AI Job Displacement: Why Raimondo Ties It to the China Race

AI job displacement is now part of the US-China AI race debate. On September 16, 2026, former Commerce Secretary Gina Raimondo said "you're not going to beat China if you have destabilizing unemployment." The data so far is mixed. There are no economy-wide job losses, but entry-level hiring in AI-exposed roles has dropped sharply, and that's where teams deploying AI have the most say.

Key Takeaways

• At POLITICO's Decoded Summit on September 16, 2026, Gina Raimondo argued the US can't beat China while facing "destabilizing unemployment," and Republican Eric Holcomb agreed.
• The two lead RAISE US, a nonprofit launched in June 2026 with more than $500 million secured toward a $1 billion goal to retrain workers for an AI economy.
• The data doesn't show economy-wide AI job losses yet, but Stanford researchers find a 19% employment gap for young workers in highly AI-exposed jobs.
• Losses cluster where AI substitutes for people's work; where AI complements workers, employment is flat or rising.
• For teams deploying AI, that finding is a design choice: build systems that complement staff, and protect the entry-level work that trains your next generation.

What Did Raimondo Say About AI Job Displacement?

That it's a problem for national strength, not just a social one. Raimondo was Joe Biden's Commerce secretary. She made the case on stage next to a Republican former governor, and they mostly agreed.

"People say all the time, 'We have to beat China.' Sign me up for that mission. I want to beat China too," Raimondo said, according to POLITICO's report. "You're not going to beat China if you have destabilizing unemployment. That is a weaker America which will lead to regulations."

Former Indiana Gov. Eric Holcomb framed it around people. "When we say, 'America needs to win this AI race,' I take that as 'Americans need to win this race,'" he said. He added that AI "has a huge potential to raise America up" if workers are "skilled and ready for the jobs of the future."

Raimondo also tied it to safety. Failing to resolve AI safety and security issues, she said, would stifle competition with China too. Her summary: "We have to do both — go as fast as we can, but protect people and workers in the process."

The argument lands in a very different spot from the one we covered last week. The White House rejected calls to slow AI down, saying "whoever wins AI wins." Raimondo isn't asking anyone to slow down. She's arguing that a race won at the cost of widespread job losses would produce a public backlash, and regulation, that ends up slowing everything anyway.

What Is RAISE US?

A nonpartisan nonprofit set up to retrain and redeploy workers as AI changes the job market. Raimondo and Holcomb launched it on June 25, 2026.

Here's what it has announced so far:

• Leadership: Raimondo is CEO and Holcomb is co-chair.
• Money: more than $500 million secured toward a $1 billion goal in multi-year commitments.
• Anchor partners: Amazon, Anthropic, Microsoft, and the OpenAI Foundation, plus more than two dozen other companies and philanthropies.
• States: initial partnerships with Arkansas, Connecticut, Maryland, and Utah.
• Programs: career navigation tools, apprenticeships, wage insurance pilots, and support for displaced workers starting businesses.
• Success measure: whether workers "land and keep good jobs," not how many people complete a course.

That last point is the one worth noticing. Retraining programs have a long history of counting enrollments rather than outcomes. Committing publicly to job placement and retention is a harder test to pass, and an easier one to check.

The honest read: there's a fair critique here too. Some coverage framed the effort as companies cutting jobs paying for the fix. The firms funding retraining are also building the tools that change the job market, so they benefit from a smoother transition and from less public backlash. That doesn't make the programs useless. It does mean the outcome numbers, once published, matter far more than the pledges.

How Much AI Job Displacement Has Actually Happened?

Less than the headlines suggest across the economy, and more than the averages show for young workers. Two respected research groups look at the same labor market from different angles.

Source	What it measured	Finding	What it means
Yale Budget Lab	How fast the overall occupational mix is changing	About 1% shift since 2022	No economy-wide disruption yet
Stanford Digital Economy Lab	Employment of workers aged 22 to 25 by AI exposure	About a 19% gap in highly exposed jobs by mid-2026	Entry-level hiring is taking the hit
Stanford Digital Economy Lab	Employment by how AI is used	Declines where AI substitutes, flat or rising where it complements	Design choices shape the outcome

The Yale Budget Lab tracks whether the mix of jobs across the economy is shifting in ways that match AI's arrival. Its finding so far is mostly stability. The occupational mix has moved about 1% since 2022, well below the shifts seen in the early years of computers and the internet.

The Stanford Digital Economy Lab zooms in on young workers. Its August 2026 update, by Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen, finds employment for workers aged 22 to 25 in highly AI-exposed jobs about 19% below where it would be had it matched less-exposed peers. Between November 2022 and June 2026, employment for that age group in the two most exposed groups of jobs fell about 11%, while it grew about 10% in the least exposed.

The honest read: both findings can be true at once. The overall job market can look steady while the entry door into specific careers quietly narrows. That pattern is easy to miss in unemployment figures, because young people who never get hired don't show up as layoffs.

Why Would AI Job Displacement Weaken the US Against China?

Raimondo's case has three parts. Each can be checked against what's already happening.

• Public support. AI needs data centers, power, permits, and political tolerance. If voters link AI to lost jobs, that support drains away quickly.
• Regulation. Raimondo warned that a weaker economy "will lead to regulations." Rules passed in response to a backlash tend to be blunter than rules planned in advance.
• Talent. Entry-level jobs are how people become experienced engineers, analysts, and managers. Cut the bottom rungs, and the senior talent pool thins out a few years later.

There's a real counterargument. The White House says speed is what wins, as we covered in our China AI slowdown analysis. In that view, fears about disruption shouldn't slow American companies down. Some economists add that past tech waves created more jobs than they destroyed over time, even when the change was painful.

The honest read: Raimondo and the White House disagree less than it first seems. Neither wants to slow AI development. The dispute is over whether the transition for workers is a sideshow or a risk to the race itself. Given how politically sensitive job losses are, the second view is at least worth planning for.

What Does AI Job Displacement Mean for Teams Deploying AI?

That the outcome isn't fixed, and some of it is in your hands. Stanford's finding that employment holds up where AI complements people, and falls where it substitutes, turns a policy debate into a design decision.

Every AI rollout means choosing which work gets automated and which gets supported. Those choices add up across thousands of companies. Here's how to make them on purpose:

• Default to complementing, not replacing. Build agents that draft, check, and gather information for people, with humans deciding. We cover the patterns in designing AI agents with human review loops.
• Protect the learning work. Junior staff learn by doing the routine tasks AI now handles well. If you automate all of it, give them other ways to build judgment, like reviewing AI output under supervision.
• Measure role change, not just headcount. Track what share of each role's time shifts to review, exceptions, and customer work. That tells you whether AI is making jobs better or hollowing them out.
• Budget for reskilling up front. Treat training as part of the rollout cost, not an afterthought. It's cheaper than rehiring, and it keeps the knowledge of your process in-house.
• Keep humans accountable for outcomes. An agent that acts without a responsible person creates legal and quality risk as well as workforce risk, as we discuss in AI governance.

How Do You Design AI That Complements Workers?

By deciding who owns each decision before you automate anything. Whether AI replaces or supports a role is usually settled at design time, not after launch.

A practical way to sort the work:

1. List the tasks in a role. Break the job into concrete steps, from gathering data to making calls to talking with customers.
2. Mark what needs judgment. Anything involving trade-offs, exceptions, relationships, or accountability stays with a person.
3. Automate the preparation. Let AI collect, summarize, draft, and check, so people spend their time on the marked tasks.
4. Add review gates. Put a human approval step wherever AI output reaches a customer, a system of record, or a financial decision.
5. Re-measure after 90 days. Check whether people spend more time on judgment work, or whether the role just shrank.

The honest read: this approach doesn't guarantee jobs are protected. Companies still make headcount decisions for business reasons. But it does change what AI adoption looks like inside a team, and Stanford's data suggests that the difference shows up in who stays employed.

What Should You Watch Next on AI Job Displacement?

Evidence, not speeches. The next year will show whether the entry-level squeeze spreads, stalls, or reverses. A few signals will tell you more than any summit panel.

• RAISE US outcome data. The group says it will judge itself on whether workers land and keep good jobs. Its first published results will show if that bar is being met.
• The next Stanford update. If the 19% gap for young workers keeps widening, the problem is growing. If it narrows, hiring may be adjusting.
• New graduate hiring this fall. Campus recruiting in software, finance, and customer support is an early warning for AI-exposed entry-level roles.
• State pilot results. Wage insurance and short-time work pilots in partner states will show which kinds of support actually help displaced workers.
• Policy moves in Washington. Watch whether workforce proposals get attached to AI bills, or stay separate from the race-with-China agenda.

The honest read: the debate is running ahead of the data. That's normal for a new technology, but it cuts both ways. Claims of mass job loss aren't supported yet, and neither are claims that nothing is changing. Teams that track their own role changes will have better answers than either side.

How Van Data Team Helps Teams Deploy AI Responsibly

We help teams roll out AI agents that make people better at their jobs. We don't want them to quietly remove the roles that train future experts. That means mapping tasks before automating them, adding review loops where judgment matters, and tracking how roles change over time.

AI job displacement is now a question for governors, CEOs, and national strategy. For the teams actually deploying AI, it's also a set of design choices made one workflow at a time. If you want help making them well, our work on human review loops for production AI agents and AI agent evaluation is a good place to start.