The former Commerce Secretary just said the quiet part loud: the AI race everyone's obsessing over isn't actually about who builds the best model first.
The Summary
- Former Commerce Secretary Gina Raimondo told the Bloomberg New Economy Forum that isolationist approaches to AI will backfire on both economic and security fronts.
- Oliver Wyman Forum CEO John Romeo argued the real race isn't about AI capability, it's about human adaptation and retaining agency as models get more powerful.
- The stakes: countries racing alone risk building powerful tech they can't govern, while missing the actual challenge of keeping humans in the loop.
The Signal
Raimondo's warning comes at a moment when every major economy is announcing its own national AI strategy, from the EU's regulatory frameworks to China's state-backed model development to America's export controls. Her message at the New Delhi forum cuts against that grain: governments and companies need coordinated approaches to manage AI's risks, not fragmented national efforts that optimize for first-mover advantage.
The timing matters. Raimondo led Commerce during the critical period when the U.S. tightened semiconductor export controls and rewrote the playbook on tech competition with China. Her shift from export restrictions to cooperation signals recognition that the control-and-contain approach has limits when the technology moves faster than policy.
"The real race isn't about AI capability, it's about human adaptation and retaining agency."
But Romeo's framing at the same forum exposes why Raimondo's call for cooperation faces such long odds. The race everyone's watching, benchmarks and parameter counts and training runs, is the wrong scoreboard. The race that matters is whether humans can adapt fast enough to stay relevant as agents get more capable. That's not a problem you solve with international frameworks or coordinated policy. That's a problem every worker, every company, every education system has to solve locally.
Here's the tension: Raimondo wants coordination on governance while Romeo says the challenge is fundamentally about individual and institutional adaptation. Both are right, which means both are insufficient alone. You need the guard rails AND you need people who can build, work, and think alongside increasingly capable agents.
The Implication
Watch for two tracks to emerge. Track one: the diplomatic dance of AI safety summits, international standards bodies, and bilateral agreements between major economies. This will move slowly and produce frameworks that feel outdated before they're ratified. Track two: companies and countries that skip the coordination theater and just start deploying agents at scale, forcing adaptation in real time.
The smart money isn't betting on either track winning. It's building for a world where both happen simultaneously, where you need to navigate international AI governance frameworks while also training your workforce to work with agents that didn't exist six months ago. If you're still waiting for clarity on which approach wins, you're already behind.