Press & Citation

The headline score is a relative AI exposure rank, not a prediction of job loss. Accurate phrasing: “this occupation is more exposed to current AI capabilities than N% of Singapore occupations”. Please avoid: “X% chance of losing your job”, “X jobs will be lost”, or any framing that treats the score as a forecast. Research finds several possible channels, including changes in hiring, wages, task mix, and role design; the index does not predict which channel will dominate.

So, K. (2026). AI Work Index (V7 release). https://aiworkindex.com — data vintage 2026-06-11.

@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}
}

Attribution is required (MIT licensed). Link to the occupation page you reference so readers can see the evidence, ranges, and caveats behind the single number.

• Downloads — full datasets (JSON/CSV), the release manifest with checksums, and every validation artifact are on the data page.
• Methodology — formulas, thresholds, validation results (including the honest negatives), and version history are on the methodology page; exact constants in the appendix.
• Reproducibility — the pipeline is deterministic (no LLM assigns any score) and open source at GitHub.
• Independence — self-funded; no sponsors, advertisers, or commercial relationships with data providers. Corrections are public in the changelog.

Built and maintained by Kirill So. For interviews, data questions, or corrections: LinkedIn · GitHub issues. Custom cuts of the data for stories are possible — the pipeline is parameterized.

Current release: V7 · 2026-06-11 · 221 automated validation checks