I don’t think anything I’m going to say here is going to be particularly shocking or Earth-shattering to anyone, but I’m sort of frustrated and feel like saying it, and I have to do a Cold Start anyway, so I may as well put it out here: AI is making my job harder. I know that’s the opposite of what the conventional wisdom may be, but for those of us who give a brace of BMs about, you know, things that actually exist, AI is a plague.
Though, to be honest, the more I think about it, it’s not that AI itself is the plague; like all problems, it comes back to us humans. AI is just a tool, for better or worse. It’s the way humans are using AI that’s the real issue. I’ll explain.
I’m bringing this up because of an Instagram post I happened to see that caught my attention like a cat grabbing a dropped piece of sashimi. It was a post that was seemingly designed in a lab to get my attention: it was about a Brazilian Volkswagen, specifically a plan to update the Type 2 Bus in the late 1990s in an attempt to keep it competitive. Here’s the post:
View this post on Instagram
On the surface, it’s a fascinating concept, and one that feels extremely plausible. Brazil overall has a long history of taking deeply obsolete designs and grafting on more modern faces and other details in objectively hilarious ways to try and drag old cars into the present, like what they did with the Ford Falcon:
And Volkswagen Brazil certainly had plenty of experience modernizing VW hardware that traces back to the 1930s; take what they did with the Beetle chassis and drivetrain to transform it into a far more modern-seeming vehicle like the Brasilia:
So, the idea itself is certainly plausible. And VW absolutely did undertake styling studies to explore how the Type 2 Microbus design could be updated at various times; look at these, for example:
This concept was known as “Heckmotor mit Lufteinlässen an der Front,” which just means it has an engine in back and sucks in air at the front. It seems to be a mid-’60s design study. The compartment at the front – presumably for the spare wheel – is interesting, too.
This 1975 update clearly prefigures the Vanagon-type design that the Type 2 would become. I show you these because it just reinforces that what this AI-generated post is something that is, fundamentally, possible.
So, when I first saw the post, I was excited; a 1997 project to update the old buses that Brazil was still producing? I’m fascinated; this is catnip.
And they look like something VW Brazil may have done: just the front clip was modified, with new faces that incorporated more modern lighting equipment and a bonded-type windshield – as opposed to the old held-in-with-rubber-grommet style – that a ’90s car would have. They seem plausible!
But then I looked closer, and saw the expected AI tells in the small text and logos:
Yep, that’s AI. What about the other one?
Crap. AI, too.
And the interior: this looks like what I would imagine a modernization attempt to the Type 2 interior would, in fact, look like:
Honestly, thank every deity for AI still being bad with text and logos, because those are still the quickest signs of AI bullshit.
It’s all very well done, and that’s the problem. Well, it’s a problem because nothing in the caption suggests it’s AI at all. Now, I am all for making shit up – I love “what-if”-type speculation, and we do a ton of that here on this site – I’ve done it before, and you’ve seen the great stuff The Bishop does here. But we always make it clear it’s fiction, we’re speculating and extrapolating and imagining. That’s fun.
Putting stuff out without the clarification that it’s AI is simply deceptive. It muddies the water of truth and fiction, and that’s just going to bite us in the ass. It already is.
It’s not AI’s fault that it’s weirdly good at creating hypothetical cars like this; the fault is how we use it, and that it’s being used to deceive instead of inform or entertain or spark thought. I hate that when I see anything online now that looks interesting, there’s that seed of doubt that all too often proves to be accurate. It used to be fun to stumble upon something you didn’t know before online, and now, well, who the hell knows what’s real or not.
Being real still matters. It always will. We will never use AI to deceive like this; when I want to deceive, I’ll do it by hand, like our ancestors did.
Stupid AI. I wanted these weird bus concepts to be real.
On the Telo Discord server, there was a rash of AI images for a bit, including some from a person associated with the brand. He took the feedback, but it was wild to me that he didn’t see the problem with images that didn’t properly reflect the product and gave misleading visual cues as to the size of the vehicle.
On the plus side, not only did he stop, but the people using AI to slap liveries onto the pickup seem to have stopped, too. Reminding people that it is not only wasteful, but also incorrect, seems to help. So this article is keeping up the good fight.
Damn it man! Where are the happy little clouds with Gonzo floating through?
I think the positive thing about AI being this terrible is that the start date for the Terminator future we were all raised to worry about is moving farther back by the day
My Facebook feed insists on showing me posts of beautiful redheads with perfect complexion and figures. Not porn, just “photos”. Of course they’re fake AI generated ( down to the freckles ) always with the background out of focus. What blows my mind is the comments shown are from people thinking these are real girls.
And no matter how many time I tell FB to hide this slop it keeps popping up. F:&@k AI.
I hate the AI writing style that has infected most every piece of content online, and I hate that the tells are going to become useless as AI gets better, and actual humans start mimicking the AI content.
It’s not x, it’s y.
Let that sink in.
It feels like we’re regressing back to the olden times when writers were paid by the word count, not by the number of book sales.
Also, “this is x dressed up as y.”
We’re in a phase where people think AI is going to help them sell stuff, and they’re blind to how offputting the slop is.
Like all the overlong sponsored Facebook posts in which every sentence is its own paragraph.
Every one.
No matter how long.
No matter how short.
It’s like they prompted AI to pretend it’s a veterinarian writing the saddest possible story of how all the cats were dying because they had the wrong kind of water fountain.
Tragic.
Preventable.
Wait, are we talking about cats, or…?
I was scratching my head at new “renders” of the new Jaguar sedan this morning, wondering if they were GenAI or not. I feel like I shouldn’t have to do that, and I hate having to always question what I’m seeing, but it had that glazed over, too perfect look, so here I am…
AI suxxxxxxxx
I wish those modern updates to the old cars were real though. I think modern facelifts of really old cars are cool when not done by AI
I’m with you on AI . . . YouTube is stinking with slop that the algorithm is trying to feed me.
…but 370k views from a YT account that is 3 months old and titled ‘Cars Through Time’ or ‘Real Time Vehicles24’ can’t be wrong, right?
But yeah – lots of Azerbaijani infographic slop and lazy Indonesian AI cartoons have proliferated recently. All it needs is a ‘AI generated’ icon, but that’s a step too far for YT.
ENGAGEMENT! BRAIN ROT! WASH & REPEAT
Yep, a simple little icon indicating AI content would make me as happy as getting a free taco.
But could AI accurately render an ‘AI generated’ icon?
Not just fake vehicles, AI is being used in the development of real cars and trucks.
The headlights look like NASCAR decals as well. The perspective changes correctly between shots, but each one looks very flat.
It’s definitely getting a little harder every day to tell though. Humans are incredibly good at spotting visual deception (so we don’t get eaten by the predator trying to hide in the bushes), so maybe we’ll stay ahead of the tide for a while. Eventually we’ll have to treat images like text, in that everything we see needs to have solid sources and citations corroborating it, otherwise it shouldn’t be believed.
Luckily, we already hold every piece of text to this standard, right?
PREACH BROTHER!
Leading candidate for the best username on the site BTW
You had me at ‘kvetch.’ 🙂
This is why I love this website. There are decades of incredible automotive anachronisms, we don’t need to generate fake crap to find the joy that already exists.
As new models are unknowingly trained on existing AI slop, the outputs will get progressively worse. We may already be in peak AI, and it’s all downhill from here.
Kind of related, I grew up being warned by my boomer parents and relatives that television would rot my brain and I should go outside and play. Now it seems many a boomer just sit around all day watching the “news” network talking heads as they feed them a steady stream of party affiliated propaganda, without any critical thinking about what they’re being told.
And I’m guessing many of them completely miss the irony.
Like this positive feedback loop in AI, the boomers seem to have lost the plot, and we’re all worse off for it.
My parents in 1996: “Don’t believe everything you see on the internet, honey.”
My parents in 2026: “FreedomEagle76 dot Facebook said that we’re all getting checks any day now.”
I have noticed that Gemini has been trained on reddit and thinks stuff on reddit is factual. We are EFFED.
I wonder what would happen if you trained AI on parodies of AI.
Not only can AI be used to (mostly, for now) convincingly lie, it also steals from other people’s work to “create” and the data centers used for generating it pollute the environment and are being built in areas with no input from the people who live there and would be most negatively impacted by them. And it’s taking away jobs.
This x1000
They should be charged $5 per kwh and $40 per gallon of water used, and also have a multi billion dollar liability insurance policy. They should also be required to hire a certain number of employees and pay them well. For example, pay at least 500 people at least $100k a year each, plus benefits.
The money raised would be put in a few places: a fund for the community, a fund to compensate people whose work was stolen, and a fund for unemployment and job training
You are right to be bothered by this. I reassured myself that “ at least it wasn’t a matter of life or death,” but then I sobered up. Yep, it’s bigger than life or death; it’s Brazilian VW styling history. Oh, the humanity!
Went to Wharton, was a good student, went there, went there, did this, build a fortune – i’m like the best AI you’ve ever seen. I’ve said this before and I’ll say it again, I can say anything and everything like no one has ever seen. I’m proud of my net worth. I’m not doing that to brag, because, you know what, I don’t have to brag.
Thank you for your attention to this matter.
The windshield wipers on the one with the round-ish headlights should be a dead giveaway.
The more I interact and use AI tools the more I think AI acts like a lazy teenager.
No kidding. Last year I would have considered myself a ‘late adopter’ of AI. Now, not so much.
It now seems like I have to question every answer Chat or Claude gives me now, and find that more and more it ignores parts of my questions and/or conversation. As in: the most important parts… as though it has latched onto the least important aspect of the conversation and would rather annoy you than answer the question.
EXACTLY like my 15-year-old boy.
Thanks for the supporting info. Gemini is doing the same and ends with some extension questions and vailed flattery.
It’s like I told my boss, “if this program was an intern, we wouldn’t even let him finish out the summer, never mind have him back next year.”
Is that CPATCHA working for anyone else? I keep clicking the squares with the “Ford Falcon” in them, but nothing is happening.
“Click on the images of a lackadaisically assembled vehicle”
“Click on images of motorcycles”
Promptly shows a bunch of maxi-scooters and images with portions of motorcycles
What counts here?
So I’m not the only one – I feel better now. I’m often fuming at my screen. “It’s a Burgman…it isn’t a motorcycle damnit!”
What’s even worse is that in the near future we’ll look back on this era as the good old days when we could at least tell that something was AI generated.
Yeah, soon enough all these “tells” will be addressed and we’ll no longer be able to discern what’s real or not. I hate it so much.