Strain Says AI Still Missing From U.S. Productivity Data

Michael Strain told a Georgetown audience that AI has not shown up in U.S. productivity data, despite claims it will remake the economy. Strain, director of economic policy studies at the American Enterprise Institute and a professor of practice at Georgetown’s McCourt School of Public Policy, said the tools have not lifted output per hour in the aggregate numbers, have not sent unemployment soaring, and do not yet rank with writing or the printing press.

On September 10, 2026, he was named one of 20 commissioners on the Commission on AI and the Future of the American Workforce, co-chaired by former Commerce Secretary Gina Raimondo and former House Speaker Paul Ryan. The talk was the public version of the case he now takes into that room: history is full of “end of work” panics, and the 2026 books still look ordinary.

Strain Says the Boom Is Missing From the Books

He put three tests on the table and said AI is failing the first one. Firm-level studies can look strong. The national accounts do not.

When will it translate into productivity gains. First of all, it hasn’t yet, and I think there’s pretty broad agreement among economists on that point. There are some studies of specific firms or small groups of firms that do show that, but, when you look at the aggregate productivity data, it’s just not there.

Michael Strain, professor of practice, Georgetown University event

He was just as blunt on jobs. A collapse in work, he argued, would force a brutal choice between hunger in the streets and a much larger safety net, and he does not believe the country is heading there.

STRAIN’S THREE TESTS

• Productivity: He said the gains investors and executives expected have not appeared in the aggregate data.
• Employment: He said AI has not caused a meaningful rise in unemployment, and his worry is shrinking.
• Scale: He said domestication of plants and animals, and later the printing press, remade daily life; AI will be a big deal without reaching that rank.

Lower production costs, he added, would put downward pressure on prices if firms actually use the tools. “AI will lower production costs, or it wouldn’t be used,” Strain said. He also refused the extinction talk that still clings to the sector. “I don’t think it’s going to take all our jobs. I don’t think it’s going to kill all people on Earth,” he said. “I don’t think any of those things are going to happen, but I think it will be a big deal.”

Has AI Raised U.S. Productivity Yet?

The Bureau of Labor Statistics, in its revised second-quarter report dated September 3, 2026, said nonfarm productivity rose 1.4 percent at an annualized rate, as output rose 1.7 percent and hours worked rose 0.3 percent. From a year earlier, output per hour was up 2.2 percent. That is growth. It is not a break from the country’s long-run pace.

From the fourth quarter of 2019 through the second quarter of 2026, labor productivity has grown at a 2.1 percent annualized rate, the same as the rate since the first quarter of 1947, and above the 1.5 percent rate of the 2007-2019 cycle. The first quarter of 2026 was weaker, at 0.3 percent. Private nonfarm total factor productivity, a broader measure that nets out extra capital and labor, rose 0.8 percent in 2025 after 1.5 percent in 2024, according to the BLS annual TFP release. Labor productivity itself rose 2.2 percent in 2025 after 3.0 percent in 2024.

THE PRODUCTIVITY PRINT AGAINST ITS OWN HISTORY

Manufacturing looked firmer in the quarter, with productivity up 2.4 percent as output jumped 5.4 percent, the largest output gain since the second quarter of 2021. That still leaves Strain’s core point standing: the national print has not broken into a new regime. Unit labor costs in the nonfarm sector rose 1.2 percent in the quarter and 1.4 percent over the year, as hourly compensation rose 2.6 percent. The next look at the third quarter is due November 5, 2026.

The Same Lag Followed Electricity and Personal Computers

In 1987, Robert Solow quipped that the computer age was visible everywhere but in the productivity statistics. Erik Brynjolfsson, Daniel Rock, and Chad Syverson later framed that gap as a clash between hype and the national accounts in their modern productivity paradox working paper, arguing that a general-purpose technology shows up late because firms must rebuild processes around it. They pointed to electrification, when it took a generation, in Paul David’s account, before factory floors were laid out to use electric motors instead of a steam shaft.

That is the pattern Strain was reaching for when he talked about domestication and the printing press. Those breaks changed where people lived and how politics worked. “I think it is going to be a big deal, but maybe not as big of a deal as some folks in Silicon Valley say,” he said. On whether the current wave is overhyped, he would not give a verdict. “I think nobody really knows the answer to that question,” he said.

THE QUIET YEARS BEFORE THE BOOKS MOVE

1. 1987: Solow says you can see computers everywhere except the productivity statistics.
2. 1990: Paul David describes a generation of factory redesign before electricity paid off in measured output.
3. March 2023: Goldman Sachs economists write that generative AI could affect work equal to 300 million jobs worldwide and could lift global GDP by 7 percent.
4. 2024: Daron Acemoglu publishes a task-based estimate that puts the 10-year GDP lift in a much smaller band.
5. March 19, 2026: BLS reports total factor productivity up 0.8 percent in 2025.
6. April 24, 2026: Goldman’s labor desk publishes a U.S. payroll estimate far below the 2023 global headline.
7. September 3, 2026: BLS revises second-quarter nonfarm productivity to 1.4 percent.

On August 23, 2026, Strain posted that Sam Altman was right about ambitious timelines and still right that the technology will leave society better off. The wiring can be real while the calendar stays wrong.

Sam Altman: “We’ve all been too ambitious on timelines, even with this incredible technology.”

He’s right that many technologist have been too ambitious on timelines. And he’s right that generative AI is amazing technology that will make all of society better off.…

— Michael R. Strain (@MichaelRStrain) August 23, 2026

Goldman’s Job Count Moved From 300 Million to 16,000 a Month

The 2023 Goldman note, “The Potentially Large Effects of Artificial Intelligence on Economic Growth,” is the figure that still travels. It treated 300 million as work that could be affected worldwide, not as a same-year layoff list, and it paired that exposure with a 7 percent lift to global GDP if the tools delivered. By April 24, 2026, the same shop was measuring what had actually hit U.S. payrolls. Economist Elsie Peng wrote that AI had reduced monthly payroll growth by 16,000 jobs over the prior year and had raised the unemployment rate by 0.1 percentage point.

That 16,000 figure is already moving around as a monthly layoff wave. Goldman’s own note is narrower. It is a drag on hiring growth, not a headcount purge, and it does not fully count extra work building data centers. In roles where the tools help people rather than replace them, the same team found about 9,000 jobs a month added, and a small lowering of unemployment. The hit, Peng wrote, appears to be falling largely on younger, less-experienced workers. Telephone operators, insurance claims clerks, and bill collectors sit at the high-substitution end. Education workers, judges, and construction managers sit at the high-augmentation end.

AI augmentation that makes workers more productive can reduce the number of workers needed to produce a fixed amount of output. But by lowering the cost per unit of output, it might also increase demand for what they produce enough to generate a net increase in their employment.

Elsie Peng, economist, Goldman Sachs Research

Strain’s historical bet lines up with that small unemployment move. “If we go from a society where 85 percent or 90 percent of prime-age adults had a job to a society where 40 percent of prime individuals had a job, then our choice is going to either be that people are starving to the death on the streets or we’re going to have to really ramp up our safety net spending,” he said. “My level of concern is shrinking, not growing. I really approached this question with a historical mindset, and concerns about the end of human work have been with us for a long time.”

Why 73 Percent of Young Adults Still Expect Fewer Jobs

The hiring drag on younger workers is the piece of Goldman’s 2026 note that matches what people tell pollsters. A Pew Research Center survey of 3,488 U.S. adults, taken June 22-28, 2026, found that 71 percent of adults expect fewer jobs in the United States over the next 20 years because of AI, up from 64 percent in 2024. Only 5 percent expect more jobs. Ten percent say it will not make much difference.

WHAT THE JUNE 2026 PEW SURVEY FOUND

• Under 30, jobs: 73 percent of adults 18-29 expect fewer jobs, up from 61 percent in 2024, a 12-point rise.
• Under 30, mood: 55 percent of that age group are more concerned than excited, up from 31 percent in 2021; 11 percent are more excited, down from 25 percent.
• All adults, mood: 52 percent are more concerned than excited, up from 37 percent in 2021; 9 percent are more excited.
• Other ages, jobs: 74 percent of adults 30-49, 72 percent of those 50-64, and 63 percent of those 65 and older expect fewer jobs.

For the first time in Pew’s series, a majority of adults under 30 say they are more concerned than excited. Their job-loss view now sits in line with workers in their 30s through 60s. Strain’s shrinking personal concern and that 73 percent figure can both be true at once: the national unemployment rate has not broken, and the people most exposed to white-collar entry jobs are already pricing a thinner ladder.

Georgetown has spent the past year adding AI fellows, a planned university-wide framework, and a partnership with Google’s Gemini assistant. Students walking through those programs are being trained on tools whose payoff, in Strain’s telling, has not yet cleared the national accounts.

Acemoglu Puts the Decade Gain Near 1 Percent of GDP

Daron Acemoglu, an MIT Institute Professor and a 2024 Nobel laureate in economic sciences, has been the main academic counterweight to the 7 percent global GDP claim. In “The Simple Macroeconomics of AI,” he estimates a modest GDP gain of 1.1 to 1.6 percent over 10 years, with a roughly 0.05 percent annual gain in productivity. U.S. GDP growth has averaged about 3 percent a year since 1947, with productivity growth about 2 percent a year, so his band is a bump on the existing path, not a new one.

I don’t think we should belittle 0.5 percent in 10 years. That’s better than zero. But it’s just disappointing relative to the promises that people in the industry and in tech journalism are making.

Daron Acemoglu, Institute Professor, MIT

He puts the near-term, profitable slice of work at about 5 percent of the economy, mostly office tasks built on data summary, visual matching, and pattern recognition. “I think most companies are going to be doing more or less the same things,” he said, looking out to 2030. “A few occupations will be impacted, but we’re still going to have journalists, we’re still going to have financial analysts, we’re still going to have HR employees.” His complaint is direction, not existence: too much effort on replacing workers, not enough on giving them better information.

WHERE THE FORECASTS PART WAYS

• 2023 Goldman note: Work equal to 300 million jobs worldwide could be affected, with a possible 7 percent lift to global GDP if the tools deliver.
• Acemoglu, 10-year U.S. GDP: 1.1 to 1.6 percent in total, about 0.05 percent a year on productivity, with roughly 5 percent of tasks in profitable range.
• Strain at Georgetown: No breakout in the aggregate productivity data, no mass job collapse, a big deal that still falls short of the printing press.

Ben Moll, the Sir John Hicks Professor of Economics at the London School of Economics, put the same lag in plainer arithmetic on September 9, 2026. Double-digit GDP growth, he wrote, rests on assumptions that will not all hold in the next 10 to 15 years. “AI will affect GDP growth with long and variable lags,” he wrote. A large rise in the level of GDP spread over a decade or two is the outcome he called more likely, which is what 4 to 5 percent growth is, not 15 or 30 percent a year.

The Commission Strain Joined Has a Year to Write the Playbook

The AEI-Urban Institute commission launched on June 11, 2026, and on September 10 it named its 20 commissioners, Strain among them, with a year of playbooks promised for jobs, skills, benefits, and earnings. AEI president Robert Doar called Strain and fellow AEI scholar Brent Orrell two of the country’s leading experts on AI and the American workforce. The group is built to move in real time, which is another way of saying the evidence is still arriving.

Strain’s own close at Georgetown left the same door open. He will not sign the jobless-future brief, and he will not sign the civilizational-break brief either. He will sign a slower one: costs fall if the tools are used, prices feel some of that, and the national productivity file, for now, still reads like 2.1 percent history rather than a new age. The BLS prints the next chapter on November 5, 2026.