OpenAI’s new AI image generator is potent and bound to provoke
- Thread starter JournalBot
- Start date
And it's real easy to be 'snarky' on the internet than come up with actual solutions.
https://www.amazon.com/Factfulness-Reasons-World-Things-Better/dp/1250107814
Yeah this is a bit of a myth. The world has been getting better. This nostalgia based "I hate this timeline" nonsense is completely divorced from facts on the ground.
Now, does this mean that we won't ever have difficult times again? Of course not, Trump may indeed bring fascism and get us into WWIII but Pax Americana was never going to be eternal.
You think it's just openai now? Look at deepseek: even if we somehow manage to stop Google, Meta, and OpenAI, we can't stop China and the rest of the world.
The genie is out and all the snark in the world can't put it back into the bottle: either we adapt or we die.
I'm confident in our ability to adapt - well, I'm confident until we hit ASI at which point, all bets are off.
Everyone (rich and poor) is likely screwed at that point.
Upvote
-18
(3
/
-21)
I noticed this too, thanks for putting the pics side by side.
Yeah, changing the beard shouldn’t affect the background of the image (unless the beard blocks a bit more of it). But this demonstrates the weakness of AI art, that it’s constantly starting from scratch instead of modifying what’s already there. But that’s because there are no in-between or earlier stages, there’s just a final product.
Upvote
10
(11
/
-1)
Truth! A smart machine would ask you to clarify what you mean by “autoharp.” Maybe it’s a thing that exists in the real world, already documented? Maybe it’s a word you completely made up? Why you are wanting it is tells as much as the word itself. Theory of mind. Context awareness. Are you being serious? Or just wanting fun?
(Me, I had to #GoogleIt: “autoharp” is a real thing with a name that’s fantastical. Old joke: “Hurrah, we have achieved AGI!” “How do you know?” “The computer’s just asked for the back of an envelope.”)
But being a confident fool’s not the generators’ fault. It’s the programmers’, who lack imagination to consider their outputs could be wrong. It’s the brittle bright-eyed AI startup execs’, already visualizing their first IPO, their first mega-yacht. It’s the users’, delighted by fresh novelty, unpredictability, the chance their next click of the button might just yield exactly what they want.
All human bugs though. Significantly harder to solve.
Upvote
3
(6
/
-3)
Bit of a weird hypothetical.
1. Henson may not have to go to a studio - he can probably work on creating his own thing and maybe a YouTube channel (like those babyshark/Pink fong etc which are insanely popular and profitable among kids).
2. Even if Henson's muppet Benj wouldn't exist in that form. A creative mind like Henson would find someway to use this technology to do something truly creative with it. You're seriously underestimating human intelligence and ingenuity here.
Upvote
0
(2
/
-2)
Upvote
1
(1
/
0)
I think you must have linked the wrong image, I refuse to believe the original prompt was anything other than Jared Leto ‘Damaged’ Joker.No ChatGPT is needed to call out the cynics in the room. But if one were to ask where I draw my writing style from, I’d suggest trying to reverse-engineer the 'prompt' behind this image.
Upvote
1
(2
/
-1)
As someone who, several career changes ago, went to college for visual design - it is deeply saddening.
Design, at least in the semi-analog way I was taught it, emphasized critical thinking and visual problem solving that I don't see taught outside of graphic design courses and similar. I cannot stress enough how some of the foundational design courses I took years ago still influence my work for the better. This has held true between actual design roles, working in marketing, becoming a developer, and moving into IT.
It fucking sucks that soulless image generation will eventually snuff out human design roles, then the education pipeline for those roles, and finally the courses which teach design fundamentals.
Upvote
16
(16
/
0)
\>create image generator
\>show it off by generating something in Ghibli style
\>get mad that you’re trying to “cure cancer” when every product you’ve released is a bullshit generator and people use it to generate bullshit
\>whine about it in a green-text post on X
EDIT: ugh can't figure out how to do the > without it quoting
\>show it off by generating something in Ghibli style
\>get mad that you’re trying to “cure cancer” when every product you’ve released is a bullshit generator and people use it to generate bullshit
\>whine about it in a green-text post on X
EDIT: ugh can't figure out how to do the > without it quoting
Last edited:
Upvote
8
(10
/
-2)
Do you not actually see the expressions on their faces? It has created its own interpretation where the boyfriend has a clear expression at being caught by his girlfriend.
Now, when I prompted it to give me a version of the meme, it gave me one where the boyfriend does look distracted.
Again, it has actually got a bloody perspective on the whole thing and that is actually impressive.
If you remember, in the original meme, the boyfriend is confidently staring at the other girl, here the boyfriend is coyly sneaking a peek - which is actually closer to real life.
And you think all it does is "picture of two women and man standing in roughly these positions"
Last edited:
Upvote
-15
(2
/
-17)
And how do you know how these future jobs are going to be affected and how the workers will react? Do you have some oracle of doom that the rest of us don't have access to?
Upvote
-12
(0
/
-12)
No, it doesn’t. Computer programs don’t have a perspective on anything.Do you not actually see the expressions on their faces? It has created its own interpretation where the boyfriend has a clear expression at being caught by his girlfriend.
Now, when I prompted it to give me a version of the meme, it gave me one where the boyfriend does look distracted.
Again, it has its own take on how the boyfriend looks - (in the meme, the boyfriend is confidently staring at the girl, here the boyfriend is coyly sneaking a peek).
Again, it has actually got a bloody perspective on the whole thing. And you think all it does is "picture of two women and man standing in roughly these positions"
Upvote
8
(14
/
-6)
How long till Hollywood realises that if AI can generate still Images, scripts, and footage of deceased actors, the next step will be generating entire films in the style of Director X .. and shortly after that the film industry dies.
Upvote
3
(4
/
-1)
If you're independently wealthy, or doing art while you hold down another job, that may work. However, most great art that I'm aware of was created by artists who worked at it full time for their entire lives, supported by selling their output (or having wealthy patrons). That model is on its way out. If you can't support yourself as a professional artist, there will be a lot less incentive to pursue the arts, a lot fewer artists, and ultimately a hollowed-out culture controlled by the people who create the AI. There will be no more art for art's sake; the "best" art will be only what's the most popular, and makes the most profit, and AI will be the most effective at creating it..
Upvote
4
(6
/
-2)
The handwriting, to me, seems like a step backward. Overall, it's less natural, as if it was selected off the fonts menu in a word processor. It look like the text was handed off to another piece of software that rendered it in an appropriate typeface, then handed back to the image generator and pasted in.
Upvote
3
(3
/
0)
If I laid out flesh and bones into something shaped somewhat like a person I wouldn’t be making a real human, just like if someone makes a computer program that looks somewhat like it’s conscious, they’re not making a conscious computer program.
Upvote
3
(7
/
-4)
That image is from DALL-E, their old image generator. The free version of ChatGPT doesn't use the new generator yet. Here's what native GPT-4o makes:
Still a little janky, but much, much closer.
Upvote
8
(8
/
0)
Coming soon: influx of "businesses" that will take a $100 fee to not create an image of yourself you may not like in OpenAI
Upvote
3
(4
/
-1)
We're still going to need people on UBI sitting in their basements typing and looking at screens.
Upvote
4
(4
/
0)
Hooh-boy, ’cos if you thought that not-far-above-subsistence-living was some sort of panacea then Marx and Trotsky have a lifestyle to sell you. Pre-industrial revolution, that idyllic “own schedule” generally consisted of “all your life” with diptheria and “died in childbirth” as a fun bonus.
(Talking of which, there’s a significant body of art romanticizing this happier gentler past—no AI slop on these hands! As a genre it’s even coming back in fashion now. Et tu, Art?)
The one thing we really ought to all agree on is that Life Itself is Unfair and that making life better [for everybody] is a genuine Hard Problem. For anything else, we have Etsy now.
Last edited:
Upvote
-1
(4
/
-5)
Wait you just said 'or having wealthy patrons' - why can't they have wealthy patrons now?
Art was never really sustainable for a majority of artists throughout history. Even now, only certain arts can sustain you (film/television mainly).
There is a reason for the stereotype of the penniless artist - let's not pretend like art was ever really sustainable as a profession.
Upvote
-3
(6
/
-9)
Hoever, in the example image in the article (which the OP is referring to), the boyfriend is not even looking remotely closely at the other girl.Do you not actually see the expressions on their faces? It has created its own interpretation where the boyfriend has a clear expression at being caught by his girlfriend.
Now, when I prompted it to give me a version of the meme, it gave me one where the boyfriend does look distracted.
View attachment 106111
Again, it has actually got a bloody perspective on the whole thing and that is actually impressive.
If you remember, in the original meme, the boyfriend is confidently staring at the other girl, here the boyfriend is coyly sneaking a peek - which is actually closer to real life.
And you think all it does is "picture of two women and man standing in roughly these positions"
Upvote
6
(6
/
0)
Saying that Altman/OpenAI is playing the role of creative savior is excessively charitable--they're playing God. Altman observed how it went and listened to society and then ignored everyone and stole the internet's content. Even had the temerity to suggest it was necessary for them to do so. What a farce.
I think this is exactly right. 100% agree.
Upvote
7
(8
/
-1)
The beauty of Art: its interpretations.
Ex: https://www.thecollector.com/picasso-and-african-art/
See also:
View: https://www.youtube.com/watch?v=l9Aj7W3g1qo
Upvote
-2
(0
/
-2)
Two things:
1. In the example image in the article, it looks like the boyfriend is actually looking back nervously at his girlfriend (having been caught).
2. As my image shows, the model is perfectly capable of producing the right expressions - it isn't an inherent limitation.
Upvote
-12
(1
/
-13)
I still have very mixed feelings on the future of AI in creative and professional endeavors, but from this point on "raising the bar" I have to ask: How do people people get the practice and experience to become "exceptional" if all the demand for entry-level effort has been consumed by a "good enough" AI?
Where does the next generation of Senior Programmers and Senior Artists come from if there is no pipeline of Junior Programmers and Junior Artists?
Upvote
12
(12
/
0)
It's all wrong, though. Multiple vanishing points in different places. Humans figured this out a few hundred years ago.
Upvote
16
(16
/
0)
Upvote
8
(8
/
0)
What's the new version of the Monkey Wrench Gang gonna be called?
Upvote
3
(3
/
0)
I'm wondering how long it takes before the first person is convicted of a crime by virtue of evidence in the form of fabricated A.I. imagery or video. A.I. audio is it's own issue as well.
It feels like for every upside there are multiple downsides. It's disappointing, to say the least.
I was hoping for more of a Star Trek TNG future, what with the one piece jumpsuits and all, not some dystopian Snow Crash crypto bro fever dream. And yet, here we are.
Upvote
5
(6
/
-1)
Right - and may I know how you can ever tell the difference between a conscious person and a computer program that pretends to be?
Douglas Hofstadter, the godfather of cognitive science, who literally wrote the book on this recently admitted that he was wrong and that LLMs do have some sort of consciousness.
Now, for my part, I don't know (and I don't really care) if they do or they don't but I know that dismissing their output like I presume to understand the essence of consciousness is not the right answer.
Upvote
-12
(0
/
-12)
What's also notably absent is how much it costs openAI to A) develop and B) run this model. OpenAI spent $9 billion in 2024 to make $4b, a net loss of $5 billion. There is a good argument that a significant fraction of their costs is not just building the models, but running them also. That implies that they are losing money with each user, even those with paid subscriptions. Increasing the user base does not help that issue, only charging more or reducing costs dramatically would help.
My point is that you could pay armies of illustrators and copy editors for $5 billion a year, and those people certainly provide more original styles and content than any generative AI. There is no way that OpenAI will replace anybody unless they fix this problem.
Upvote
9
(10
/
-1)
Upvote
-11
(0
/
-11)
Upvote
-7
(1
/
-8)
You can care about this because it makes climate change much worse and enables Trumpism to spread propaganda
Upvote
7
(8
/
-1)
Upvote
2
(2
/
0)
Err, what?
I suggest you re-read the text that you quoted. It says that the safeguards are still in place, not that anybody has set fire to them.
Upvote
-5
(0
/
-5)
One of the biggest horrors is that it destroys-by-proxy. Rarity is a quality, that gets obliterated once this machine starts putting out "80% facsimiles" by the hundreds of thousands. The originals drown in them, and get diluted out of existence together with the ability to savour the style.
It's all so tragic...
Upvote
17
(18
/
-1)
Yes, humans were bad at perspective too, and they mimicked it the best they could, like AI image generators, until they understood how vanishing points work. Probably some day AI will understand perspective or at least be able to approximate it close enough that we can't tell. But in the meantime, I'm still questioning why we're spending so much money and resources on it. What's the end use?
Upvote
7
(7
/
0)