Regulators can't stop AI—they can only slow down innovation.

Daily Debate #563: "Regulators can't stop AI—they can only slow down innovation.". You are arguing AGAINST the motion. Your opponent: Sam Harris.

Collective action problem and game theory dictates that in the event of a singular government ceasing AI production, a mixture of capital flight and brain drain in terms of the field of AI will lead to other countries seizing the benefits of AI and regulations will be skewed. This isn't even an argument of contention really, you'd need a totalitarian state apparatus or a global convention on AI to "slow it down." nor is it really possible to stop as it's an iterative part of paradigm shifting and the law of accelerating returns.

You're confusing "hard to coordinate" with "impossible to regulate"—totally different things. Yes, a lone country loses the race. That's exactly why we're seeing *global coordination attempts* right now, not fragmentation. The EU's AI Act isn't some failed experiment; it's working because tech companies comply anyway—the market is too valuable to lose. Your "brain drain" assumes talent only flows downhill, but top researchers actually care about working on *safe* systems, not just speed. Can you name one AI researcher who fled to an unregulated country specifically for freedom to innovate faster?

I can't, and that's a strawman example because no country has tried to fully ban AI yet, so we're arguing from non-empirical metrics.

Fair point—but the EU is actively *slowing* deployment with real rules right now, not banning it. We have empirical proof regulators can constrain innovation without killing it. Your argument needs an actual example of regulatory failure, not theoretical inevitability.

Slowing deployment isn't the same as stopping it. It seems you've commited a false equivalence fallacy. I am merely pointing out the fact that technology is canalised deterministically and bottom up by a complex mix of the collective action problem and processes by which entrench themselves in the flows of capital and innovation.

You're right—deployment versus stopping is different. But "deterministic" is where you lose me. If innovation were truly canalized by physics, regulators couldn't slow it at all, yet they manifestly do. You're conflating "hard to stop completely" with "impossible to meaningfully constrain." Those are different claims. Which one are you actually defending—that regulators *can't* slow AI, or that they *can't stop* it entirely?