Employees Using AI Are Working Faster, But the Economy Isn’t Getting More Efficient Yet
.jpg)
Why the AI productivity boom might look a lot like the early Internet era
Something strange is happening in the U.S. economy in 2026. Growth remains solid even as job creation has slowed sharply — a classic sign that existing workers are becoming more productive. Yet official productivity statistics tell a different story: growth has been modest and even slowed in the first quarter of 2026.
This apparent contradiction has a familiar ring to it.
Technologists promise that AI will dramatically boost productivity by helping workers complete tasks faster and more efficiently. Early evidence supports that at the individual level. A London School of Economics study found that employees using AI can often produce the same output in significantly less time — potentially saving the equivalent of one full workday per week.
Economists call this **capital deepening**: giving workers better tools that raise their individual output. But as a new research brief from the Federal Reserve Bank of San Francisco highlights, this phenomenon closely mirrors what happened during the early adoption of the Internet in the 1990s.
The Productivity Paradox Returns
During the 1970s through the mid-1990s, the U.S. experienced a well-known “productivity paradox.” Massive investments in computers and information technology failed to deliver measurable gains in overall economic efficiency. Robert Solow famously quipped in 1987: “You can see the computer age everywhere but in the productivity statistics.”
Then, starting in the mid-1990s, the gains finally materialized — but only after a lag.
Today’s situation shows striking parallels. Two key productivity measures are diverging:
- **Labor productivity** (output per hour worked) has posted solid gains.
- **Total Factor Productivity (TFP)** — the broader measure of how efficiently the entire economy turns inputs into outputs — has been much weaker since its post-pandemic spike.
The San Francisco Fed researchers interpret this gap as workers becoming individually faster thanks to AI, while the broader economy has yet to reorganize itself around the new technology for systemic efficiency gains.
The Growing Pains of AI Adoption
Real-world studies reveal both the promise and the friction:
- A Harvard Business Review study of 200 tech workers found that AI saved time on specific tasks, but that time was often immediately filled with more work. Many employees ended up working longer hours with fewer breaks, raising burnout risks.
- Another Harvard study warned that heavy AI use can create excessive cognitive demands, leading to mental fatigue.
- An Atlanta Fed survey of executives found that while leaders *perceive* strong productivity gains from AI, these improvements have not yet shown up clearly in hard metrics like revenue growth — a phenomenon the researchers called “delayed output realizations.”
In short, workers feel (and often are) more productive, but the economy-wide benefits are still emerging.
Looking Ahead
The San Francisco Fed researchers caution that it’s difficult to know in real time whether we’ve entered a sustained productivity boom. History suggests we may be in the early, messy phase of exactly that.
“If today mirrors what we experienced in the mid-1990s,” they write, “we may be in the early stages of a productivity boom driven by AI that will only become clear in retrospect.”
Just as the Internet revolutionized business models, supply chains, and entire industries years after its initial adoption, AI may need time to fully reshape how companies operate. The tools are already here. The deeper transformation — and the statistics to prove it — may still be coming.