A story about an AI generated article contained fabricated, AI generated quotes.
Archived version: https://archive.is/20260215215759/https://www.404media.co/ars-technica-pulls-article-with-ai-fabricated-quotes-about-ai-generated-article/
A story about an AI generated article contained fabricated, AI generated quotes.
Archived version: https://archive.is/20260215215759/https://www.404media.co/ars-technica-pulls-article-with-ai-fabricated-quotes-about-ai-generated-article/
I wouldn’t go that far. The article was posted Friday afternoon, and blew up over the weekend. Once the problem was known, the article was taken down quickly. We’ll see what happens when the editorial staff is back in the office in Monday.
They already posted their response: https://arstechnica.com/staff/2026/02/editors-note-retraction-of-article-containing-fabricated-quotations/
EDIT: it’s the lack of acknowledgement that they didn’t discover it but the contributor had to go in and correct it, how they locked and deleted the article, etc. I was expecting a bit more tbh
Benj Edwards, one of the authors of the offending article, has posted an explanation, taking the blame and clearing his co-author.
So he used an AI tool to “organize” references and it hallucinated crap that made it into the human-written article because he never reviewed the output for accuracy.
This guy writes about AI for a living, he knows it hallucinates, and he even acknowledges the irony but never explains why he thought experimenting with AI was a good idea to begin with. Am I supposed to assume his judgment was impaired by being sick?
Thanks for linking this. I hope ars makes it more visible. I’ll have to take Benj’s word.
That’s the thing with trust, hard to build, easy to burn.
Ah, that’s new from this morning. Seems I was a few hours out of date.
I hope for more. If they don’t have something substantial very soon they’ve got some serious problems maintaining the standards they profess to have, and we all should question the validity of their content.