I came across an article today about a wild animal encounter in Poland: "Wild Video Captures Giant Bison Flattening Tourist, Refusing to Let Him Escape."
.jpg)
As the title suggests, a tourist got too close to a European bison, which proceeded to knock him flat. Several nearby people acted to save the man from additional pummeling. But while reading the article, it soon became apparent it was written by AI.
I've touched on my gripes with AI writing before (most recently here), but it seems AI has "improved" in the sense that it seldom uses the identifying "trigger" words it once did (unwavering, intricate, commendable, innovative, meticulous, notable, and versatile).
So how did I know the above-linked article was AI?
A couple of things. First, AI loves to use similes (comparing two things using "like" or "as"). From the article: "Video shared on social media shows the bison looming over the fallen tourist like a living wall of muscle, making it painfully clear who controlled the situation."
No journalist worth his salt would ever use a simile in a news article, at least nothing that florid.
Second, AI tries to tell the reader how to feel about the subject in question through the use of over-the-top adjectives. Examples from the article:
• "A tourist learned a brutal lesson..."
• "The terrifying encounter was captured on video as stunned witnesses scrambled to save the man..."
• "Without much warning, the hulking creature surged toward the man..."
• "...making it painfully clear who controlled the situation..."
• "Nature photographer Grzegorz Ryński recorded the frightening scene..."
Et cetera ad nauseam.
Now I'll give this article one benefit of the doubt: It might well have been translated from Polish into English. Had AI not been available for this purpose, likely the subject would never have hit an American news site, so there's that.
But sadly, AI is taking over the written word in so many categories, including journalism. Aside from concerns about factual accuracy (and I have no way of know what, if anything, was factually inaccurate about the tourist/bison piece), AI writing is just ... bad. I know that as soon as I see similes and unnecessary adjectives in a so-called news article, I shut down.
Part of this aversion to AI writing is obvious. I'm a writer. I can see the handwriting on the wall for my beloved craft. I'm grateful beyond words that both my nonfiction editors (the various magazines I write for) as well as my fiction editors (Harlequin et al.) have no intention of using AI-generated content anytime soon.
I hope.
"The Shadow that bred them can only mock, it cannot make: not real new things of its own.”
ReplyDeleteTolkien said it in a different context but it applies quite distinctly to ai in my opinion.
The major AI content producers have figured out the giveaways of AI-generated writing and are spending a LOT of time and money to train their models to some across as “normal”. I’m skeptical.
ReplyDeleteI've been doing some work in that vein. Once you are used to it, AI writing becomes very easy to point out. Dead giveaway to me are the one-sentence paragraphs.
DeleteA great deal of actual reporters with news agencies also use AI to write all their articles and it's very apparent in the wording. I live in south central Idaho where we had the shooting a little over a week ago, and the vast majority of all the articles you could find online about it were obviously completely or partially wrote using AI. The easiest way to tell was the elaborate descriptions of the amazing scenic sites around the area, they really played it up how picturesque the area is. Along with information that is completely wrong for the area, but when you search the internet it is stuff that the AI version tells you.
ReplyDeleteMaybe you just have bias here? You spot BAD AI writing and assume all AI writing is bad. Not realizing that "ok" AI writing propably passes you.
ReplyDeleteIn my country, our official media uses AI. News always say clearly: "This was made with AI, and checked by human reporter". Without that text, I propably couldn't tell, because those news are always decent news language.
Also, if someone burns his house with bad oil lamp usage, do you thinkg oil lamps are bad? No. Same with the news you quoted: AI writes the style you tell it to write. If one wants "neutral news text", it wont use "like", or exaggerate or anything.
Here, I asked ChatGTP to write that article short and newslike, avoiding similes:
"A tourist was injured after a European bison charged him in Poland’s Białowieża Forest.
The incident, which was recorded on video, occurred while the man was standing about 20 feet from the animal. The bison knocked him to the ground and prevented him from getting up when he tried to move.
Nature photographer Grzegorz Ryński witnessed the incident and used his vehicle as a barrier between the tourist and the bison. He encouraged the man to remain still and eventually helped him reach the vehicle and move away from the animal.
According to Ryński’s wife, the tourist suffered a deep cut to his calf and severe bruising and was taken to a hospital.
Ryński said the bison should not be blamed, noting that the animal was relatively calm compared with some others. Wildlife experts generally advise people to keep a safe distance from wild animals, including bison, which can react aggressively when they feel threatened."