A high-up employee at one of The Autopian’s competitors recently told me about something bothering them: Their site publishes so many crap articles every month that they now have to spend significant portions of their days replying to car company PR reps complaining about all the misinformation being “written” by outside contributors paid pennies to churn out slop. This is the poisoned well of the internet that we’re all being forced to drink from, and it’s killing us all. Let’s talk about this, and also about a recent MotorTrend article admitting to the use of AI.
As the publisher of this site, I have to contend with the fact that, for all the work we do to be model citizens of the web (we don’t review products we don’t use, we try to pay a living wage, we don’t overwhelm the site with ads, we don’t use AI to write stories), we’re up against sites that pump slop into the ecosystem and surround that slop with as many ads as they can.
While not all of this slop is written by AI, my suspicion is that the writers of this crap often use artificial intelligence to create their articles. There was a recent Reddit thread accusing Jalopnik of using AI, and I replied in the piece that there’s no way the actual journalists there are behind this trend (we know many of the journalists; we think they’re great).
What seems to be happening in various parts of the industry is that individuals outside the normal editorial channels are being given the right to publish by the revenue side of the business, and these low-paid writers (often from abroad) are using AI to complete their articles.
One recent example from Jalopnik is “Here’s What Costco Customers Really Think About Its Wheels,” an article written by UmmeAimon Shabbir, a Pakistani writer whose LinkedIn profile describes her job as “creating engaging content for Google’s Spotlight.”
It’s a strange article that doesn’t seem to demonstrate any real awareness of how buying tires from Costco really works. The top comment asks: “Anybody remember when Jalopnik articles weren’t AI-generated slop?” Another example, from the same writer, describes how Colin Chapman created a car “with two chassis stacked together.” There’s no sourcing for any of the article, so it’s possible it’s based on this article by Huibert Mees, the suspension designer who actually looked at the cars in person in order to write about them.
The apparent slop isn’t isolated to Jalopnik. You might say the internet is in its “slop era.” Some of it’s AI; some of it’s human-made, designed to game the algorithm of the day. It’s hard to tell, and genuinely, that’s the problem. Take, for example, Motor1, which features Google-baiting articles like this one: “‘They Lying! That’s a Tactic:’ Woman Pays Car Note for Brand-New Kia Twice a Month. Then She Gets a Letter 5 Years Later” that are being called “ownership stories.”
Autoblog seems to have some of this, plus a lot of commerce pieces built around trying to sell you tools they don’t seem to have personally tested.
All of this falls under the banner of “slop.” These are articles either being written by a computer or very specifically for a computer (in this case, the one that decides which stories you see or don’t see), or both. It’s not about journalism or quality; it’s about cheaply producing enough content to make money. It’s a sign of how challenging this industry has become.
The only thing I can say positively about the sites doing this work is that they’re smart enough to try to hide it. These aren’t efforts championed by journalists; at best, these are desperate attempts at revenue designed to keep journalists employed.
Ed Loh, the Editor-in-Chief of MotorTrend, seems to be taking a different approach. He’s openly admitting to using AI in a recent post about his EV-focused podcast with Jonny Lieberman.
The podcast went on hiatus for a while but is back, and one of the big changes, he notes, is that the company is now a part of Hearst Magazines, which also publishes Car And Driver and Road & Track, as well as owns BringATrailer.
The other change?
Another big change is that we’re openly embracing artificial intelligence (AI). The InEVitable is, after all, a vodcast about the future of mobility, and as you no doubt have heard, ad nauseum, THE FUTURE IS AI. So, we decided to put that to the test in the most meta way possible, by actively using commonly available AI tools in the production of The InEVitable. Jonny and I remain your 100 percent human (I think) hosts, interviewing mostly human guests, so be not afraid. The main point of the podcast (the guests, our conversation and insights) are still human, it’s the supporting elements that will be AI-assisted.
For instance, parts of this article were written with the assistance of Hearst Magazine’s ChatGPT. We created the social media clip via an agentic AI video editor called OpusClip. The summary of this article used by Apple News was created via an AI tool in our content management system. We plan on using additional AI tools to assist with future episode descriptions, thumbnail image creation, and audio clean up. Why? For all the reasons AI tools are being touted—speed, efficiency, and experimentation.
What Loh doesn’t mention here is that his company’s new owner, Hearst Magazines, is part of a company that just sold the rights to its works to OpenAI, creator of ChatGPT.
Here’s a Hearst press release on the topic:
“Our partnership with OpenAI will help us evolve the future of magazine content,” said Hearst Magazines President Debi Chirichella. “This collaboration ensures that our high-quality writing and expertise, cultural and historical context and attribution and credibility are promoted as OpenAI’s products evolve.”
In exchange for money, Hearst seems to be agreeing to hand over all of its content to OpenAI (the creator of ChatGPT). This is the “suing for peace” version of dealing with AI, and I’m sympathetic to it. Maybe that’s what we’ll have no choice but to do, because some of these companies are training their models on our content already, and we’re not getting anything out of it.
The amounts Hearst got weren’t disclosed, but it’s assumed by Axios that Hearst is being “compensated millions” for partnering with OpenAI. This is all to say that this is the first time I’m seeing this from a Hearst Autos site, but I doubt it’ll be the last.
This seems bad. I’m not anti-AI, and in fact use AI tools in my own life (Perplexity and Gemini, in particular). When AI is used to take some of the toil out of creating art, that doesn’t bother me, but Loh conflating the creation of graphics and the writing of content with “audio clean up” seems disingenuous to me.
In fact, if you get to the bottom of the post, you’ll see it’s not just the above:
Editor’s Note: This article was written with the assistance of artificial intelligence (AI). Future articles in this series will also be AI-assisted.
Loh doesn’t outright say the article was written by AI, just that it was “assisted,” so we don’t know the degree to which robots are writing what you and I are reading. Perhaps we should applaud the transparency here: at least Loh is admitting to using it (though without the context of the OpenAI partnership). However, if we take what Loh is saying at face value, it does seem like he’s at the very least not using AI begrudgingly.
This makes sense given that in a previous piece on Motor Trend, he said:
“They don’t care who generated the illustration…They love it or they hate it. AI doesn’t come up. I think eventually, writing is going to get that way. [Readers] just want a cool story told, they don’t really care how.”
There’s more to the story, and to his credit, Loh mentions ethical concerns and says he does think humans will remain important, but that quote above bugs me because I don’t feel that way about what we write here at all. You readers really, really do care about who is writing the words you’re reading, and if you found out that it was a robot, you’d rightly feel betrayed.
I think I need to be clear about this: That will never happen here, and if you don’t think this is ok industry-wide, if you think it’s important that sites like this one exist, then you should become a member. Right now.
It used to be my belief that it doesn’t matter if all these sites create slop or use AI, because it’ll eventually differentiate our real work. I’m more bothered by this now because it’s immediately impacting how we make money.
All this slop is being created at an exponential rate and swarming the usual networks where our content is shared. These sites are also the ones most likely to be overwhelmed with ads. This means that not only is it harder for our content to be seen, but when it is seen that the value of the ads we show is lower because there’s a glut of inventory.
I went to a conference for publishers this week in Georgia, and there’s going to be more crap like this, not less, in the future. It’s InEVitable…
If we are going to survive the storm, we need more help. Please consider becoming a member if you haven’t, if only to show that there are still people who think good journalism is worth paying for.
Hat tip to Alberto for pointing this out!
I miss the old magazines from the 1970s and 1980s that had long articles that are high calibre and informative like Mercedes Streeter’s articles today.
Today, I lost interest in reading the “news clips” of Motor1, The Drive, and others.
Been reading since the site launched, but felt this was a good time to officially join the band wagon. Used to read MotorTrend all the time in high School, very sad
I had thought to myself “How bad could Jalopnik actually be” because I had basically abandoned it just before Autopian started, in between when Bradley was writing badly conceived Twitter hot takes disguised as articles (and picking fights with people in the comments if they questioned him) and when they brought him back for a spell to do the same thing a second time. Then I just checked.
Good lord that site is fucked now. Of the past dozen articles it looks like maybe two weren’t written by AI, and of them one of them was one of their horrendous listicles. The truly fascinating bit are the articles correcting misconceptions that no one was asking about but then the article just paraphrased a Wikipedia page that is questionable as it was. Might as well watch an iilluminaughtii video at that point.
Man the AI articles on Jalopnik these days will make you long for Bradley’s hot takes.
Even after becoming a member here at Autopian, I still frequented Jalopnik daily. But about 6 months ago, the slop over there just became overwhelming. I would poke through a dozen articles and only end up reading 1 or 2. I said enough was enough and i haven’t been back there since.
I still read several columns in Jalopnik but am greatly disappointed in some of the articles I see there.
I very quickly got to the point where I wouldn’t open any of his articles even if it was something that could have been of interest.
And then came the even worse slop and I am the same as you – I refuse to even open the site or any links that friends share.
I spent three weeks on an offshore rig last summer. There’s nothing to do except work, eat and sleep. Although 180 miles out in the gulf, they had pretty amazing wifi and it seemed like every single person was scrolling TikTok or going through Instagram / Facebook / whatever all the time. One guy was showing me pictures of his new girlfriend, very much enhanced using photo tools and filters. It’s a continuous series of enhanced images and short videos (that match a shortened attention span) flashing by at high speed. Reading an actual book must seem like living in slow motion when you’ve become habituated to such high speed “content”. Photography required knowledge of film and exposure and it was somewhat expensive and good results required real effort, talent and practice. Digital cameras changed all that and “democratized” photography – cheap, unlimited, easily manipulated, perfectly exposed. And the results were predictable. Make a list of all the things disrupted by that cell phone in your pocket: music, photography, information search, reading, data handling in general and to a massive extent.
You are a slave to you cell phones battery life.
Go outside. Leave your phone at home, you don’t need it – it needs you. No one is going to call you except spammers anyway. If it’s your mom, she’ll leave a message. Walk, ride your bike, look at clouds, chat with a human in person. Delete all your social media crap, including LinkedIn (which is useless). Yes, all of it. No YouTube, no TikTok. After a while you’ll wonder why you ever watched them. Learn a skill, be good at something, while you still can and before these vampires suck away all your time. Don’t feed the machine.
Or, just keep what you want and use it in moderation. It’s no different than any other addictive behavior or item.
And speak for yourself on the calls…I went for a walk yesterday with my wife and came back to five missed calls from work.
What you do drastically changes whether you need to be reachable 24/7 or not. The vast majority of jobs do not require a work phone, so it’s sound advice.
There’s a middle ground that works for me: I disabled youtube search history but subscribe to a few dozen channels from creators I appreciate. When they upload, it shows up on my subscriptions page. No more algorithm, no more suggestions for channels I’ve never heard of that may/may not be using AI to read me wikipedia pages.
The real problem with AI for you guys is that when people search for information that happens to be in an Autopian article, it will show up in an AI summary at the top of the search results, and there are better than even odds that the person will just read the summary and never click through to the site.
AI is going to kill what’s left of the internet as we know it because websites will eventually only get crawler traffic.
And, also, when someone searches “2025 Camry review” or whatever, google will either serve up one of the handful of reviews written by a human or one of the hundreds of AI slop posts plagiarizing their work. To the average person who just wants to know the MPG or trim levels, they will neither know or care (whether the article came from a human or whether the info it spits up is even real). Maybe one day there will be the equivalent of a blue check mark for content written by people.
The best part is that the parasite is going to kill its host and then there will be no way to get another host.
Journalists are already getting fired because their bosses want to save money by having AI pretend to do their jobs. I was a journalist for many years. I quit before the AI thing started because the salary was lousy. I went from barely scraping by to doing quite well.
Many of the journalists who lose their jobs are going to do the same thing because there are places where you make more working a register at McDonalds than you do as a journalist. And once AI runs out of content to plagiarize, and also once AI companies start charging what they actually need to charge in order to make a profit, all these organizations that were so eager to fire their $35,000/year writers are going to want to hire them back, only they won’t want to go back because life is much more profitable outside of the industry.
What a time to be alive…
Hearst Media can go fornicate with themselves. They ruined BaT, Roadkill and so much more.
One very solid reason I subbed to this site, is that the content is actually engaging, not some very basic we-know-it kind of news & articles, and its very readable that you all poured in hearts into the pieces you all have written here.
Hold the line. We’re here as an escape from the AI slop.
If someone didn’t care enough to write something, why would I care enough to read it? If someone didn’t care enough to make an illustration, why would I want to see it? If someone didn’t care enough to compose a song and put thought into it, why would I want to listen to it?
The humanity of any created work is everything. That’s what gives it meaning and integrity. It’s utterly worthless if it wasn’t created by a person, and computers will never be people.
Thank you Autopian for not subjecting us to AI slop. AI is not the future, and the crash can’t come soon enough.
This is exactly right. I read autopian posts because they contain actual information from real people. I don’t give a rats ass what AI model gleaned from reading millions of reddit shitposts and then presented to me as facts. AI is not intelligent. It’s just an algorithm created to generate summaries that are best described as a sped-up search result (in the case of google AI summaries), but the result isn’t always correct. Art generated by a machine is just slop pushed out to fuel an insatiable burn for stimulation.
I’m an architect and recently had a client for our firm send us hundreds of “photos” for inspiration for his house. The problem is, it was 99% AI images and most of those pictures look “real” to a layperson but an architect can spot the fakery very quickly and most of what is being shown is impossible for various reasons of structure, materials or space planning. It’s the first project we’ve had where there’s a prevalence of fake photos being sourced as inspiration (by the client, not us), and it’s tricky how to explain to people that they’ve been duped into believing in a fake reality.
AI sucks man.
I liked The Oatmeal’s take on AI “art”:
A cartoonist’s review of AI art – The Oatmeal
Just read that. I agree wholeheartedly with it, it is really beautifully put.
I’ve never once considered subscribing to news of any kind, I think I only ever bought two magazines that didn’t come with a toy (newsstands really knew how to sell me a weekly rubber dinosaur with a glorified pamphlet of factoids about it). Paywalled articles from the likes of WP or NYT were always met with ctrl+W and I’d start my search over.
The Autopian is an outlier, there’s a special sauce here that makes me gleefully part with my money. I suspect that sauce is goopy, pungent, more than a little bit carcinogenic, and oozing from at least one of the gaskets of each staffer here.
Renewed.
In the article above, the two links below point to the same thing. In other words, the second link is bad. Maybe an AI tool could identify (and fix) these issues prior to posting?:
Another example, from the same writer, describes how Colin Chapman created a car “with two chassis stacked together.” There’s no sourcing for any of the article, so it’s possible it’s based on this article by Huibert Mees, the suspension designer who actually looked at the cars in person in order to write about them.
If anyone can find the second article, please post as I’d like to read it
I think it’s referring to this one:
https://www.theautopian.com/how-lotus-revolutionized-formula-1-with-ground-effect-aerodynamics/
Tiktop AI stories are poison. There was a time when I valued reading Motor1. That time is over.
Having some oil in my blood from a young age, and now living in the Information Age, I get a fair number of various car “news” sources on my device article feeds.
I try to read something from all the sources (with The Autopian being my favorite). And I get caught by a baity title now and then and end up on some absolute nonsensical slop piece. After that happens twice (everyone gets a second chance) I tell my purveyor of daily stories to not recommend that source any more. Jalopnik is getting really close to being blocked, but I can usually identify the garbage by the title. Motor1 got blocked a while back.
I read an article (mostly) about Harley Davidson’s logo (someone sent it to me) and I got less than halfway through and realized it was fully AI generated slop. It was a Jalopnik article which didn’t surprise me. It was so terrible. Probably partially written by the same person who works the help desk a SiriusXM.
I appreciate what you are doing here, especially with respect to adds. But I disagree with your contention that using robots would be bad. With all due respect, the quality of the writing here is far from the best I have, or routinely, read. The style of this site strikes me as more (untrained writer) blog+, than professional journalism.
David Tracy even wrote once that he didn’t need trained journalists or authors on this site – only people with car enthusiasm. This was memorable to me, as Car and Driver (in perhaps the “Golden age” of car journalism) noted once that they could hire an endless stream of car enthusiast that wanted to drive and write about cars, but that finding ones with actual writing abilities was the task for the editorial staff – and they felt strongly that well-written, entertaining articles was their secret sauce.
Lately, I’ve noticed the quantity of articles on this site has increased, but not the quality. There are more articles on the site than I can read in a day, so I have to be selective. There are a number of writers I skip over, as the articles are too long (I’m all for in-depth articles, but editing an article to make it concise and focused rather than a James Joyce stream of consciousness piece is greatly appreciated). You can pick up a 25 year old C&D article today and enjoy it – because it is well written – even if the car they are writing about is of no interest today and the performance metrics were long ago surpassed.
It seems to me your best writer, an actual trained journalist, which showed, Beverly Braga, no longer writes for the site. I’ll also make special mention of Hubert Mees, who brings expertise I lack (unlike DTs articles which are more engineering 101 level vs Mee’s graduate level discussions) and which are clearly edited and reviewed (perhaps by him, or the editorial staff), unlike DT’s articles which are full of typos and too long. The Bishop’s articles are also great. And while Mercedes’s articles are impossibly long, if I have time, I will read them because they are so though and well-researched. And Matt’s TMD are something I always read, not for the writing, but for the expertise / insight.
In short, most of your writers would probably benefit from AI editing (or, better yet, real human editing but there is a substantial time and cost difference to get what is, perhaps, nearly the same end effect). What is the value proposition or secret sauce of this site? You mention community – and that is great – but if you could add “outstanding pieces that are well-written, professionally edited, and entertaining” that would be better. To paraphrase car and driver’s editor from so many years ago, it would be great if the vision statement of this site were to provide outstanding pieces that are well-written, professionally edited, informative and entertaining – that happen to be about cars. If AI can help achieve that, that would be wonderful
I agree with every word of this. Really nice comment.
I can kinda get behind most of what you said, except for the AI editing part. AI text is universally worthless, and every effort it makes to spice up your writing comes across as awkward and uncanny. AI articles are always making weird nonsensical analogies and clearly were not written by anything with an understanding of what it’s like to be human.
Yeah, the articles on here are long and some of the writers could stand to take a writing course, but I value the fact that it’s a real person writing about their experiences, or news from a perspective of someone who knows what they’re talking about, far more than I value perfect writing skills. If it gets dry (which it usually only does in the long copy-pasted quotes from other sources), I’ll just skim until I get to the meat of the article, and go back and re-read something if I miss any context.
So yeah, if they want to spice up their writing a bit, investing in sending their writers to writing courses would probably help, but I’d rather they do nothing than try to use AI. And Torch can keep doing whatever he wants, his weird nonsensical analogies are actually funny 🙂
Counterpoint, while I agree that editing is helpful, I believe that AI editing would strain out the personalities which is part of what makes the Autopian what it is. AI is based on a melding of every bit of what it has been fed, which takes personality right out of it.
Weekend AI generated filler slop. I’m suddenly seeing it everywhere in the last several months. What gives?
Low cost engagement farming, for filler as you said. Otherwise they might have to pay people to work on the weekend, or hire more journalists to write more content that can be spread through the weekend.
It’s a shame, because weekend gigs are often how early-career writers got their feet in the door and honed their skills.
The ghouls pushing for this slop simply don’t care. They don’t care if we get anything valuable from it. They don’t care about building a readership. They’re completely detached from journalism’s role in society. The herbs of the publishing world are as worthless and soulless as the slop they favor.
And the largest problem, people no longer care for the most part. Idiocracy is more a documentary by the day.
Reading ye olde lighting website these days is depressing. I know a lot of good folks still work there and write for them, but all the filler and “light scrape” articles (versus deep dive, is what I mean) really drown it all out.
Much of what I am convinced are AI generated or written articles have the byline of people that are not in North America. I believe the Jalopnik writers in NA are unionized, so I think much of this is financially driven, not only to get the ad clicks, but I’m sure those writers in other countries are not being comparably compensated.
As proof of authenticity, I would like a video linked to the bottom of Torch’s articles that shows him actually typing out his charming lunacy. I picture him in a dim room drinking warm Kool-Aid out of an old anti-freeze jug while kackling maniacally at his computer as he slams the keys.
That dim light is coming from an incandescent bulb inside a VW taillight mounted to the wall.
I use AI regularly–it’s part of my profession. (I do CS research.) I notice that the closer I am to knowing the content, the more likely I will discover errors, which happens the majority of the time (when I know the content well). This has made me very cautious when I know the content less well.
These are infrequently significant errors, but usually at the edges of the material or in minor details about meaning. But they matter.
I would rather read an article written by a human that had mistakes than an AI article at all. This is because those mistakes are more complex, assuming the person has tried to do an honest job. And they will generally at some point get noted in the comments and the person can then respond appropriately (either with a correction, apology, or further explanation).
I will say that I’m surprised Jalopnik is still alive given how many times they’ve been resold. It still has good content. It also has some really bad junk. You can usually tell just by looking at the title of the article.
Ugh, the difference between European and American semi trucks article they have on their was so bad and I thought it was AI slop.
They just published an article titled, “Why Boeing Is Building The F-47 Stealth Fighter Instead Of Lockheed Martin.”
In the article, they first claimed that Northrop Grumman builds the F-16 (they don’t), then they wimp out answer the title question and say, “It’s classified.”
Now, I don’t know much about the military industrial complex, but I do know the government asks for bids, and the lowest bidder wins. Boeing won this time. Yes, the program is classified, but they won because their design is cheaper. That’s it.
AI slop garbage over there. It’s pathetic.
I’m still upset autocorrect changed “there” to “their”. In all seriousness, I’m not surprised. The worst is, I keep getting recommendations for an AI bullshit site that does car renders. “What a modern ******* would look like.” I keep blocking them and they keep popping up. We’re using up lots of water and electricity for this garbage?
God that’s even more depressing: we’re racing to build more and more data centers that collectively are using more kW than most countries on earth, all in the service of making crap no one wants.
Not only that, it drives up the prices of water and electricity for the consumers. That water and land could be used for food production, housing, etc. I’m not one to stifle the growth of text but when it’s competing with people for resources, that’s a problem.
So much this. It terrifies me that people are taking AI answers from Google at face value given the number of times I know it was wrong. I also work in tech and they’re pushing AI tools on us very hard, and in my experience at least 90% of what I get back from AI is wrong to one degree or another. Some are blatant mistakes, but the more concerning ones are where it sounds plausible but has no basis in fact. AI is like a really effective liar – it surrounds the lie with truth to make it sound more believable.
the term AI researchers use is “confabulation”. Because it’s programmed to have an answer whether or not it actually knows the correct one.
My favorite was when it correctly identified that the question was not appropriate for this particular bot, then proceeded to answer it (wrongly) anyway.
They have admitted that they use standard human test writing methodology. An answer if you don’t have a clue is scored higher than no answer or admitting you don’t know. Sometimes works when a human is writing a test, but for AI it trains it to hallucinate.
AI sucks at technical writing. While the language and wording might be close to reality, it almost always misses critical information that an actual experienced person would pick up on immediately.
Motor1 recently had an article on how to replace a damaged rim. They have now updated it to attribute AutoZone, but I can’t find a single article on AutoZone that matches what they have on Motor1. It appears to be a conglomeration of about 3-4 articles. The important thing is that they say to park the vehicle on level ground, set the park brake and chock the wheels. Then remove the air from the tire and loosen the lug nuts (but don’t remove them). Then use tire irons to dislodge the tire from the rim, and then use your hands to finish removing the tire from the rim. Apparently at no point is there a need to jack up the car. It never actually mentions removing the lug nuts completely.
At some point some wanna be DIYer is going to follow those instructions and hurt themselves.
Mid to late boomer here (68). This is the only automotive website I subscribe to, and I only rarely read a review, usually of an old one of a car I used to own or wanted to own, anywhere else on the webs. All of those predate AI.
I was a newspaper reporter (waiting to get a photographer job, which I eventually got) fresh out of college, so I know (or knew) how the sausage was made.
I have a news aggregation app on my phone that I use daily. There are so many articles that feel like they were AI generated or written by people who really ought not to be in journalism.
I also subscribe to the NYT, Slate and WaPo, but I’m thinking of dropping the last due to the ethical issues I have with its owner vs. journalism.