Opinion | This politician made an embarrassing blunder. The reaction from the masses gives me hope

By Justin Ling

“Public confidence in the office of an advocate matters,” said MLA Bill Oliver last month in the New Brunswick legislature. He continued: “Here’s a more natural, flowing version of that section that reads like a legislative speech rather than a series of short points.”

Oliver had unwittingly read some output from a large language model (LLM), perhaps ChatGPT. Like an actor mistakenly reading stage directions aloud, he powered through the speech — which opposed legislation establishing a public advocate for energy utilities — without missing a beat. Weeks went by, and Oliver’s oratory went unremarked upon.

That is, until a Reddit user posted a video clip to a New Brunswick subreddit. “Embarrassing” was the most charitable adjective the user deployed. Replies to the post were even more brutal.
“If there are no consequences for this, then we hold our high-schoolers to a higher standard than our elected officials,” one user wrote. “If he was a lawyer, he’d be sanctioned,” another chimed in. More comments rolled in: “amateur hour,” “idiotocracy,” “depressing,” “good grief.”
I was pleasantly surprised to see all these negative reactions to Oliver outsourcing his role to a robot. They expose, I think, a growing divide in our society: between the elites, who are only too happy to delegate their duties to the Borg; and the masses, who find this objectionable. The latter camp is correct — and the elites should take heed.

AI, we know, has done many wonderful things, from discovering new planets to revolutionizing medical imaging to improving web development. It will probably do many more. But it has also loosed a wave of lazy, low-value slop on the world.

Oliver’s vague, hand-wavy speech was laden with telltale signs of AI-generated pablum. When I fed the utilities advocate bill into ChatGPT and asked it to prepare a speech opposing the bill, tailored to sound just like Oliver, ChatGPT spat out a pretty similar homily. The AI-detection service Pangram puts the likelihood Oliver resorted to AI at 100 per cent. (An interview request sent to Oliver’s office went unanswered.)

I strongly doubt Oliver was the first politician in Canada to surreptitiously outsource his work to a chatbot. He was just the first to get caught.

AI advocates frame the technology as a path toward make us smarter: better read, better researched, more efficient, more eloquent. According to their logic, we should want parliamentarians to supercharge their research and write their speeches with AI. But this raises a disquieting question: If AI can do politicians’ jobs, why have politicians at all? (Wait, don’t start cheering yet.)
Despite the hype, LLMs have real drawbacks — limitations that make them particularly ill-suited to real-life debates on policy, politics and ethics. They hallucinate facts and figures and cite them with an air of confidence; they render important debates flatter and duller.

And because LLMs are trained on a corpus of existing knowledge, we shouldn’t look to them for novel policy ideas. What’s more, we need to be far more skeptical of their corporate masters. “How independent can an advocate truly be,” asked Oliver in his speech, “if the advocate’s resources ultimately depend upon the very government whose policies may occasionally require criticism?”

It was a facile criticism of the utilities advocate bill, but it makes a great point about the dangers of American-owned AI. Uploading our politics to social media has been disastrous, and things are unlikely to get better if we become dependent on Trump-aligned Silicon Valley oligarchs.

But even if these shortcomings were addressed, we’d still hate the idea of chatbots doing our politicians’ thinking for them. For decades, Canadian voters have been taking parliamentary democracy less seriously, and Canadian politicians have taken it less seriously in turn. That elected representatives will do only the bare minimum of critical thinking is now our expectation. No wonder they’re using ChatGPT to write speeches.

There’s an easy fix for this particular problem, one that’s been floated repeatedly for ages: ban pre-written speeches in the legislature. Force MPs to argue on their feet, with only a few notes at their disposal. That would make reciting AI slop into Hansard more difficult.

Expecting our politicians to eschew AI in their day-to-day work is not Luddism. If they can’t do their job without recourse to ChatGPT, perhaps they shouldn’t have one.
Source: https://www.thestar.com/opinion/star-columnists/if-politicians-cant-get-the-job-done-without-chatgpt-maybe-they-shouldnt-have-the-job/article_ae1c5ff2-fc93-4f8e-8401-ab47a1e0b279.html