You know that sinking feeling when you realize the person in charge has no idea what they’re doing? Multiply it by a thousand. Because what happened in Canada’s Parliament this week isn’t just incompetence—it’s a quiet, terrifying coup against representative democracy itself.
Let me set the scene. A Canadian legislator stands up in the House of Commons to deliver a floor speech. But instead of arguing, debating, or representing constituents, they read aloud what is unmistakably an LLM-generated response. The telltale signs: formulaic phrasing, neutral tone, vacuous platitudes, and a complete absence of personal conviction or local reference. The most damning part? They didn’t even read it beforehand. One top comment on the initial report summed it up: “He didn’t even read it before hand? That should be immediate grounds for dismissal.”
Now, you might think this is a one-off embarrassment. A lazy staffer. A bad day. But look closer. This isn’t laziness. It’s a silent transfer of democratic authority. When a legislator reads an LLM output, the people aren’t being represented by their elected official or even a human staffer—they’re being represented by a weighted average of scraped internet data. The machine doesn’t know your town, your struggles, your hopes. It knows patterns. It predicts the next token. And that token is now law.
We’ve all felt the creeping dread of AI infiltrating our lives—art, writing, customer service. But this is different. This is the core of governance: the deliberation, the debate, the human judgment that democracy is built on. And it’s being outsourced to a chatbot. When a legislator reads an LLM output, the people aren’t being represented by their elected official, but by a weighted average of scraped internet data. That’s not representation. That’s automation.
You’ve probably noticed politicians seem more robotic lately. More scripted. Less willing to engage in real debate. But did you ever think they’d literally become robots? This is the logical endpoint of a political class that has abandoned critical thinking in favor of efficiency. Why wrestle with a complex policy when you can ask the AI to write a nice-sounding paragraph? Why take a stand when you can generate a perfectly neutral, non-offensive statement?
But here’s the twist: the problem isn’t just that politicians are lazy. It’s that we’ve normalized the idea that text generation is equivalent to thinking. We’ve accepted that a statistical model can produce something that looks like reasoned argument. And when the people who make our laws adopt that same mindset, we’ve lost the plot entirely. Democracy dies not with a bang, but with a hallucinated citation.
This is a crisis of accountability. A speechwriter is a human being who can be questioned, fired, or held to account. An LLM has no ethics, no conscience, no responsibility. When a politician reads a chatbot’s output, who do we blame? The machine? The staffer? The system? The answer is: no one. And that’s the point. It’s a perfect accountability vacuum, and power rushes into it.
So what do we do? We demand transparency. We push for laws that require disclosure of any AI-generated content in legislative proceedings. We force politicians to actually read, understand, and defend the words they speak. And we stop pretending that efficiency is a virtue when it comes at the cost of representation.
If we don’t act, the next speech you hear from your representative might not be theirs at all. It might be a machine’s. And you’ll never know. When you outsource your thinking, you outsource your power. And power, once given away, is rarely returned.
FAQ
Q: Is this really a big deal? Politicians have always used speechwriters.
A: Speechwriters are human beings with judgment, ethics, and accountability. An LLM has no understanding, no responsibility, and no conscience. It's a fundamental difference between a tool that assists deliberation and a tool that replaces it. When a machine writes the words, the human in the room becomes a mere mouthpiece, and the democratic contract is broken.
Q: What can we actually do about this?
A: Demand transparency. Push for laws that require disclosure of any AI-generated content in legislative speeches, reports, or communications. Support open-source AI detection tools. And most importantly, hold elected officials accountable: ask them directly if they wrote their own words. If they can't answer, they don't deserve your vote.
Q: Maybe LLMs write better than humans. Shouldn't we welcome the efficiency?
A: Efficiency without accountability is a recipe for disaster. The goal of representative democracy isn't efficient text generation—it's thoughtful deliberation, compromise, and human judgment. A machine can't represent you because it can't care about you. Prioritizing speed over substance is how we end up with laws written by statistical averages, not by the people we elected.