Someone might be on the verge of being caught using ChatGPT to half-ass their job again.
Yep, people are seeing hints that some of the batshittery of executive orders that Trump is conducting some form of depraved performance art from via scrawling his Big Name all over in front of alt-right pilled Elon-hyped enthusiastic crowds might in fact have been ‘inspired’ by your local friendly AI chatbot. Or, god forbid, Grok, which might explain a lot.
From Futurism:
…legal experts have called attention to some curious common threads: bizarre typos, formatting errors and oddities, and stilted language – familiar artifacts that have led to speculation that those who penned them might have turned to AI for help.
There’s the one about further ruining Alaska’s environment which lists 6 Public Land Orders, all of which are numbered 1.
“The weird typos and formatting errors could lead to confusion down the road,” Stern wrote of the bungled numbered list. “If the Secretary of the Interior invoked his authority under Section XV(1) of this order, which of the 6 different subsections labeled 1 would [he] mean? And which number controls when a subsection has two different ones?”
There’s the comedy skit that renames Gulf of Mexico to Gulf of America - you know, ‘problem solved’, for no known or imaginable problem - which it is alleged by some provides in its description of the location a book-report style description of the area right outta the ChatGPT-a-likes:
“I struggle to believe,” agreed Stern, responding to Melkonian, “that a human, let alone a lawyer, wrote this 7th-grade book report-style description of the Gulf.” (Indeed, when we asked ChatGPT for a “description of the importance of the Gulf of Mexico,” it hit almost all the same notes.)
And more:
Other orders feature questionable errors and structural choices. The order to withdraw America from the WHO, for instance, includes some inexplicably bolded punctuation, while others, like one effectively withdrawing the US from a global corporate tax deal, fail to maintain uniform formatting standards throughout.
Now I personally doubt that the reason we see so many senseless and oftentimes cruel orders emanating from the Arena of Nightmares is because a chatbot went wild rather than that a small pustule of x.com-ravaged human brains sploooshed them out willy-nilly with minimal care taken. That they come from robot minds at least in intent would be most likely nothing but wishful thinking, although it’s very feasible that the average “knowledge worker” of course uses this technology as one of many tools.
The article does freely admit that there’s no way to know for sure from just looking at this text.
In a murky digital world, it’s often hard to tell: is what I’m looking at AI-generated? Or is just poorly executed human work?
But perhaps the fact that these allegations are passing through people’s heads says a lot - both about the quality of output we’ve come to expect from current LLMs, and also the new US administration.
To that end, is it possible that the Trump administration’s newly-signed executive orders were all crafted by humans, sans AI? Sure. Either way, though, the initial expert reviews of the executive actions are in – and according to those, they’re weird and sloppy. And even if they’re not AI, they feel like AI.