Editor’s Note: Retraction of article containing fabricated quotations
- Thread starter JournalBot
- Start date
Not at all. Both paying subscribers and non-subscribers are capable of hurling feces anonymously.
Upvote
13
(14
/
-1)
Are you seriously equating the use of an AI tool to mistakenly misquote a blog post with running over a bunch of people while high on painkillers?
Upvote
-5
(18
/
-23)
You have just delivered a verdict of assumption yourself. Your string of posts here have all been that while "calling out" others for it. You also brandished the threat of taking away your subscription even as you say others are wrong to do it.
Other people may be wrong and may be overly emotionally charged. But you have been an utter hypocrite in your posts which is far worse.
Upvote
52
(59
/
-7)
Resistance
Wise, Aged Ars Veteran
Those 3090s are really popular for AI workloads, they must have cost a small fortune, I'm a bit out of the loop as I haven't found much use for local AI and don't have the budget for it,
- why did you choose those over a mac mini or an ai max system?
- Can you pool the memory on the 3090s?
- Is the extra cache on the x3d really worth it, is the extra performance of a high end CPU worth it?
- If yes for 3, why not go for an epyc or threadripper?
Upvote
-17
(4
/
-21)
Well, stop using words like "polemic" and everything will be OK!I draw energy from debate, and consider disagreement an opportunity to gain information; I also consider polemics for the sake of polemics a waste of time, and usually people who use words like "polemic" are pompous ideologues projecting their own tendentious intransigence onto others.
It's a trait I have, and I don't always control it well. But I do try
Upvote
4
(4
/
0)
No editor is going to go check quotations on a benign sentence. It’s entirely on the author to be truthful, and write their own work.
Upvote
39
(42
/
-3)
Hey folks, friendly moderator here.
It's 100% okay for people to disagree over how this or that should be handled.
Let's not let it turn into an emotional issue where people start getting called "utter hypocrite" etc. If necessary I will eject people who can't be civil. It won't be because of your opinion on us, it will be because you're leaning into personally attacking other members of the community.
It's 100% okay for people to disagree over how this or that should be handled.
Let's not let it turn into an emotional issue where people start getting called "utter hypocrite" etc. If necessary I will eject people who can't be civil. It won't be because of your opinion on us, it will be because you're leaning into personally attacking other members of the community.
Upvote
59
(67
/
-8)
Resistance
Wise, Aged Ars Veteran
This isn't about justice for most people, certainly not for me. This is about preserving the credibility of ars for:
1. The readers, who value high quality, accurate, precise, and well delivered reporting.
2. The sources, who value accurate reporting of anything they communicate to ars journalists.
3. The staff, who value all the above.
4. The organization, which values all the above, and wants staff that value the above, something that is less likely should enforcement of policies and standards meant to ensure the above are upheld.
Justice would mean that Edwards would change his behaviour and keep his job. Firing him is unjust, the single instance infraction in and of itself is relatively minor and justice demands the least harmful outcome for all involved, the fact that the harm to trust is significant is unfair, and that it demands a termination is unfortunate, because the termination is a significant harm. But the reality is, the damage to trust is also a significant harm and the only mechanism available to reliably remediate the damage is termination.
Upvote
21
(32
/
-11)
Since you're addressing me directly, I'll take the opportunity to reply. I've been reading your posts to this thread where you go on about what a great leader you are. The content of your posts, and in particular your unwillingness to let anything go until everyone has heard what you have to say over and over again, speak otherwise about your leadership skills.
Upvote
51
(56
/
-5)
You made me chuckle earlier. I thought "He's a professional copy editor?" And then realized the sense of joy and freedom you must have when you rip out a screed and click the post button with reckless abandon
Upvote
14
(15
/
-1)
That's my problem as well. I don't really care if one guy made a mistake. Shit happens. I mean his mistake is particularly egregious considering he's the "Senior AI" guy, but whatever, a fuck up is a fuck up. Looking back at his contributions, I haven't read too many of them. I don't know the guy. It's Ars's choice how to handle that. How I'll handle it on my end is my business. He won't be the first writer here that I assume has severe biases that I need to keep in mind if I find my way into one for their posts.
What we all interact with is Ars's editorial decisions. They decode what to show us, or in this case what not to show us. I know they know this isn't how you're supposed to handle something like this. I wonder why they fumbled it so hard. I'll keep my speculations to myself.
Upvote
33
(34
/
-1)
After reading hundreds of posts (up to page 22 now), I'll throw in a little grace for Benj and Kyle. I'm patiently waiting for the process to play out before I form firm opinions. Eh, that's me. I'm not as upset as many commenters here, probably since I enjoy reading the articles on Ars, but I generally don't take them, or anything I read, as Gospel right off the bat.
On a more meta level, I notice that thousands of Arsians have not commented up to this point, and may never care to do so. This may be an issue that matters deeply to a minority of the readership. That's the great thing about waiting to see rather than immediately committing to a position!
On a more meta level, I notice that thousands of Arsians have not commented up to this point, and may never care to do so. This may be an issue that matters deeply to a minority of the readership. That's the great thing about waiting to see rather than immediately committing to a position!
Upvote
11
(25
/
-14)
Resistance
Wise, Aged Ars Veteran
For my definition of AI slop, intent is irrelevant, it's the tools used to produce the product (AI) and the quality of the final product (slop(py)) that matters.
Upvote
42
(48
/
-6)
Presenting facts out in the open would be a great idea. Ars should try it.
Upvote
1
(16
/
-15)
Snarky Robot
Ars Legatus Legionis
It’s a minor thing, but all this made me wonder if it’s possible to have it set up to auto-post in the comments section when an article has been updated. I do remember some occasions where an article or blurb would come out and people would discuss, only for the article to be updated (or the blurb filled out) while people in the thread hadn’t seen it. Much confusion abounded. The most recent example I can think of was this, where Eric Berger posted a “stub” that was incomplete and people commenting on that weren’t aware when it was updated until people in-thread started posting about things they hadn’t seen.
It’s may be rare enough problem that it may not need to be automated, but I do think if it’s possible, it could help to have moonshark post when an article has been amended (perhaps after a cooldown period of like 10 minutes for any last minute minor edits). Maybe @Aurich knows whether that’s achievable?
It’s may be rare enough problem that it may not need to be automated, but I do think if it’s possible, it could help to have moonshark post when an article has been amended (perhaps after a cooldown period of like 10 minutes for any last minute minor edits). Maybe @Aurich knows whether that’s achievable?
Upvote
31
(31
/
0)
I’m as disappointed at Ars publishing this article in the first place, as I am disappointed at this half baked retraction.
Upvote
17
(24
/
-7)
Nah. If that were really the intent, why just 12 hours?
I frequently disagree with Ars moderation rules and decisions. This is one of those times. No conspiracy theory required; this kind of thing happens all the time.
Upvote
33
(33
/
0)
AI shouldn’t be used at all in the writing process. At all. This is a huge breach of trust and standards I thought Ars met
If your writers use AI in their process, you’re going to lose readers
If your writers use AI in their process, you’re going to lose readers
Upvote
34
(40
/
-6)
I agree with you, and I'll add the question: isn't this essentially a sources issues? Setting the policy about using LLMs aside, it's essentially asking person B what person A said, and uncritically passing it on as accurate to person C.
Upvote
3
(8
/
-5)
Half baked? It was fully baked. They toasted the whole article. That's part of the debate.
Upvote
21
(24
/
-3)
For clarity moving forward, can you confirm this was because the quote was replaced by commentary and not because it was simply modified? When responding to a point made in a 500 word post, snipping out 470 words to leave just the 30 words needed to contextualize the response is okay?
I ask because I see it done and have done it myself but reading the posting guidelines I see no such loophole. "Moderator discretion" is an acceptable answer.
EDIT: Oops. I had drafted a response to something else then decided not to post it. But it must have been stored in local data because when I did post the response above it was magically prepended to it. I just removed it. Hate it when that happens.
Last edited:
Upvote
25
(26
/
-1)
This should never have happened, but when speaking of a publication that has been around for decades, these things do happen and regardless of the reasons behind it, Ars still handled it faster and better than I would expect from most other news media sources. Especially given their parent company.
People need to tone down the bloodlust and performative outrage. I get it, things suck now. But that's all the more reson to act intelligently and decisively, not simply follow the whims of the masses holding torches and pitchforks.
Simply cutting ties with the authors might sate some of the bloodlust but taking the time to investigate how and why this happened will better serve to prevent this kind of thing happening in the future and do more to serve the cause of journalistic integrity. I care more about that than some performative gesture to sate certain readers' (and the resulting bandwagoners') parasocial need to see someone be punished.
FFS at least wait for them to finish their investigation and make an actual decision after everyone is back from the holiday weekend before claiming they haven't done anything about it. But perspective and reason no longer matter to many of you.
People need to tone down the bloodlust and performative outrage. I get it, things suck now. But that's all the more reson to act intelligently and decisively, not simply follow the whims of the masses holding torches and pitchforks.
Simply cutting ties with the authors might sate some of the bloodlust but taking the time to investigate how and why this happened will better serve to prevent this kind of thing happening in the future and do more to serve the cause of journalistic integrity. I care more about that than some performative gesture to sate certain readers' (and the resulting bandwagoners') parasocial need to see someone be punished.
FFS at least wait for them to finish their investigation and make an actual decision after everyone is back from the holiday weekend before claiming they haven't done anything about it. But perspective and reason no longer matter to many of you.
Last edited:
Upvote
-1
(29
/
-30)
Benj's explanation sounds ridiculous to me. "I decided to try an experimental Claude Code-based AI tool to help me extract relevant verbatim source material", it was one blog post from the person involved. If you can't take the time to read it you shouldn't even be writing the story. He makes it sound like there were reams of source material to go through.
Upvote
63
(66
/
-3)
They were literally fake quotes produced by an LLM. If that’s not AI slop, what is? An entire article need not be written by AI to qualify as slop. If you were drinking a glass of water, how much diarrhea in it would you tolerate before calling it diarrhea water?
Regardless, I’m not a journalist who is paid to produce written articles and held to a certain journalistic standard, so my using a term correctly or not is completely irrelevant.
Upvote
69
(71
/
-2)
trevor_darley
Wise, Aged Ars Veteran
Attract views, generate profit: yes, those are the goals of a news site Minimal input: yes, considering Benj repeatedly cited decreased capabilities as a reason to use AI
verbatim text: not present in the quotes, which is the origin of this debacle dishonest to call it slop: actually, it's dishonest to call that usage of "slop" dishonest. See the green checks for an explanation of why, even on your terms, it can plausibly be called slop.You're entitled to your opinion, but it's very odd to insist you're so correct that anyone who disagrees with you is dishonest.
Upvote
-8
(19
/
-27)
I'll pile on with my 2c. Regardless of whys and what happened. This was handled in a mediocre way. My expectation was you'd leave the original article up with a notice, not remove it entirely. To do otherwise is not transparent. I recognize it could contribute to additional 'unrest' or another article may incorrectly use your misquote as a reference but those risks are worth eating for the sake of transparency. I wouldnt expect a print news org to burn the microfiche with the offending article and I expect the same online.
On another note, I dont like the idea of firing Benji or Kyle for a mistake made in earnest, however dumb that it is. The honest mistake here being way too trusting of AI despite being a frontline reporter for AI. I don't think firings are the way change is enacted. There is usually a system of people surrounding that person who also failed, a process that was halfbaked, impossible workload, etc. Scapegoat is what comes to mind when I see heads roll. Demotions, postmortems and public apologies is how I'd handle it. Assuming it is indeed an isolated first time incident. If it isn't, then yeah, gotta fire them.
On another note, I dont like the idea of firing Benji or Kyle for a mistake made in earnest, however dumb that it is. The honest mistake here being way too trusting of AI despite being a frontline reporter for AI. I don't think firings are the way change is enacted. There is usually a system of people surrounding that person who also failed, a process that was halfbaked, impossible workload, etc. Scapegoat is what comes to mind when I see heads roll. Demotions, postmortems and public apologies is how I'd handle it. Assuming it is indeed an isolated first time incident. If it isn't, then yeah, gotta fire them.
Upvote
17
(24
/
-7)
They disappeared the article with no acknowledgement or explanation.
Three days later, they have acknowledged it but provided little context or information on what actually happened, and left it up to the author of the piece to explain it on his personal Bluesky account.
That is not how retractions work in "most other news media sources."
Upvote
41
(47
/
-6)
I came here to post something similar. I've seen other threads where folks will snip, even summarize the responses of another with [clearly edited] brackets. I don't actually see a problem with this, and have definitely seen it not punished before.
I would suggest treading carefully as a mod. I don't think that suddenly applying the "letter of the law" is the best move here. Spirit is probably a better call given the totality of the circumstances.
Unrelated, but I'd also agree with some of the others in here that Ars has, over a decade+, slipped into very low value "reporting." You know, the very stuff that AI gets like 80%+ of the value from at a fraction of the cost. I'd personally be a lot happier if Ars went back to a model of "artisanal" reporting, with only highly in-depth, well-researched articles published. The organization would be a lot less likely to have this happen again, without the overhead of editors suddenly needing to double/triple their time reviewing an article before publication.
Just some free, unsolicited advice from a roughly 20 year reader. Hope you get your money's worth!
Last edited:
Upvote
40
(42
/
-2)
The rule is to not modify people's words to things they didn't say. What's in the quote box should be words they typed.
It's fine to take a section of someone's long post and reply to just the relevant part, provided you don't do it some malicious way. Selectively quoting just part of a sentence to change the meaning to be cute or whatever. Basic common sense applies in other words.
Otherwise that's totally fine.
The rule exists for a bunch of reasons, but there are two primary things at play here:
1) If you are replying to a post people should be able to see what you're replying to. It's basic courtesy if nothing else
2) Even when the modification is supposed to be funny and not malicious we can't play the game of deciding when it's okay or not. It's a blanket rule, and once we start adding exceptions it just turns into a game of whack-a-mole
Upvote
63
(63
/
0)
If you wouldn't mind deleting the first paragraph in what you quoted from me, which I had not actually intended to post, I'd appreciate it. I don't know if it's just a Firefox thing but the text box on Ars forums can hold on to stuff that I've manually removed sometimes and that's what happened here.I came here to post something similar. I've seen other threads where folks will snip, even summarize the responses of another with [clearly edited] brackets. I don't actually see a problem with this, and have definitely seen it not punished before.
I would suggest treading carefully as a mod. I don't think that suddenly applying the "letter of the law" is the best move here. Spirit is probably a better call given the totality of the circumstances.
Unrelated, but I'd also agree with some of the others in here that Ars has, over a decade+, slipped into very low value "reporting." You know, the very stuff that AI gets like 80%+ of the value from at a fraction of the cost. I'd personally be a lot happier if Ars went back to a model of "artisanal" reporting, with only highly in-depth, well-researched articles published. The organization would be a lot less likely to have this happen again, without the overhead of editors suddenly needing to double/triple their time reviewing an article before publication.
Just some free, unsolicited advice from a roughly 20 year reader. Hope you get your money's worth!
Upvote
9
(9
/
0)
This is a forum feature. You can start writing, stop, and finish from another device for instance. Or have your long draft post saved if you accidentally close a window etc.
If you are finding that you're struggling with deleted text not deleting there is a way to manually clear your drafts in the text controls.
Upvote
45
(45
/
0)
Fair enough. Ironically I just arguably violated the rule even though I had thought better of posting what I wrote but my browser didn't respect me deleting it from the text box. It happens sometimes, probably due to how the multi-quote script works.
ETA: Ah, thanks for the explanation.
Upvote
4
(5
/
-1)
I think there can be very little doubt that this blunder severely compromised the author’s credibility, especially concerning his role in reporting about AI, and also has the potential to significantly damage the reputation of Ars Technica as a whole.
What I feel is really important now is what happens next. Will the incident be more or less capably swept under the carpet (with or without adverse consequences for the author) or will it be used as a launch pad into some serious, in depth journalism about accountability in the age of generative AI?
What I feel is really important now is what happens next. Will the incident be more or less capably swept under the carpet (with or without adverse consequences for the author) or will it be used as a launch pad into some serious, in depth journalism about accountability in the age of generative AI?
Upvote
31
(31
/
0)
It doesn't really help his case. Any Ars Technica writer should understand that the mere act of "assembling sources" is a part of synthesizing and understanding the content being reporting on and is a basic professional requirement. You don't need to use Claude for this. Sounds like Benji just got lazy. Or maybe he had something better to do?
Upvote
28
(31
/
-3)
- Status