Last reviewed:
What is AI-augmented productivity? Definition and business implications
AI-augmented productivity refers to measurable production gains (volume processed, speed, quality) that a worker achieves when they benefit from generative AI assistance. Real-world studies show an average gain of about 15%, strongly concentrated on the least experienced profiles.
Three real-world studies, published between 2023 and 2025, have solidly established the order of magnitude of gains. Brynjolfsson, Li and Raymond, on 5,172 customer support agents at a large Fortune 500 firm (published in Quarterly Journal of Economics, 2025): 15% more tickets resolved per hour on average, about 30% for the least experienced agents, marginal gains for experts. Cui, Demirer, Jaffe et al. (2025), on three randomised controlled trials (Microsoft, Accenture, Fortune 100), 4,867 developers: 26% more tasks completed with GitHub Copilot. On adoption, the Federal Reserve Bank of St. Louis (FRED series based on the work of Bick, Blandin and Deming) measures 45.2% of US workers aged 18 to 64 using generative AI for work in the second quarter of 2026. The academic consensus is now established: generative AI produces real and measurable gains, heterogeneous across profiles.
Concrete example
Illustrative case. A 1,200-employee European insurance company measures its gains over 9 months after deploying an AI assistant in four trades. Customer service (180 agents): 13% more cases processed, consistent with the Brynjolfsson study. Software development (45 developers): 22% more features shipped, close to the Cui et al. result. Legal (12 lawyers): no measurable gain, consistent with the hypothesis that expert profiles benefit little. Marketing (28 people): 35% more content produced, exceeding the studies due to less constrained creative work. Total: 28 full-time equivalents of capacity freed, reinvested into business growth rather than headcount reduction.
FAQ
What is AI-augmented productivity?
It refers to the measurable production gains (volume processed, speed, quality) a worker achieves with generative AI assistance. Real-world studies establish an average gain of about 15%, strongly concentrated on the least experienced profiles.
What productivity gain can we realistically expect?
On average 15%, according to the Brynjolfsson, Li and Raymond study on 5,172 support agents, but about 30% for the least experienced agents and marginal gains for experts. On software development, the Cui et al. study measures 26% more tasks completed with GitHub Copilot.
Why do gains differ so much from one team to another?
Because they depend heavily on workforce composition. A deployment on an experienced team yields modest gains, while a deployment on juniors or operational staff yields more substantial gains (about 30% for the least experienced agents in the Brynjolfsson study). Targeting the right teams changes the economics of the project.
How can we be sure the measured gains are real?
You need a solid business baseline before deployment. Without a robust metric established upfront, you will not know whether your 14% gain comes from AI or from other factors such as seasonality, training or staff turnover.
Beyond efficiency, does AI have a lasting HR effect?
Yes. The Brynjolfsson study documents a tacit knowledge transfer from senior to junior profiles: AI spreads the experts' good practices to the less experienced. This is a more durable strategic effect than short-term efficiency alone.
See also
Further reading
Generative AI at Work, Brynjolfsson, Li & Raymond, QJE, 2025
Sources
- Generative AI at Work, Brynjolfsson, Li & Raymond, Quarterly Journal of Economics vol. 140 n° 2, 2025. https://academic.oup.com/qje/article/140/2/889/7990658
- The Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software Developers, Cui, Demirer, Jaffe et al., Management Science (the former NBER w32760 URL pointed to a different paper). https://doi.org/10.1287/mnsc.2025.00535
- Generative Artificial Intelligence, Adoption Rate for Work: Employed Adults (series RPSGENAIUSAGESHAREWORK, Real-Time Population Survey, workers aged 18 to 64), FRED, Federal Reserve Bank of St. Louis, Q2 2026 value released 4 August 2026. https://fred.stlouisfed.org/series/RPSGENAIUSAGESHAREWORK
- The State of Generative AI Adoption in 2025, Bick, Blandin & Deming, Federal Reserve Bank of St. Louis, 13 November 2025. https://www.stlouisfed.org/on-the-economy/2025/nov/state-generative-ai-adoption-2025