Backstory here: https://www.404media.co/ars-technica-pulls-article-with-ai-fabricated-quotes-about-ai-generated-article/
Personally I think this is a good response. I hope they stay true to it in the future.
Backstory here: https://www.404media.co/ars-technica-pulls-article-with-ai-fabricated-quotes-about-ai-generated-article/
Personally I think this is a good response. I hope they stay true to it in the future.
Link to the archived version of the article in question.
I actually like the editor’s note. Instead of naming-and-shaming the author (Benj Edwards), it’s blaming “Ars Technica”. It also claims they looked for further issues. It sounds surprisingly sincere for corporate apology.
Blaming AT as a whole is important because it acknowledges Edwards wasn’t the only one fucking it up. Whatever a journalist submits needs to be reviewed by at least a second person, exactly for this reason: to catch up dumb mistakes. Either this system is not in place or not working properly.
I do think Edwards is to blame but I wouldn’t go so far as saying he should be fired, unless he has a backstory of doing this sort of dumb shit. (AFAIK he doesn’t.) “People should be responsible for their tool usage” is not the same as “every infraction deserves capital punishment”; sometimes scolding is enough. I think @totally_human_emdash_user@piefed.blahaj.zone’s comment was spot on in this regard: he should’ve taken sick time off, but this would have cost him vacation time, and even being forced to make this choice is a systemic problem. So ultimately it falls on his employer (AT) again.
I agree with you. For better or worse, I have to imagine a lot of people who’s job relies on pumping out regular articles use LLMs to get the ball rolling. Which is what appears to have happened here.
Just to be clear, the article itself was written by him; he was just experimenting with an AI tool to extract quotes (because learning about AI tools is literally his job), and because he had COVID at the time he got mixed up and pasted paraphrased quotes rather than original quotes. (Arguably he should not have been experimenting with a new tool while sick, but I am willing to cut him some slack because he was probably not thinking clearly at the time.)
The serious thing here is actually not so much that he used an AI tool at some point in the process but that fabricated quotes ended up in a published article.
Thanks for clarifying, after I left that comment I realized I had the order of events reversed!