Corporate executives are increasingly framing productivity through the lens of artificial intelligence, but they are describing the payoff as something still to come, according to new research from the Federal Reserve Bank of St. Louis.
Researchers analyzed roughly 490,000 earnings call transcripts from 5,198 publicly traded U.S. companies spanning 2000 through 2025. Using an open-source large language model, they flagged every productivity-related sentence, determined whether it also referenced AI, and classified whether the speaker was describing past, present or expected gains.
The shift in how executives talk is dramatic. Before ChatGPT’s late 2022 debut, the share of productivity sentences mentioning AI was near zero. It climbed through 2023, plateaued in 2024, then accelerated again in 2025, reaching about 15% of all productivity discussion by year-end. The language itself has evolved, too, moving from the vague “generative AI” toward more hands-on phrases like “using AI” and “AI tools.”
But nearly all of that talk is aspirational. About 95% of AI-related productivity sentences referred to future gains, compared with roughly three-quarters of non-AI productivity sentences. Sentiment was similarly lopsided: 95% of AI-related sentences described productivity as increasing.
The optimism has not yet registered in the aggregate numbers. The researchers cite San Francisco Fed data showing utilization-adjusted total factor productivity grew just 0.07% over the four quarters ending in the first quarter of 2026. A related San Francisco Fed study found that firms expressing positive AI sentiment have posted substantially higher capital spending and R&D growth, with the activity concentrated among large technology companies building AI infrastructure.
For companies weighing AI investments in contact centers, compliance review and account scoring, the research offers a reality check worth heeding. Vendors and boards alike may treat AI efficiency gains as a settled fact, but the executives running America’s largest public companies are still describing those gains to investors in the future tense. Capacity plans, staffing decisions and pricing models built on the assumption that AI has already delivered its productivity dividend are getting ahead of the evidence. The smarter posture, the data suggests, is the one those executives are actually taking: invest, experiment and reorganize now, while treating the payoff as a forecast rather than a result.
