Re:MSN Publishes AI Generated Obituary for Former NBA Player
by hedora on Friday September 15, 2023 @03:04AM
(#37519518)
The unrelated New York Post article on that page is nearly as bad.
Are there different grades of syndicated articles now or something?
Edit: The Futurism article points out MSN fired all their editorial staff a few years ago. It’s not surprising people just shovel garbage into their feed, though I thought the New York Post had higher standards. (I never thought I’d read, let alone write, that sentence…)
Re:MSN Publishes AI Generated Obituary for Former NBA Player
by vintermann on Friday September 15, 2023 @04:43AM
(#37520186)
This isn't AI generated, it's synonym-spun.
It's an old spammer tactic, to take a real article from somewhere else and randomly replace words with (things they hope are) synonyms. "Dead" becomes "Useless", "played" becomes "performed", "passed" becomes "handed", "men's" becomes "males's" etc. It doesn't matter if it turns into garbage, as long as web spiders don't spot it and see it as original content. It's written to fool robots, not people.
I guess it could be that a language model spat out a synonym-spun article out of whole cloth. They are very familiar with the genre, probably because there's a lot of it in their training data.
But it could also be a hand-written article that got passed through a synonym spinner for some reason.
Re:MSN Publishes AI Generated Obituary for Former NBA Player
by ImAnAmateur on Friday September 15, 2023 @02:11AM
(#37519135)
What baffles me is how this was published. Which human at MSN didn't read this before putting it out there?
I feel sure in saying that someone had the chance to read it but didn't... but the idea that it was published automatically after being prompted is so bad that I'm struggling to find words for it. It's like having a workplace near-miss because someone set up industrial equipment wrong. It's like a comedy prequel to Terminator that shows how Skynet was born. It's like turning your bread to ash because you set your toaster to the wrong setting and stepped outside.
Re:MSN Publishes AI Generated Obituary for Former NBA Player
by rvz on Friday September 15, 2023 @02:22AM
(#37519202)
Oh dear.
If there were no humans that reviewed this article and it was generated and published automatically and unattended, then it shows that not only AI systems can't be trusted, you still need at least 1 journalist to supervise and review the AI generated page before it gets published.
This is why fully replacing all humans with AI in lets say journalism won't happen anytime soon with rubbish like this although for sure there will be less human journalists needed.
But even worse if a human reviewed it and published this anyway. But again we don't know for sure.
Re:MSN Publishes AI Generated Obituary for Former NBA Player
by AraceliHarker on Friday September 15, 2023 @02:37AM
(#37519308)
Even before ChatGPT, Microsoft often used Bing Translate for languages other than English. There also seemed to be no one to check the translations, so nonsensical expressions like the ones in this case were commonplace. Now, with the advent of ChatGPT, its influence has spread to English content as well.
Re:MSN Publishes AI Generated Obituary for Former NBA Player
by AraceliHarker on Friday September 15, 2023 @02:33AM
(#37519278)
Since the structure of the TMZ article and the MSN article is almost the same, I think the writer named Race Track used Copilot or ChatGPT to change the expression to plagiarize. As a result, the word "dead" was converted to "useless."
Re:MSN Publishes AI Generated Obituary for Former NBA Player
by perihelions on Friday September 15, 2023 @04:02AM
(#37519944)
There's so many levels of dehumanization here—the writer, the reader, the subject, all deeply disrespected. It reminds of Orwell's vision of 1984, where all of the writers and thinkers have been fired, and everything people read was garbage extruded by writing-machines. And that was just fine with the decision-makers, because the machines were ideologically safe (aligned?), and thinking coming out of the proleteriat class didn't matter anyway—it was, at most, just a nuisance to be managed. If your utility to society isn't in thinking, then (one might reason) society loses nothing if it erases your humanity entirely—erases your curiosities and hopes and aspirations and wonders and replaces with them with synthetics, printed to a pattern-design. And the machine patterns never change, because, the thought-world mapped out for your managed exploration is small and circumscribed.
Are there different grades of syndicated articles now or something?
Edit: The Futurism article points out MSN fired all their editorial staff a few years ago. It’s not surprising people just shovel garbage into their feed, though I thought the New York Post had higher standards. (I never thought I’d read, let alone write, that sentence…)
It's an old spammer tactic, to take a real article from somewhere else and randomly replace words with (things they hope are) synonyms. "Dead" becomes "Useless", "played" becomes "performed", "passed" becomes "handed", "men's" becomes "males's" etc. It doesn't matter if it turns into garbage, as long as web spiders don't spot it and see it as original content. It's written to fool robots, not people.
I guess it could be that a language model spat out a synonym-spun article out of whole cloth. They are very familiar with the genre, probably because there's a lot of it in their training data.
But it could also be a hand-written article that got passed through a synonym spinner for some reason.
This is good old-fashioned plagiarism, with a cheap twenty year old trick to fool plagiarism detectors.
I feel sure in saying that someone had the chance to read it but didn't... but the idea that it was published automatically after being prompted is so bad that I'm struggling to find words for it. It's like having a workplace near-miss because someone set up industrial equipment wrong. It's like a comedy prequel to Terminator that shows how Skynet was born. It's like turning your bread to ash because you set your toaster to the wrong setting and stepped outside.
If there were no humans that reviewed this article and it was generated and published automatically and unattended, then it shows that not only AI systems can't be trusted, you still need at least 1 journalist to supervise and review the AI generated page before it gets published.
This is why fully replacing all humans with AI in lets say journalism won't happen anytime soon with rubbish like this although for sure there will be less human journalists needed.
But even worse if a human reviewed it and published this anyway. But again we don't know for sure.
Edit: answer is in the article, kind of...:
"In this case, the content was generated through a combination of algorithmic techniques with human review, not a large language model or AI system."