Data & analytics
UK AI Economic Measurement Lab
An independent research prototype exploring how artificial intelligence could be identified and measured within UK National Accounts frameworks.
- Status
- Prototype Live research prototype (v0.2.0, experimental)
- Access
- Open to try
- Type
- Independent research
- Role
- Sole author: research design, data layer, statistical methods and web application.
- Built with
- Next.js
- TypeScript
- Python data pipeline
- Vercel
- Link checked
- Loads with full content (10 October 2026)
The problem
AI activity sits inside broad product and industry totals, so published figures mix AI and non-AI output. Anyone trying to reason about AI's economic footprint needs a transparent way to separate candidate denominators from scenario estimates.
Who it is for
Economists, statisticians and policy analysts interested in how AI activity shows up in economic statistics.
What was built
A set of linked research tools: an AI production-stack explorer, a supply-and-use explorer covering 23 AI-relevant CPA product groups, a disaggregation lab with four scenario methods and sensitivity curves, a business-classification benchmark, an SNA 2008 / ESA 2010 asset-boundary decision tool, adoption context and a register of measurement gaps.
Key features
- Supply-and-use explorer for 23 AI-relevant product groups
- Four disaggregation scenario methods with parameter sensitivity
- Rule-based and statistical business classifier with a published benchmark
- Asset-boundary decision tool with worked case studies
- Method cards, data-quality scoring and a source register
Results so far
Live and publicly browsable. The repository records 38 automated statistical tests and a documented route-by-route deployment check.
Limitations and context
- Not an ONS product, not endorsed by ONS and does not produce official statistics.
- Decompositions are illustrative research scenarios, not estimates of AI output.
- The public repository holds documentation; application source is not published there.