Sean Thomas's piece in The Spectator (link below), captures the dizzying media circus following Dario Amodei's call to "pace the frontier" and slow down AI development. The scene carries a distinct absurdity: Silicon Valley CEOs lining up on Sunday talk shows to solemnly warn us about the apocalyptic power of the very engines they are building. The impulse to urge everyone to hit the pause button, step back, and actually think through safety and governance sounds completely reasonable on its face. Yet, the moment that high-minded plea collides with real-world geopolitical momentum, the entire "let's slow down" premise collapses under its own weight.
Asking the American AI sector to pause or moderate its pace completely ignores the global landscape. You cannot hit pause in a vacuum when China is aggressively subsidising its own AI infrastructure and pushing model capabilities forward with state-backed urgency. To Peking and Shenzhen, a voluntary Western slowdown isn't a sign of high moral reflection; it is a massive strategic gift. It offers a clear window to close the capability gap, establish global standards, and dominate dual-use military applications without needing to outpace Silicon Valley in real-time.
This reality isn't lost on Washington, which is why any call for restraint immediately meets a brick wall in American domestic politics. Political leaders and policymakers, particularly those aligned with an aggressive, competitive posture against foreign adversaries, view AI development through the lens of a zero-sum arms race. When the overarching mandate is to win a technological cold war, asking developers to slow down sounds less like prudent governance and more like unilateral disarmament. A pause in California doesn't mean a pause in Beijing; it just means changing who holds the lead.
This dynamic traps the "AI apocalypse alchemists" in a fundamental paradox. They are asking for a coordinated global brake on an engine that is inherently fuelled by competitive panic. While Amodei and his peers call for independent auditing and international pacing frameworks, the structural incentives of global power guarantee that no such brake can ever be applied evenly. So long as the alternative to moving too fast is letting a geopolitical rival win the race, the mandate will always be to ship models faster, train bigger clusters, and sort out the consequences later.
https://spectator.com/article/dario-amodei-and-the-ai-apocalypse-alchemists/