Georgetown Bets on AI While Strain Finds No Productivity Jump

Michael Strain told a Georgetown audience that artificial intelligence still does not show up in the country’s output-per-hour numbers, even as the same campus takes applications for paid AI fellows and hands students Gemini licenses.

Strain, director of economic policy studies at the American Enterprise Institute and a professor of practice at the McCourt School of Public Policy since 2023, said firm-level studies can look promising and the national file does not. He also said his worry about a jobs collapse is shrinking, not growing.

The Aggregate Data Still Has Not Moved

The question hanging over the room was timing. Investors and executives have spent three years waiting for a productivity boom after ChatGPT arrived in late 2022. Strain said that wait is not over.

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 event

The official file is not frozen. In its revised second-quarter 2026 release, the U.S. Bureau of Labor Statistics said nonfarm business productivity increased 1.4 percent at an annual 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.

Over the current business cycle, which the bureau dates from the fourth quarter of 2019 through the second quarter of 2026, nonfarm productivity has grown at a 2.1 percent annual rate. That matches the long-term pace since the first quarter of 1947 and sits above the 1.5 percent pace of the 2007 through 2019 cycle. Strain’s claim is narrower. He is not saying output per hour is zero. He is saying the AI stamp still is not readable in the totals.

THE PRODUCTIVITY FILE, Q2 2026

The same release put the labor share, the slice of output that shows up as worker pay, at 52.8 percent in the second quarter, the lowest reading in a series that begins in 1947. Real hourly compensation fell 3.3 percent in the quarter. Those figures describe a squeeze on labor’s cut of output. They do not, on their own, prove that generative AI caused it.

A Kansas City Fed industry breakdown earlier in 2026 found labor productivity had moved above its pre-pandemic trend, with the pickup concentrated in a thin set of industries, and concluded that AI still explained little of the aggregate gain. A St. Louis Fed review of earnings calls found utilization-adjusted total factor productivity up only 0.07 percent over the four quarters ending in the first quarter of 2026, with most AI talk about productivity aimed at future gains rather than ones already booked.

He Is Already on a National Jobs Commission

The remarks landed while Strain is being asked to help write the policy response to the shock he says has not arrived in the totals. On September 10, 2026, the AEI-Urban Institute Commission on AI and the Future of the American Workforce named 20 commissioners to advise policymakers on jobs, skills, benefits, and pay. Strain is on that list. So is Harry Holzer, another McCourt public policy professor.

Former Commerce secretary Gina Raimondo and former House speaker Paul Ryan chair the group, which launched on June 11, 2026, and is supposed to convene over the next year. The charge is practical playbooks, issued as the effects become clearer. That is a hard assignment if the national productivity file, as Strain told campus, is still blank on AI.

THE DATES AROUND THE REMARKS

1. June 11, 2026: The AEI-Urban commission on AI and the American workforce launches.
2. September 3, 2026: BLS publishes the revised second-quarter productivity and costs report.
3. September 10, 2026: The commission names its 20 members, including Strain and Holzer.
4. September 12, 2026: Strain’s campus remarks on missing aggregate gains are published.
5. September 28, 2026: Georgetown AI Fellows applications close.

In late August, Strain wrote that OpenAI chief executive Sam Altman was right that many technologists have been too ambitious on timelines, and that generative AI is still a technology that will make society better off. That is the same split he brought into the Georgetown room. The tools are real. The national clock has not jumped.

Georgetown’s Gemini Licenses and $2,500 Fellows

The university is not waiting for that clock. The President’s Office launched an AI Fellows program that pairs faculty members with AI-fluent undergraduates on a course for a semester. Each fellowship pays $2,500 for about 130 hours of student work across fall and spring. Faculty and student applications are due September 28, 2026, with selections due by October 5.

University Information Services already offers an enterprise Gemini license to staff, faculty, and students after a 2025 pilot. The campus instance is sold as a way to draft, summarize, and search mail and docs without feeding chats into Google’s training runs. Faculty still set the rules in each class.

WHAT CAMPUS IS STANDING UP

• Paid fellows: Undergraduates work with faculty to rethink assignments, test tools against real course tasks, and set rules for AI use, for $2,500 and about 130 hours.
• Gemini licenses: The university rolled enterprise access to Google’s assistant after a 2025 faculty and staff pilot, with student claiming opened in March 2026.
• A campus framework: University leaders have said they will write a university-wide AI framework for teaching, research, and operations.
• New credentials: University notices describe a College of Arts and Sciences AI certificate, an MBA AI core requirement, and a master’s in AI management launched in 2025.

Those moves treat AI as something students must learn to use at work. Strain’s lecture treated it as something the national accounts have not yet absorbed. Both can be true at once, and that is the awkward part. A campus can train people on a tool whose economy-wide footprint is still small.

Why Young Adults Expect Fewer Jobs

The students in those fellowships are the age group now most convinced the tool will erase work. Pew Research Center surveyed 3,488 U.S. adults from June 22 to 28, 2026, and found that 73 percent of adults under 30 think AI will lead to fewer jobs in the United States over the next 20 years, up from 61 percent in 2024. Among all adults the share is 71 percent, up from 64 percent. Only 5 percent expect more jobs.

For the first time since Pew began asking in 2021, a majority of adults under 30, 55 percent, say they are more concerned than excited about AI in daily life. That is up from 31 percent in 2021. Only 11 percent of that age group are more excited than concerned. Overall, 52 percent of adults are more concerned than excited, up from 37 percent in 2021. Colleen McClain and Eugenie Park wrote the analysis.

Strain’s thermometer runs the other way. “My level of concern is shrinking, not growing,” he said. “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.” The gap is not a polling error. Young workers can feel hiring freeze in their own inboxes while an economist watches the unemployment rate and the output-per-hour index and sees no collapse.

The household survey still looks ordinary. BLS said the unemployment rate in August 2026 was unchanged at 4.1 percent, with payrolls up 162,000 and 7.0 million people unemployed. Strain said AI has not moved that rate in a meaningful way. He also sketched the nightmare he does not expect: a drop from 85 percent or 90 percent of prime-age adults in work to 40 percent, which he said would mean people starving to the death on the streets or a huge jump in safety-net spending.

A $1 Trillion Build Meets a 1.4 Percent Quarter

Money is not waiting either. Goldman Sachs Research, in an August 2026 note by Joseph Briggs, put global 2026 AI-related spending at 1 trillion dollars of AI-related investment, including $581 billion in the United States. That is capex, chips, data centers, and software, not a productivity print.

The same bank’s March 2023 note, “The Potentially Large Effects of Artificial Intelligence on Economic Growth,” by Joseph Briggs and Devesh Kodnani, is still the scare line everyone quotes. It put the exposure at the equivalent of 300 million full-time jobs worldwide, with up to 25 percent of current work able to be done by AI, about two-thirds of jobs in the United States and Europe exposed to some automation, and 18 percent of work globally. It also said widespread adoption could lift global GDP by 7 percent and raise U.S. labor productivity growth by just under 1.5 percentage points over 10 years.

THE MONEY AND THE PRINT

• The 2026 build: Goldman’s August forecast is $1 trillion of AI-related investment worldwide this calendar year, $581 billion of it in the United States.
• The 2023 jobs line: The bank’s earlier note put exposure at the equivalent of 300 million full-time jobs, not a count of layoffs already filed.
• The Q2 print: BLS recorded a 1.4 percent quarterly productivity rate and a 2.2 percent gain from a year earlier, with hours barely up.
• The labor share: Workers’ slice of output fell to 52.8 percent, a series low, while unit labor costs rose 1.2 percent in the quarter.

Oxford Economics, in a September 9, 2026 note, said AI’s mark has been clearer in investment than in realized productivity and put a baseline U.S. gain around 3.5 percent over the next decade. That is a large number in a textbook and a modest one next to Silicon Valley speeches. It sits closer to Strain’s “big deal, but maybe not as big” line than to a 300 million-job panic.

He Still Wants Productivity Growth of 5 Percent

The campus paper’s frame, that AI might be economically small, is not quite Strain’s own writing. In a March 26, 2026 essay for Project Syndicate, posted by AEI, he asked how much productivity growth a society should even want, then answered high.

I hope that the AI revolution ushers in substantial increases in trend productivity growth. The benefits of productivity growth as rapid as 5% or 6%, much faster than the 3% pace of the internet boom, would clearly outweigh the costs, given our system’s ability to absorb and address the consequent disruption.

Michael R. Strain, director of economic policy studies, American Enterprise Institute

In that essay he compared a 3 percent internet-boom pace, which he said would double American living standards in 24 years, with a 1.5 percent post-2008 pace that would take 47 years. He productivity growth as rapid as 5 percent or 6 percent would, in his view, be worth the mess because modern safety nets and schools can handle faster change than 18th-century Britain could.

That is not a man betting against AI. It is a man saying the national accounts have not caught the story yet, and that if they do catch it at 5 percent, the country should take the disruption. At the Georgetown event he refused the hype binary. “Is it overhyped? I think nobody really knows the answer to that question,” he said. “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. I don’t think any of those things are going to happen, but I think it will be a big deal.”

He put the technology below the domestication of plants and animals and below the printing press, which he said is hard to imagine the American Revolution without. “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.”

The Safety Net Case He Says Is Fading

The policy sting sits in that 40 percent hypothetical. If prime-age work really collapsed, Strain said, the country would either watch people starve or “really ramp up our safety net spending.” Because he does not see that collapse in unemployment or in the totals, he is also arguing against a crash program of new transfers sold as an AI emergency.

He offered a second channel that gets less airtime than layoffs. “AI will lower production costs, or it wouldn’t be used,” he said. Lower costs, he argued, put downward pressure on market prices. If that shows up, the consumer gain is quieter than a pink slip and easier to miss in a campus jobs conversation.

None of this settles whether the 300 million-job exposure table from 2023 was a useful stress test or a slogan. It does fix the present tense. Payrolls are still adding jobs. The unemployment rate is 4.1 percent. Output per hour is running at its long-run 2.1 percent pace, with a soft 1.4 percent quarter inside it. Young adults, by Pew’s count, are more sure than they were two years ago that AI means fewer jobs. Georgetown is paying undergraduates $2,500 to help professors rebuild courses around the same tools.

Applications for those fellowships close September 28, 2026. The national productivity release for the third quarter is scheduled for November 5. Strain’s own test is simple enough to repeat. When the aggregate file moves, the argument changes. Until it does, the campus is training students for a boom the government ledger has not printed.