentially no rank association, while internal checks establish reproducibility rather than external validity
- No hidden seniority adjustment — entry-level sensitivity remains unknown until a suitable open occupation-level series exists
- 88 synthetic roles — modern job titles (AI Engineer, Prompt Engineer) scored as weighted occupation blends
Frequently asked questions
What is the AI Work Index?
The AI Work Index ranks 562 Singapore occupations by relative AI exposure and publishes 88 synthetic modern-role estimates separately. Demand and adoption inform context and likely pathways, not a hidden headline multiplier. No LLM is used in the scoring pipeline.
How is the AI Exposure Rank calculated?
V8 converts a multi-source AI exposure signal into a within-Singapore percentile rank. It separately ranks substitution pressure and augmentation potential. Demand, adoption, attrition and transitions remain visible context. The score is not a probability or employment forecast.
Is the AI Work Index open source?
Yes. The entire scoring pipeline, data, and website are MIT licensed and available on GitHub at https://github.com/kirso/aiworkindex. Anyone can reproduce the results by running the deterministic scoring script.
Author & Independence
The AI Work Index is built and maintained by Kirill So as an independent project.
Self-funded, no sponsors. No advertisers, no paid placements, and no commercial relationship with any data provider or government agency. The full pipeline is open source and every score is reproducible from public inputs.
Contact: LinkedIn · GitHub issues · Press & citation
Corrections: mistakes are fixed publicly. Every methodology revision and score change is recorded in the changelog with per-release score diffs.
How to cite
So, K. (2026). AI Work Index (V7 release): structural AI displacement pressure for 562 Singapore occupations. https://aiworkindex.com
@misc{aiworkindex,
author = {So, Kirill},
title = {AI Work Index (V7): structural AI displacement pressure for 562 Singapore occupations},
year = {2026},
url = {https://aiworkindex.com},
note = {Data vintage 2026-06-11}
}License & Credits
MIT License. Adaptable for other countries via ISCO-08 crosswalks.
Made by Kirill So with Claude (Anthropic) & Codex (OpenAI). Data from MOM, BLS, O*NET, Felten et al. (2021), Pizzinelli et al. (2023), Anthropic Economic Index, Eloundou et al. (Science, 2024), ILO, and Stanford DEL.