AI productivity
Updated
What it is
AI productivity refers to measurable gains in output per worker or per dollar of input attributable to the adoption of artificial intelligence tools across the economy. It is a widely debated macroeconomic question because corporate and executive expectations of AI's benefits have so far outpaced economy-wide productivity statistics.
Why it matters lately
A Federal Reserve Bank of St. Louis study of roughly 490,000 earnings-call transcripts from 5,198 public U.S. firms (2000-2025) found that AI-related productivity commentary rose from near zero before ChatGPT's late-2022 debut to about 15% of all productivity discussion by end-2025, but roughly 95% of those AI-productivity statements describe gains executives expect in the future rather than gains already realized. Separately, an NBER study found firms that increased their share of AI-related workers from 2018-2024 saw productivity growth roughly one percentage point higher per year for each one-standard-deviation increase in that investment.
Asymmetric Analytics view
Inference (AA): ELSI treats the gap between executive AI-productivity expectations and realized aggregate productivity data as a key indicator of whether AI-driven growth narratives are getting ahead of measured reality.