When New Brunswick Member of the Legislative Assembly Bill Oliver stood before the chamber to address the intricacies of an advocacy office, he did not merely commit a standard rhetorical gaffe. He pulled back the heavy velvet curtain covering modern governance, exposing a quiet reality where elected officials outsource their thinking to silicon processors.
"Here is a more natural, flowing version of that section that reads like a legislative speech rather than a series of short points," Oliver read aloud into the legislative microphone.
The room kept moving. The broadcast rolled on. But online, the footage detonated.
This incident is not an isolated comedy sketch. It serves as a symptom of a profound institutional rot. Modern politicians no longer write their own words, and frequently, they do not even read them before stepping onto the floor. The reliance on large language models to draft policy arguments, constituent replies, and floor speeches has reached a saturation point. When the output of a language model is copied blindly into an official binder, governance ceases to be an act of human deliberation. It becomes automated theater.
The Mechanics of Lazy Legislation
To understand how an elected official ends up reading chatbot conversational filler into the parliamentary record, one must examine the workflow of a modern legislative office. Staff budgets are tight. Legislative calendars are relentless. Politicians face an endless demand for commentary on everything from complex budgetary oversight to local infrastructure.
Enter the automated drafting assistant.
A staffer feeds a few rough bullet points into an interface. The machine responds with paragraphs, often prefaced with conversational pleasantries or meta-commentary like instructions on tone. Under pressure, the staffer copies the entire response block without performing even a cursory scan. The text goes straight to the printer. It lands on the politician's desk minutes before they rise to speak.
The politician, trusting the apparatus behind them, glances at the page for the first time under the chamber lights. They see familiar political vocabulary. They begin to speak. Trust replaces verification.
When the machine's conversational throat-clearing emerges from the mouth of a lawmaker, the illusion shatters. Citizens are left staring at a stark truth. The laws governing their lives are increasingly authored by algorithms that have no stake in the outcome.
The Erosion of Democratic Authenticity
Rhetoric in public office was once valued for its personal friction. The process of writing a speech—drafting sentences, arguing with oneself over phrasing, weighing the political cost of a specific noun—forced politicians to engage directly with the substance of their arguments.
That friction is gone.
By outsourcing the syntactic structure of debate to automated models, politicians detach themselves from the cognitive labor of governing. If an elected representative does not bother to read a speech prior to delivery, they are not representing their constituents. They are acting as an organic reader piano.
The public backlash online following the New Brunswick episode captured a visceral public exhaustion. Voters understand that politicians use ghostwriters; that practice is centuries old. But a human ghostwriter possesses judgment, political intuition, and accountability. An algorithm possesses none of these traits. It optimizes for linguistic probability, stringing together platitudes that sound authoritative while remaining entirely weightless.
The Broader Institutional Danger
The fallout extends far beyond a viral clip on social media. As artificial intelligence tools become standard equipment within government offices, the baseline quality of legislative debate degrades. Nuanced, localized policy concerns are flattened into generic, globally averaged prose styles dictated by model training sets.
Legislative bodies exist to argue specific regional realities. When those arguments are processed through automated text generators, they absorb the bland homogenization characteristic of machine-written content.
Accountability evaporates in the cloud. If a poorly phrased policy or an erroneous factual claim slips into a speech via an automated tool, who is responsible? The software vendor? The exhausted junior staffer? The politician who treated their own legislative address like a teleprompter run-through?
The systemic vulnerability exposed by this incident suggests that public oversight mechanisms are wholly unprepared for the automated transformation of political communication. Lawmakers draft regulations to govern artificial intelligence in industries ranging from finance to healthcare, yet their own internal operations remain dependent on the exact technologies they seek to police.
The microphone remains active. The servers keep processing queries. Somewhere in a legislative annex, another prompt is running, waiting for a binder, waiting for a floor, waiting for someone to stand up and read it to the world.
Canadian Politician Gets Roasted for Seemingly Reading From AI Prompt
This video captures the immediate social media reaction and breakdown of the legislative session where the automated prompt was broadcast live into the record.
http://googleusercontent.com/youtube_content/1
