There are definitely starting to be some cases where I don't pick up on something being AI-generated, but that is the most AI-generated-ass-pizza I've ever seen.
I work in marketing so I look at a LOT of stock photos. And every time I go, "Ah yeah AI garbage", i find the source date is from 1990s or something. So my paranoia is way up there now where I don't think I'm going to be able to tell in a few years.
I wonder if those dates are always accurate. Is the source date something searchable or likely to come up in results when a user searches for something "90s"? Maybe it really is AI garbage that's just tagged 90s.
all entirely whole pepperoni slices and far too many of them to be mass-market or chain pizza.. and no burnt bits anywhere. it wasn't even modeled after a real slice.
Also way too thick of crust for this style of pizza. The toppings-sauce-crust ration would be way off. Terrible to me, but some people really like bread so maybe they would be into it.
I’m kinda into pizza this way, sometimes. The frozen pizza I prefer has a thick crust kinda like this AI crust. But I 100% understand why people wouldn’t like it; I figure I’m in the minority. At least there’s enough of us that the grocery store sells a frozen pizza that I like.
Yeah that’s why I added the comment at the end. An ex really liked thick crust. You might also like that Boboli (?) make it yourself pizza. That crust is way too thick/bready for me.
I’m into a New York style pizza, which is pretty thin and usually a topping/sauce ratio I like.
Shimmer! Thats a great descriptive word for it. Its something i always notice and give away the game. Its what tells me i am looking at i before i have studied the image for weird errors.
Exactly. They love/hate xy so much, they just Start saying the most uninformed stuff.
For example „Linux is great because every Supercomputer runs on it”. I have so many arguments how this is an utter nonsensical argument ,starting with „so you think every computer on this world uses pretty much unique Hardware which only works with software written for itself and therefore requires a tailored costum Linux kernel?” and ending with „So most enterprise businesses use windows 7 to this day, so its the best OS ever right?”
Not that I amost die of too high blood pressure every single time I have to use a windows machine, but these kinds of arguments are just nonsensical and don't help anyone who's just hearing about linux
It’s a plausible trap. Depending on the architecture, the image decoder (that “sees”) is bolted onto main model as a more discrete part, and the image generator could be a totally different model. So internally, if it’s not ingesting the “response” image, it has no clue they’re the same.
Of course, we have no idea, because OpenAI is super closed :/
Of course, it's an image generated by AI that has been trained to generate images that can't be distinguished from reality. So of course the AI thinks it's real lol
Weirder in that it gets better at "photorealism" (textures, etc) but subjects might be nonsensical. Only teaching it how to avoid automated detection will not teach it to understand what scenes mean.
I believe most image generating models are too small (like only 4GB RAM). Deepseek R1 is 1.5TB ram (or half or quarter that at reduced precision) to get some semblance of "general knowledge". So to get the "semantics" of an image right, not just the "syntax" you'd need bigger models and probably more data describing images. Of course, do we really want that?
There are definitely starting to be some cases where I don't pick up on something being AI-generated, but that is the most AI-generated-ass-pizza I've ever seen.
Yeah, that AI "shimmer" makes it really obvious, but I have to admit that I've been tricked some recently.
I work in marketing so I look at a LOT of stock photos. And every time I go, "Ah yeah AI garbage", i find the source date is from 1990s or something. So my paranoia is way up there now where I don't think I'm going to be able to tell in a few years.
The models are trained with a lot of stock photos.
I wonder if those dates are always accurate. Is the source date something searchable or likely to come up in results when a user searches for something "90s"? Maybe it really is AI garbage that's just tagged 90s.
all entirely whole pepperoni slices and far too many of them to be mass-market or chain pizza.. and no burnt bits anywhere. it wasn't even modeled after a real slice.
Also way too thick of crust for this style of pizza. The toppings-sauce-crust ration would be way off. Terrible to me, but some people really like bread so maybe they would be into it.
I’m kinda into pizza this way, sometimes. The frozen pizza I prefer has a thick crust kinda like this AI crust. But I 100% understand why people wouldn’t like it; I figure I’m in the minority. At least there’s enough of us that the grocery store sells a frozen pizza that I like.
Yeah that’s why I added the comment at the end. An ex really liked thick crust. You might also like that Boboli (?) make it yourself pizza. That crust is way too thick/bready for me.
I’m into a New York style pizza, which is pretty thin and usually a topping/sauce ratio I like.
Shimmer! Thats a great descriptive word for it. Its something i always notice and give away the game. Its what tells me i am looking at i before i have studied the image for weird errors.
like a photo of a plastic replica run through 500% strength instagram filters
Tell me about it! I feel like I have to hyper analyze just about anything I post now and even then get fooled sometimes
What's an ass pizza?
These violent delights have violent ends
My reaction to the later seasons
Which AI is the swirly butthole?
I think it's all of them, now that you mention it.
Whoa. Mind blown. Does it stand for Anus Intelligence after all?
Mind=brown
It's a top down view of the vortex that your personal data is being sucked into
Cave Johnson would've liked a word about that.
The biggest giveaway here is ChatGPT answering any question with one word without being specifically told to do so.
This. I feel like a lot of AI criticism posts are faked
People be misinforming way more than AI itself
Exactly. They love/hate xy so much, they just Start saying the most uninformed stuff.
For example „Linux is great because every Supercomputer runs on it”. I have so many arguments how this is an utter nonsensical argument ,starting with „so you think every computer on this world uses pretty much unique Hardware which only works with software written for itself and therefore requires a tailored costum Linux kernel?” and ending with „So most enterprise businesses use windows 7 to this day, so its the best OS ever right?”
Not that I amost die of too high blood pressure every single time I have to use a windows machine, but these kinds of arguments are just nonsensical and don't help anyone who's just hearing about linux
I'd go back to 7 in a heartbeat.
That's so weird. Every time I start a chat I say
If I'm getting slop it rather be concise and funny.
What if they asked AI to generate a shitpost and this image is part of the end result...
I didn't ask for "plastic toy pizza", just an actual edible pizza.
It's so perfectly fake. To the point it becomes really obvious.
did you actually get it to generate a picture and then ask if the same picture is generated and it said no?
This is a LemmyShitpost, so, yes absolutely.
Indeed.
It’s a plausible trap. Depending on the architecture, the image decoder (that “sees”) is bolted onto main model as a more discrete part, and the image generator could be a totally different model. So internally, if it’s not ingesting the “response” image, it has no clue they’re the same.
Of course, we have no idea, because OpenAI is super closed :/
"Open" AI. Yeah, open my ass.
Insert Man Ray meme
Of course, it's an image generated by AI that has been trained to generate images that can't be distinguished from reality. So of course the AI thinks it's real lol
It's been trained to generate images that it thinks* can't be distinguished from reality
And if it could distinguish better, it could also generate better.
Not necessarily, but errors would be less obvious or weirder since it would spend more time in training
Weirder? Interesting, like how for example?
Weirder in that it gets better at "photorealism" (textures, etc) but subjects might be nonsensical. Only teaching it how to avoid automated detection will not teach it to understand what scenes mean.
I believe most image generating models are too small (like only 4GB RAM). Deepseek R1 is 1.5TB ram (or half or quarter that at reduced precision) to get some semblance of "general knowledge". So to get the "semantics" of an image right, not just the "syntax" you'd need bigger models and probably more data describing images. Of course, do we really want that?
Absolutely not
What is wrong with the cheese. Is this AI thing or does anyone has so fluid cheese it looks like orange goo?
The secret is the Elmer’s Glue 🙂
AI images have a tendency (much like video games) to look like moist greasy plastic
What door?