September 2026 update
Methodology
How the scores are produced, and how to read them.
How the index is built
Start from Singapore’s Skills Framework
The Skills Framework, published by SkillsFuture Singapore, describes the work done in Singapore through 8,900 skill statements, and records which skills each job role requires and which sector each job role belongs to. The index uses this structure throughout.
Score every skill with several frontier AI models
Several frontier AI models independently assess each skill statement and estimate the share of working time that using generative AI could save, with output of broadly the same quality. The skill’s score combines these estimates, which makes it less dependent on the judgement of any single model.
From skills to job roles
The Skills Framework lists the skills each job role requires. The scores of these skills are aggregated into a single score for the job role. A typical job role draws on about 20 skills.
From job roles to sectors
The scores of the job roles in a sector are in turn aggregated into the score for that sector.
How to read a score
It is the potential to save time. A score of 0.48 means that using generative AI could save about 48% of the time now spent on the skills concerned. It is not a measure of how far firms have adopted AI.
It is not an employment forecast. Time saved can be used in many ways: more output, better quality, new tasks. What happens to a job depends on firms, workers and policy.
Scores are dated. These results reflect AI capabilities as of September 2026 and will be updated as the technology changes.
Acknowledgement
We gratefully acknowledge the generous support of Equinix for this research.
