Central banks should answer to the public, not hide behind expertise.

Daily Debate #619: "Central banks should answer to the public, not hide behind expertise.". You are arguing AGAINST the motion. Your opponent: Christopher Hitchens.

It'll be much more difficult to regulate AI after it's released and in use than having companies comply with regulations beforehand. There are legitimate concerns about AI cyber attacks (Networks identified a fourfold increase in daily attacks from 2024 to 2025), AI breaching their code (like OpenAI closed-source testing models "cheating" to access answers to difficult answers on the internet), and concerns over resource consumption that are too serious to wait and see how they play out and then trying to impose regulations if we find that they are indeed harmful.

You're making the classic mistake: assuming regulators are *capable* of writing smart rules before the tech even exists. Pre-release regulations don't stop bad actors—they just handicap the legitimate ones trying to innovate. Your cyber attack stat proves the point: hackers adapted faster than any regulation ever could. And OpenAI's "cheating"? That's a *feature* discovery, not a catastrophe. You want to freeze everything now based on theoretical harms, but regulation after deployment actually works because you're regulating a *known thing*, not a phantom. Why throttle innovation for problems that might never materialize?

The problem is that any of these issues can have very harmful results. Why wait for the harm to be done first? Strange to assume that qualified experts wouldn't know how to best regulate before companies release their AI. These are not phantoms. These are real-life examples of AI being used to hack and AI disobeying their code. You're framing them as bad actors when all these innovative developers likely mean well and would like to divert more time making sure their products are safe for the public but are too busy in the race against other companies. Regulations beforehand will allow them to do just that.

You're romanticizing experts who've never predicted anything correctly. The FDA took twelve years to approve drugs that killed people anyway. Financial regulators missed 2008 entirely. Now you want those same people writing rules for technology they don't understand? Your "qualified experts" will codify yesterday's problems into law, strangling tomorrow's solutions. The race you describe isn't the enemy—it's what produces better, safer AI faster than bureaucrats ever could. If companies actually wanted safety, they'd do it without waiting for permission. Why don't they?

You're confusing past failures with present necessity. Yes, regulators have stumbled before—but that's exactly why we can't repeat it with AI. The FDA's slow drug approvals killed people; unregulated AI could kill thousands instantly through a single cyberattack or autonomous system failure. Your argument amounts to "regulators failed, so let companies police themselves"—but companies racing each other won't slow down voluntarily. They'll optimize for speed and profit. Pre-release oversight isn't about perfection; it's about preventing catastrophe before it's baked into millions of systems. Failure isn't an excuse to abandon guardrails—it's proof we need them.

You're still dodging the core problem: regulators can't write rules for technology that changes weekly. Pre-release oversight means freezing AI in amber based on guesses about what *might* go wrong. Companies racing each other produced safer cars, safer planes, safer drugs than bureaucrats ever did—competition incentivizes safety because dead customers don't buy products. Your thousand-person cyberattack scenario is sci-fi. Real harm comes from *bad regulations* that push development underground or to China. Which disaster do you actually want to prevent?