Every pundit in Ottawa and Toronto hyperventilated when a Canadian politician stood up in the legislature and read a prompt straight out of ChatGPT. The lazy consensus erupted across the usual commentary channels: look at the future arriving, politicians are embracing modern technology, AI is finally creeping into the sacred halls of governance.
They are entirely wrong.
Watching an elected official outsource their rhetoric to a stochastic parrot is not a sign of progress. It is a terrifying admission of intellectual surrender. When you let a model trained on median internet chatter draft your arguments, you aren't modernizing politics. You are hollowing it out.
I have watched consultants and tech lobbyists pitch generative text tools to government relations teams for years. They promise efficiency. They promise clarity. What they actually deliver is a terrifying homogenization of thought, flattening complex regional grievances into polite, mid-Atlantic corporate mush.
The Myth of the Natural Prompt
The specific grievance from the legislature speech was that the politician found the original bureaucratic phrasing stiff, turning to ChatGPT with a plea for a "more natural version."
Let us dismantle this worship of the conversational tone.
Language in politics was never meant to sound like a casual coffee chat. Legislation, debate, and parliamentary posturing are high-stakes theatrical mechanics designed to establish hard boundaries, legal liabilities, and ideological lines in the sand. When you ask a machine to make your speech sound "natural," you are asking it to strip away precision in favor of vibes.
ChatGPT does not understand the nuance of federal-provincial transfer payments, indigenous consultation rights, or trade exemptions. It understands token probabilities. It looks at millions of generic internet texts and spits out the most statistically average string of words possible.
Average is the exact opposite of leadership.
Imagine a scenario where every single member of parliament hands their writing duties over to the same three foundational models. You end up with a parliament of algorithmically aligned clones, all arguing with the exact same rhetorical cadence, using the exact same smooth transition words, arriving at the exact same toothless compromises.
We do not need more natural speeches. We need sharper ideas.
The Mechanics of Outsourced Thinking
The media treats AI text generation like a smarter pen. This is a fundamental category error. A pen is an extension of the hand; a language model is a substitute for the mind.
When a politician feeds a policy brief into a chatbot and asks it to rewrite it for public consumption, a black box performs a lossy compression of reality. The machine smooths out the sharp corners. It sands down the controversial edges. Why? Because the safety filters and alignment training built into these systems explicitly penalize friction, offense, and radical clarity.
AI models are optimized for pleasant consensus. Politics requires the management of bitter conflict.
If your political rhetoric reads like it was generated by a tool that also writes customer service apology emails for airlines, you have failed your constituents. You have taken complex human suffering, regional economic anxiety, and systemic friction, and you have run it through a blender of corporate neutrality.
What the Pundits Miss About Public Speaking
The conventional wisdom holds that politicians must sound relatable to survive the 15-second clip era. This drives the desperate scramble to adopt whatever tool promises to make them sound human.
The irony is thick enough to choke on. Politicians hire teams of speechwriters precisely because writing is thinking. The struggle to find the right word, to build a coherent argument, and to defend a contentious position forces the brain to test its own logic.
When you bypass that struggle, you bypass the cognitive stress test.
I’ve seen enterprise executives try this same trick. They feed quarterly earnings data into an LLM and publish the resulting shareholder letter without reading the underlying spreadsheet. Six months later, they are walking back strategic commitments because the AI hallucinated a growth narrative that the market fundamentals flatly rejected.
Now scale that up to the governance of a G7 nation.
The Real Danger of Algorithmic Governance
The danger is not that the robots are coming to take over. The danger is that humans are eagerly abandoning their own agency because thinking is hard and prompting is easy.
When a politician reads a ChatGPT prompt in a legislature, they are signaling a profound laziness. They are telling the public that the drafting of laws and the articulation of values are administrative chores to be automated away.
That is not innovation. That is abdication.
We are sleepwalking into an era where public policy is written by machines that have no skin in the game, read by politicians who refuse to read their own briefs, and consumed by an electorate too exhausted to notice the difference.
Stop celebrating this nonsense as forward-thinking. It is the ultimate intellectual retreat. Put down the prompt, pick up the pen, and try forming an original thought for once in your career.
