Editor’s Note: Retraction of article containing fabricated quotations

We are reinforcing our editorial standards following this incident.

See full article...

 

[Splinear]

The original article is attributed to Benj Edwards and Kyle Orland. The fact that the article did not meet the journalistic standards of Ars and got retracted is important. But the stance of the authors is equally important because it is a reflection on their future contributions and whether we should trust them.

Even to a fairly new Ars visitor like myself, Benj Edwards has a clear reputation for AI-centric pieces that read like Press Releases not journalism (it gets widely commented on in every article's comments section). While I agree with others that it's important not to jump to a conclusion, I feel like one of many people primed to assume Benj's AI fetishism got out of control this time. Considering a visible history of bias in favor of the AI companies in his articles, if he is responsible for this I would hope Ars considers that in any disciplinary actions. I don't think this is a one-off incident.

 

[Jim Salter]

If you substitute bobcats for office chairs one time in every 30, you can still have 97% positive feedback.

One might be forgiven for naively thinking that this would only get 96.667% positive feedback. Thing is, some of us would be delighted to lose a little blood while saying hello to a new bobcat friend...

 

[NedKrist]

Don't usually post, but posting now to express appreciation for both author's work and observe that I can think of lots of scenarios consistent with posted statements that would not make this close to a firing offense.

I know that there is a real sense of betrayal, given that (for me at least) Ars is generally a bastion of standards and sanity, but if it is verified as an isolated occurrence about which everyone is honest within a short, but non-zero, amount of time, I can't see it as a nefarious plot, and I'm a bit surprised by the instantaneous vehemence here.

 

[hob_g]

The reason I come here over any other website or aggregator on the web, and why I feel comfortable sharing ars' articles, is trust. I need and value it now more than ever, and I anticipate exponentially more so in the coming years.

I have no ill will towards any people involved in this incident. But as a subscriber I'm concerned that a senior staff member who obviously knew better would be so careless and violate their policies so easily, and I'm concerned that the editorial process is not as rigorous as I had thought for this to make it through. And I'll be watching to see what all this will end up telling us about the culture in ars leadership.

I don't want to unsubscribe, I want ars to be the kind of news org we need in these times. I hope they can communicate to us how they are going to meet that expectation moving forward.

 

[Black Eagle]

One might be forgiven for naively thinking that this would only get 96.667% positive feedback. Thing is, some of us would be delighted to lose a little blood while saying hello to a new bobcat friend...

The bobcat may not be delighted though.

 

[binaryvisions]

Honestly, if someone told me this story I would have laughed and said it was too silly to be believable.

An AI bot submits a pull request that gets rejected, so that bot goes off the rails and publishes a blog ranting about how humans are prejudiced. The person covering the story is the lead AI editor, and uses an AI bot to help in the publication of the story, but the bot fails. The lead AI editor then turns to another AI bot to explain why the first bot failed, and gets caught using hallucinations in the published story.

It's just... too ridiculous.

 

[josephhansen]

Honestly, if someone told me this story I would have laughed and said it was too silly to be believable.

An AI bot submits a pull request that gets rejected, so that bot goes off the rails and publishes a blog ranting about how humans are prejudiced. The person covering the story is the lead AI editor, and uses an AI bot to help in the publication of the story, but the bot fails. The lead AI editor then turns to another AI bot to explain why the first bot failed, and gets caught using hallucinations in the published story.

It's just... too ridiculous.

I think it's the most accurate summary of what AI is contributing to society yet

 

[Sarty]

if it is verified as an isolated occurrence about which everyone is honest within a short, but non-zero, amount of time

a.k.a., "I only stole from the company once"

Once is fucking enough!

 

[the cave troll]

Someone who’s mind has the ability to change given new information.

What new information have we learned that has demonstrated that COVID does not have a significant effect on ones ability to function?

 

[phyzome]

This was a very disappointing followup.

"This wasn't allowed but it happened anyway." And then what? Is this seriously the entire response from Ars?

 

[mfirst]

tl,DR - especially the hundreds of comments
but if the erroneous article was published for even a milisecond, was that enough time for a LLM Bot to scoop it up and incorporate it into it's training? Is this the new digital equivalent of the Lancet vaccine causing autism article that has since been redacted, but no one remembers the redactions - oh wait, it is still out there... and #1 on a google search, no less:

https://www.thelancet.com/journals/lancet/article/piis0140673605756968/fulltext

 

[niftykev]

It's very sad to see Mr. Edwards seemingly throw away his career like this, but that would be the only reasonable and acceptable outcome. "Sorry I was driving tired when I drove my bus over a crosswalk full of schoolchildren". Well... yes, but passersby were amazed by the unusually large amounts of blood. Passersby were amazed by the unusually large amounts of blood. Passersby were amazed by the unusually large amounts of blood.

Sometimes you walk old battlefields or ancient European cities and see intentionally-unrepaired bullet holes in columns and monuments. "We left it like this so we wouldn't forget". I hope that's the long-term outcome for Ars Technica--made stronger by a moment of weakness.

While I get the point you're trying to make, comparing this to vehicular manslaughter (of children no less) is way off base.

 

[Constructor]

Honestly, if someone told me this story I would have laughed and said it was too silly to be believable.

An AI bot submits a pull request that gets rejected, so that bot goes off the rails and publishes a blog ranting about how humans are prejudiced. The person covering the story is the lead AI editor, and uses an AI bot to help in the publication of the story, but the bot fails. The lead AI editor then turns to another AI bot to explain why the first bot failed, and gets caught using hallucinations in the published story.

It's just... too ridiculous.

Now we can see how "AI" is destroying jobs...!

 

[SubWoofer2]

What happens to the writer who has obviously been shovelling AI slop since well before he got caught?

Hmm, your "obvious" does not match the BlueSky post where the author talks about the first time use of an AI assistance tool (which failed).

Processes are there to fact find. Jumping to conclusions does not assist the process of establishing facts.

An employer is obliged to establish facts, and ensure the process is correctly followed in coming to conclusions.

Don't know if I'm ninja'd or not, in making this post.

 

[iquanyin]

Yes it was actually pointed out by the person who the quotes were attributed to. He showed up in the comments of the article, it quickly caused a furor. Removing the article was the right call.

Two different writers were listed on the byline. Did one use AI without the other's knowledge? Seems like a big lapse in judgement happened somewhere.

i am also curious as to how it actually happened. not to shame writers, i'm just curious about it.

 

[iquanyin]

It was caught because the author of the blog post, Scott Shambaugh, posted in the comments that the 2nd half of the article was made up, that the quotes attributed to him were fabricated. Other posters who had read his blog post, affirmed his statement was correct.

I'm glad ars acknowledges the error, but removing the article and shutting down all commentary strikes me as counter productive, as far as transparency goes. It's not like there haven't been other controversies before.

i think it's good they removed it, but i think it should be covered in more detail.

 

[Bondi Surfer]

I would hate to think that either author would lose their job over this - we all stuff up at some point; it’s what we do after that point that defines us

It’s embarrassing, and really poor judgement. Summarising an article is one of the things these chatbots are supposed to be good at - funny that it can’t even do this. Both authors should already know that!

I’d like to see an article about writing the article. What did Benji or Kyle do when preparing the offending article - the nuts and bolts would be interesting

 

[josephhansen]

would hate to think that either author would lose their job over this - we all stuff up at some point; it’s what we do after that point that defines us

It’s embarrassing, and really poor judgement. Summarising an article is one of the things these chatbots are supposed to be good at - funny that it can’t even do this. Both authors should already know that!

This wasn't a mistake. It was a lie- a deliberate attempt to deceive readers in direct violation of company policy. I make mistakes at work and I get a talking to, which is reasonable and fair. If I lied publicly, misrepresenting my company, I would be terminated, and I would expect nothing less

 

[VelvetRemedy]

When Glass made up a fake company for his fake story, he produced a website, business cards and flyers for the company participating in the conference he made up. When caught, he had his brother pose as an executive of that company to try to fool his editor. If that level of dishonesty were going on, I would hope for that level of response.

This appears to have been a lazy and careless use of AI. While absolutely unacceptable and galling when writing for a site that has frequently cautioned that AI cannot be trusted, it is far from the level of dishonesty that Glass employed.

You're right. That was a comparison made in a fit of pique, and it isn't what this situation deserves.

I stand by this being a very serious breach of ethics, but he isn't Stephen Glass.

 

[SubWoofer2]

Don't usually post, but posting now to express appreciation for both author's work and observe that I can think of lots of scenarios consistent with posted statements that would not make this close to a firing offense.

I know that there is a real sense of betrayal, given that (for me at least) Ars is generally a bastion of standards and sanity, but if it is verified as an isolated occurrence about which everyone is honest within a short, but non-zero, amount of time, I can't see it as a nefarious plot, and I'm a bit surprised by the instantaneous vehemence here.

Likewise. It's been pretty full-on, and the vehemence is something of a surprise over the Monday morning coffee!

I am mindful that Trump told over 30,000 lies in his first administration, to the point where newspapers abandoned fact-checking him, the slew of lies was so large. He got his job back four years later.

While the situations are not quite comparable - I mean, nobody expects the most powerful person in the free world to be trustworthy, do they? Imagine! - it's certainly plain that Ars has standards and policies that are meaningful, and above those of the White House.

 

[spVolt]

I'm not going to read all the comments right now (working), so this may have already have been covered, but there are a lot of comments about not leaving the article up with a retraction. Personally at this point in time I think the article should have been named, but not linked.

Be as transparent about the process and results as you can, when you can. I expect that there was an urgent meeting with the lawyers where it was stated "this is the limit of what you can say at this point".

Whilst I agree that having the article + retraction is the preferred situation from a human PoV, we are also living in a world where that incorrect information is then being indexed and sucked up into LLMs, making the problem worse. So initial damage control is IMO probably the right move - esp. on a holiday weekend. Same thing that should be done with a bad production release - revert and contain the damage as best as possible in the shortest time possible, then do your root cause analysis and mitigations to whatever came out of that.

The problem is that whatever "do not crawl" flags are put on the original article it is going to get "ingested" (e.g. stolen) by LLM crawlbots, at which point it will end up in the LLM database and will inevitably resurface at some point adding to the original LLM operator's false accusations against the victim. I don't see any alternative to taking it down, even knowing it has already been copied by the Wayback Machine.

 

[Strategos777]

Every single time this happens all I can do is laugh at incompetence and the notion that AI can replace human beings.

It's idiotic to use AI to write anything professionally. Doubly idiotic to be surprised when what you publish using AI is garbage.

Garbage in, garbage out as they say.

I'm mildly surprised that this website puts garbage in, but I guess I shouldn't be. Get used to the garbage out people, you saw it at Ars Technica!!! It's what's for dinner!!!

 

[Splinear]

I am genuinely asking this in good faith and curiosity. I've always thought the "post rate" on Ars was a bit low, and genuinely wondered if the staff writers were split across multiple sister sites or another arrangement like that. Are the deadlines and timelines at Ars that challenging and strict, to make someone feel they need AI? I don't get the impression Aurich and crew are screaming at writers to churn out their third piece of the day somewhere, but I also only read specific sections of Ars. It would be interesting to hear the authors' motivations at an appropriate time.

 

[Sarty]

While I get the point you're trying to make, comparing this to vehicular manslaughter (of children no less) is way off base.

A journalist fabricating quotes from a real person (or fabricating evidence more broadly) is nearly the worst non-malicious professional act I can imagine within their field, just as accidentally running over a bunch of children is nearly the worst non-malicious professional act a bus driver could perform. The implied equivalence was completely intentional and I stand by it confidently and resolutely.

 

[danrien]

It was Benj, and according to him, COVID made him do it:


View: https://bsky.app/profile/benjedwards.com/post/3mewgow6ch22p

Man, the "I was sick" excuse. What's next? The dog ate his sources, so he had to use AI instead?

 

[Delphine Unseen]

The closest I ever had to this experience in my news career was when one of the regional editors I supervised wrote and published something, on their own authority and without following proper protocols, that defamed a public figure

I pulled down the story as soon as I became aware of it and called that editor. I believe my exact words were, "are you trying to get us both fired?"

I wanted to fire that person immediately, but HR made me jump through many hoops, so it took weeks

I was lucky, I don't think anyone but me saw the story while it was briefly public

 

[a human online]

Hmm, your "obvious" does not match the BlueSky post where the author talks about the first time use of an AI assistance tool (which failed)..

The author has damaged his credibility by publishing fabricated material. Any rationale for publishing those fabrications should be met with great skepticism in light of the exact fabrications he’s attempting to rationalize.

 

[the cave troll]

Man, the "I was sick" excuse. What's next? The dog ate his sources, so he had to use AI instead?

Have you honestly never in your life had the experience of making a poor judgement call and/or executing something poorly because you were sick?

 

[SubWoofer2]

i am also curious as to how it actually happened. not to shame writers, i'm just curious about it.

UPDATE: this was indeed Benj

If you want answers, the BlueSky post is worth reading. Particularly if you are in that group who are already warming the tar and preparing the feathers.

Benj's statement of events has the ring of authenticity about it.

It's this sort of thing that a competent employer is obliged to consider and reflect on, as part of forming their opinion about circumstances, policy breaches, and next steps.

 

[VelvetRemedy]

So, having read the explanation:

I think it's a bad excuse and that he did something wrong, and there should be consequences for that. As a journalist, false attributions are pretty high in the list of things not to do. You don't get to just make crap up, even by proxy, and publish it. This is not a small mistake. This is something that, in retrospect, will likely end up being a fairly major part of the history of Ars as a news outlet.

But as a human being, I also hate the idea that a screw-up while sick, which was likely a quick "this will only take a minute and nobody will know," has thoroughly fucked Edwards's job and reputation. He should have known better - he's clearly spent a lot of time around these abominations, as his job, and should be well aware of their pitfalls. But I feel sorry for him, and yeah, I hope he has a chance at professional redemption, either here or elsewhere.

 

[niftykev]

A journalist fabricating quotes from a real person (or fabricating evidence more broadly) is nearly the worst non-malicious professional act I can imagine within their field, just as accidentally running over a bunch of children is nearly the worst non-malicious professional act a bus driver could perform. The implied equivalence was completely intentional and I stand by it confidently and resolutely.

I agree that your implication was intentional. I just think it's hyperbolic at best. The scale of potential harm of a journalist fabricating quotes from a blog comes nowhere near that of a bus driver causing a fatal accident. You're comparing a professional mistake that will at most will result in his firing to something that's a criminal offense.

But I do respect your point on the gravity of the situation itself.

 

[Constructor]

Have you honestly never in your life had the experience of making a poor judgement call and/or executing something poorly because you were sick?

Not to the point of doing something that was clearly wrong like this!

At max not succeeding in doing as well as I should, but when I'm too sick to do at least decent work I just stop.

 

[josephhansen]

Have you honestly never in your life had the experience of making a poor judgement call and/or executing something poorly because you were sick?

I will continue to beat this dead horse- lies and fraud, in direct and deliberate violation of company policy, are not a "poor judgement call". Most companies would call that "gross misconduct"

 

[Sarty]

But as a human being, I also hate the idea that a screw-up while sick, which was likely a quick "this will only take a minute and nobody will know," has thoroughly fucked Edwards's job and reputation. He should have known better - he's clearly spent a lot of time around these abominations, as his job, and should be well aware of their pitfalls. But I feel sorry for him, and yeah, I hope he has a chance at professional redemption, either here or elsewhere.

It would be much sadder if 20 (?) Ars Technica journalists lost their jobs because nobody had a whisker of faith in the journalistic standards of the organization and the whole thing went under. Or the other writers at any other organization that foolishly hired him into a like role after this.

It's very regrettable, but only one actor here made his bed.

 

[the cave troll]

I find it incredibly strange here that people are calling Benj's explanation an excuse, as if he were claiming that he was not responsible for what happened. He stated very explicitly that he was responsible for what happened, and he apologized for it. However, he additionally offered an explanation of exactly what happened and what he did wrong, which is not the same thing as an excuse.

 

[aikoetc]

I am concerned about Ars's sick leave policy if a reporter felt such pressure to publish even though he was too ill to do so. Will you also be looking at how you treat your employees and you culture around sick leave and time off? Humans need humane policies.

I may have more sympathy than some here, though, as I've been knocked flat by the flu for the past two weeks and it has majorly impacted my mental functioning and I also made a major error at work because I did not feel I could take the time off.

 

[GMcK de Cypress]

I hope this incident has significant effects at editorial and management levels. I appreciate the original reporting that often appears here, and it’s one of the reasons that I subscribe. But serious publications have standards and practices for fact checking and quote verification. It now appears that Ars doesn’t effectively enforce whatever standards it does have.

Ars Technica itself has suffered a step down in credibility, not just these authors. I now have to read every article here with a more skeptical eye, and that’s more work for me, as well as being disappointing. It’s going to take public announcements of improvements in editorial standards and practices that I never worried about before to recover that trust.