Google Expands Its AI Economics Team: How Will Jobs and Productivity Be Measured?
Google has expanded the advisers and research leadership of its AI & Economy Research Program. Its goal is to track how AI changes jobs, productivity, business growth, and scientific research through detailed data and economic analysis.[1]
Looking Beyond Adoption Statistics to Economic Effects
Google recently released AI & Economy ATLAS v1.0 and a public interactive website showing how its AI tools are used at work and in everyday life.[1] The team's expansion goes beyond counting usage to study future jobs, productivity and growth, technology diffusion around the world, and AI's effects on scientific discovery.[1]
The announcement emphasizes that technological transitions do not happen all at once.[1] Even the same tool can have different effects depending on company size, work practices, and access to education, so a single productivity figure cannot readily explain the entire economy.
Economists Alongside Empirical Research Leaders
Philippe Aghion, a 2025 Nobel laureate in economics, joins as a new academic adviser. Drawing on his research into innovation-driven growth and creative destruction, he will model AI's long-term macroeconomic trajectory.[1]
Visiting researcher Ajay Agrawal will work with MIT economics department head David Autor on AI's potential to expand scientific discovery, robotics, and human welfare.[1] The new empirical research leaders are Anu Madgavkar, whose work covers labor markets and technological transitions, and Daniel Rock, who has studied AI, labor, and firm productivity.[1]
| Research Area | Subjects Identified in the Announcement |
|---|---|
| Labor | How generative AI changes work and workforce structures |
| Productivity | Differences in outcomes depending on firm adoption and organizational practices |
| Diffusion | The pace of AI adoption across countries, regions, and small and medium-sized businesses |
| Science | AI's effects on research and discovery |
Madgavkar will focus on global AI diffusion, small and medium-sized business ecosystems, and generative AI's workforce effects. Rock will connect frontier-model usage data with econometrics to analyze firm productivity, labor restructuring, and scientific discovery.[1]
GamZip's Perspective: The Value and Limits of Provider Data
Real usage data held by a platform operator can reveal adoption patterns that surveys alone struggle to capture. Connecting ATLAS updates with empirical research may help identify more precisely which educational or organizational practices create differences in outcomes.[1]
However, this announcement concerns research plans and appointments. It does not present new causal findings establishing how much AI has already increased employment or productivity. Data from a company's own tools also reflects its product offering and user population, so the sample and methods need scrutiny before findings are generalized to other AI services or the entire labor market.
Google says it will seek to identify organizational practices, public policies, and educational programs that help AI enhance workers' capabilities and broaden expertise.[1] A substantive assessment should wait for follow-up papers to disclose their data coverage, comparison groups, methods of publication, and reproducibility.
What to Watch
- How much later ATLAS versions disclose about underlying data coverage and aggregation methods.
- Whether correlations between adoption and productivity are being presented as causal effects.
- Whether differences across large and small businesses, countries, and occupations are separated out.
- Whether policy and education recommendations are linked to actual empirical findings in follow-up papers.