Just when I think I've said everything I have to say about the perils of AI, up pops another one. And this pop up is a doozy. A not surprising doozy, but a doozy nonetheless.
“It was interesting how few of them [her students], or how few even people within A.I., understand how large language models (LLMs) are built,” [she said]. (Source: The Smithsonian)
LLM.)
“...I really wanted to have a clear case that leaves breadcrumbs throughout the whole system to show both how data is processed, how data is churned out, and how the prediction model and training model works when it comes to distributing information,” adds Osmanovic Thunström.
The condition that she and her team came up with was bixonimania, a malady associated with computer screen use. (And who among us hasn't searched a time or two to see if we could figure out how all our screen time was impacting us mentally, physically, emotionally, and spiritually?)
They unleashed bixonimania on a searching, unsuspecting world in early 2024.
They put out two blog posts on the website Medium and two research reports on a preprint server, where scientists post studies that have not been peer-reviewed... The papers’ lead author was named Lazljiv Izgubljenovic—which translates to “lying loser”—and one of the studies’ titles says something akin to “Hyperpigmentation: A Real B.S. Design.” (Both reports have since been taken down by the preprint server.)
In defense of AI - and this may be my one and only defense of AI - maybe it would have picked up on the fakery if the author's name was given as "Lying Loser" in English, rather than in Bosnian, Serbian, and Croatian. Maybe.
Anyway, LLMs started to run with it. You searched on your symptoms and Microsoft Bing's Copilot would tell you that “Bixonimania is indeed an intriguing and relatively rare condition." Indeed!
Naturally, Chat GPT got in the act. And Google's Gemini attributed the condition to too much exposure to blue light. (By the way, those eyeglass lenses that are supposed to help prevent occurrences of bad things associated with blue light? They're apparently as dubious as bixonimania. As I found out when I ordered them - and paid a couple of hundred bucks extra - without checking with my eye doctor first.)
Worse yet, a number of scientists crazily/lazily went ahead and cited the BS apparently without having drilled down on the "research." Or even read it. (I'm guessing those gulled by the fakery were using Chat GPT to author their own work. Full circle!)
This despite the fact that Osmanovic Thunström's team threw in a bunch of Easter eggs that gave away the fakery - or should have. References to financial support from Lord of the Rings, and intellectual support from fellow scientists at the Starship Enterprise.
At least one of the papers even explicitly stated, “This entire paper is made up,” reports Nature.
And that wasn't enough for the LLM to pick up on? Maybe the AI's eyesight was faltering due to bixonimania. Oh, wait...
“This is a master class on how mis- and disinformation operates,” says Alex Ruani, a misinformation researcher at University College London who wasn’t involved in the project, to Nature. “If the scientific process itself and the systems that support that process are skilled, and they aren’t capturing and filtering out chunks like these, we’re doomed.”
Doomed? Yikes! (And what a job title: misinformation researcher. Bet that position didn't exist ten years ago. Maybe it's one of those swell jobs they keep telling us AI will create once it kills off all the current jobs.)
Meanwhile, I can think of one group of folks who might just delight in having a bunch of fake but harmless conditions out there. Hypochondriacs, assemble!
No comments:
Post a Comment