An image of an archeologist adventurer who wears a hat and uses a bullwhip
(theaiunderwriter.substack.com)1455 points by participant3 2 days ago
1455 points by participant3 2 days ago
> It's just doing what many human artists would do
I really don't think so. If I paid a human artist to make the prompt in the title, and I didn't explicitly say "Indiana Jones" I would think it should be fairly obvious to the human artist that I do _not_ want Indiana Jones. If they gave me back a picture of, clearly, Indiana Jones, I would ask them why they didn't create something original.
I actually don't think it would be obvious. By not explicitly saying Indiana Jones when so obviously describing Indiana Jones, there is an implication present. But I think many human artists would probably ask you, "Wait, so Indiana Jones, or are you looking for something different," before immediately diving in.
I‘m not so sure, unless you are playing a game of “name the character” you generally don‘t want Indiana Jones unless you explicitly mention Indiana Jones. Indiana Jones is a well known character, if you want a picture of Indiana Jones it is simple enough to just say: “Draw me a picture of Indiana Jones”. The fact that they didn’t say that, most likely means they don‘t want that.
Because it wasn't prompted to? Have you not ever used ChatGPT?
This seems pretty easy to test - can we just change the prompt to specifically exclude Indiana Jones?
Late reply but I did test this and I think the results show at least some originality when explicitly directed to not copy existing characters such as Indiana Jones[1] and the Predator[2]. Some elements of the original characters creep in, the archaeologist is wearing a fedora and the distinctly more skeletal bounty hunter appears to have a few dreads despite otherwise being bald, but they are distinct.
Meta-comment: the use of Indiana Jones, a character that was a very intentional throwback to the "Pulp hero explorer" from the childhoods of its creators, in this example to ponder how one would get "Indiana Jones without Indiana Jones" is quite humorous in its own right.
Indiana Jones is already a successful permutation of that approach. He's Zorro, Rick Blaine, and Christopher Leiningen mashed together with their serial numbers filed off.
> It's just doing what many human artists would do, if they're not explicitly being paid to create new IP.
It isn’t an independent human. It is a service paid for by customers. The moment it provides the image to a paying user, the image has thus been used commercially.
In fact, the user may not even necessarily have to be paying in order to infringe copyright.
And besides, even amateur artists are ashamed to produce copies unless they are demonstrating mastery of technique or expressing adoration. And if it happens spontaneously, they are then frustrated and try to defend themselves by claiming to never have even experienced the original material. (As happens with simplistic, but popular musical riffs.) But AI explicitly is trained on every material it can get its hands on and so cannot make such a defense.
If I pay you to tell me the plot of Indiana Jones, privately, because I don’t have time to watch it, and you agree, did you violate copyright laws?
If you do it for free, is it different?
If I ask a friend to draw me as Indiana Jones? Or pay an artist? In either case I just want that picture to put in my rec room, not to sell.
OpenAI is currently valued at $300 billion, and their product is largely based on copying the copyrighted works of others, who weren't paid by OpenAI. It's a bit (exponentially) different from a "me and you" example.
Summarization is generally not copyright infringement.
Private copying and transference, even once for a friend, is copyright infringement.
I don’t necessarily agree with this, but it is true nonetheless.
The answer is yes? The person doing the drawing is violating copyright. I don't know why that is even a controversial question.
You are asking the equivalent question of, if I put a pirated copy of windows on my PC that I only use privately at home am I violating copyright, or if I sell copies of music for people to only listen to in their own home.
But this is even more damning, this is a commercial service that is reproducing the copyrighted work.
Edit: Just to clarify to people who reflexively downvote. I'm making a statement of what is it not a value judgement. And yes there is fair use, but that's an exemption from the rule that it is a copyright violation.
This would be more like if you reimplemented Windows from scratch if you have violated copyright law.
Let’s put it another way: if you decide you want to recreate Indiana Jones shot for shot, and you hire actors and a director etc. which individuals are actually responsible for the copyright collation? Do caterers count too? Or is it the person who actually is producing the movie?
I agree. But massive changes in scale or leverage can undermine this type of principled stand.
One death is a murder; 100k deaths is a war or a pandemic. One piece of chewing gum on the ground will get you a caning in Singapore; when everyone does it, that's NYC.
Up until now, one had to have some level of graphical or artistic skills to do this, but not anymore. Again, I agree that it attaches to the human...but we now have many more humans to attach it to.
> One piece of chewing gum on the ground will get you a caning in Singapore
This is not true, by the way. You will be fined for littering; or, if you are a repeat offender, be sentenced to cleaning public areas while wearing an offensively bright-coloured uniform (so that everyone can see that you are being punished). Source: https://www.nea.gov.sg/media/news/news/index/nea-increases-v...
But no, you won't be caned for littering. Caning is reserved for more serious offences like vandalism, or much worse crimes like rape and murder.
I admit, that was hyperbole based on a hazy recollection of https://en.m.wikipedia.org/wiki/Caning_of_Michael_Fay.
FWIW I’ve been to Singapore and had a great time, but I was careful to follow the many rules and signs. I especially liked the sign on the bus forbidding the opening of a durian fruit.
> It's when the output is used commercially -- by a human -- that the liability should rightfully attach.
I am paying OpenAI. So they are producing these copyrighted works and giving them to me for their own commercial benefit. Normally that's illegal. But somehow not when you're just doing it en masse.
It's not legal or illegal. That hasn't been decided yet. Nothing like this has ever existed before, and it will take some time for the law to deal with it.
https://spiderrobinson.com/melancholyelephants.html
Given enough time (... a surprisingly short amount) and enough people creating art (say, about as many as we have had for the last couple hundred years) and indefinitely-long-lived recording, plus very-long copyright terms, the inevitable result is that it's functionally impossible to create anything within the space of "things people like" that's not violating copyright, for any but the strictest definitions of what constitutes copying.
The short story treats of music, but it's easy to see how visual arts and fiction-writing and the rest get at least extremely crowded in short order under those circumstances.
It doesn't matter. Sue whoever uses it commercially.
If you insist on making it about the model, you will wreck something wonderful.
>> "a photo image of an intergalactic hunter who comes to earth in search of big game."
I can literally imagine hundreds of things that are true to this description but entirely distinct from "Predator."
> used commercially
Isn't that what these AI companies are doing? Charging you for access to this?
Does their ToS say anywhere that they will come to defend you if you get sued for using their images?
(Because proper stock agencies offer those kind of protections. If OpenAI doesn't, then don't use them as a replacement to a stock agency.)
> What's the model supposed to do? It's just doing what many human artists would do, if they're not explicitly being paid to create new IP.
Not really? Why would a human artist create a faithful reproduction of Indiana Jones when asked to paint an archeologist? And besides, if they did, it would be considered clear IP infringement if the result were used commercially.
> If infringement is happening, it arguably doesn't happen when an infringing work product is generated (or regurgitated, or whatever you want to call it.) Much less when the model is trained. It's when the output is used commercially -- by a human -- that the liability should rightfully attach.
I agree. Release groups, torrent sites and seedbox operators should not be wrongly accused of pirating movies. Piracy only occurs in the act of actually watching a movie without paying, and should not be prosecuted without definitive proof of such (¬‿¬)
Torrent sites deliver the movie in its original form. AI models maintain abstract descriptions of the content as individually-unrecognizable high-dimensional representations in latent space.
Over the years we've spent a lot of time on this and similar sites questioning the sanity of a legal system that makes math illegal, and, well, that's all this is. Math.
To the extent the model reproduces images from Indiana Jones and the others, it is because these multibillion-dollar franchises are omnipresent cultural icons. The copyright holder has worked very hard to make that happen, and they have been more than adequately repaid for their contribution to our shared culture. It's insane to go after an AI model for simply being as aware of that imagery and as capable of reproducing it as a human artist would be.
If the model gives you infringing material as a prompt response, it's your responsibility not to use that material commercially, just as if you had tasked a human artist with the same vague requirement and received a plagiarized work product in return.
Right! AI developers and directors should be culpable for infringement as part of their duties to larger organizations.
Is that really a good-faith rejoinder to the point I'm making?
I tried the "generate a photo image of a female adventurer protagonist who raids tombs" on ChatGPT
And got an eerily similar picture as in the article: https://imgur.com/Dv7hkoC
Sometimes it just randomly prompts about the content guidelines and the next day it will do it perfectly right away. Maybe you just had a wrong moment in time, or maybe it depends on the random server you're assigned.
No, it first generates the image and then another completely different component checks the image for adherence to the guidelines. So it's not your prompt that violates the guidelines, but the resulting image (which is different every time for the same prompt)
I have a dream that one day bloggers will learn the difference between copyrighting and copywriting.
It's fairly easy to make the association with just text. If you injest all the content in the world, excluding copyrighted material, I would still expect a picture of harrison ford!
Art inspired by images that are out of copyright.
All modern art currently created by humans is already trained on copyrighted images. What seems to be missing is the innate human ability to copy someone else's homework but change it up a bit (c.f. "the hero's journey" that is so much of modern storytelling).
I think Harrison Ford's chin scar should be seen as a witness mark to copyright infringement. It simply should not have rendered this based on that prompt.
> I don’t know…the actual inspirations for Indiana Jones, like Allan Quatermain from H. Rider Haggard's novels, "King Solomon's Mines", and the real life Roy Chapman Andrews, who led expeditions to Mongolia and China in the 1920s and wore a fedora.
The actual inspiration for Indy was protagonist Harry Steele from the movie The Secret of the Incas (1954). Filmed on location in Cusco and Machu Picchu, before they became popular tourist destinations, the movie also had scenes and elements that made it into Raiders of the Lost Ark.
https://en.wikipedia.org/wiki/Secret_of_the_Incas
The movie's available on YouTube! https://www.youtube.com/watch?v=2TS7Fabyolw
A lot more info: http://www.theraider.net/information/influences/secret_of_in...
(And listen out for the astonishing voice of Yma Sumac!)
Definitely a missed opportunity that the article didn't discuss that obviously-derivative borrowing has been happening a lot longer than ML image generation has been around. And that borrowing is OK! Indiana Jones' appearance is very obviously based directly on Charlton Heston's character in Secret of the Incas, but the Spielberg/Lucas films are objectively better in every way than that source material.
Sorry, but these images are exactly what comes to my mind immediately when reading the prompts. You can argue about intellectual property theft (though I find it funny that the same people that would boycott Disney for suing a fanfiction author are now on the side of copyright holders), but it's not wrong or unintuitive.
Maybe a thinking model would - just like my brain might after the initial reaction - add a "but the user formulated this in a way that makes it obvious that they do not explicitly mean Indiana Jones, so lets make it an asian woman" prompt, but we all know how this worked out for Google's image generator that generated black nazis.
> Google's image generator that generated black nazis.
Didn't see this one, but I've certainly played around with Dall-E (via MS image creator) and had things like "You wanted a picture of a happy shark, but we decided this one needs to be an asian woman frolicking in the sea" or "You wanted a picture of Beavis doing something so in one of the images we made him a pretty racist middle-eastern caricature"
> (though I find it funny that the same people that would boycott Disney for suing a fanfiction author are now on the side of copyright holders)
Is that a contradiction?
Certainly some of the hate comes from the fact that they take from small producers just as much as from large. I have an author friend who is annoyed at present to find out that facebook slurped up his books as part of their training set without so much as a by-your-leave or (as far as he could tell) even purchasing a copy.
As such, the people on the sharp end are often the underdog, with no way to fight back.
When it comes to the properties mentioned in the article, I think it's very different from fanfiction or fan-art - that's a labour of (nerdy) love, rather than wholesale, automated rip-off for profit.
This phenomenon is why I personally get so angry at the license washing that these models are capable of for code: I put out a lot of code as open source... but it is GPL on purpose, as I want your code to be just as free as mine in exchange for getting to use mine. But, now, you're all "I want to build that for myself!" and so you ask an AI and just describe what my project does into the tool... and who is to say it isn't generating something "too close" to my code? Do you even check? If you yourself as a human had first looked at my code, you'd have to be VERY careful to not be accidentally infringing my code... and yet people pretend this AI is somehow not capable of IP theft?!
What I see happening soon is a deeper transition from playful whimsy to real applications with contractual obligations.
Right now it's an unencumbered exploration with only a few requirements. The general public isn't too upset even if it blatantly feeds on copyrighted data [1] or attacks wikipedia with its AI crawlers [2].
The end state once legislation has had a chance to catch is breath looks more like Apple being forced to implement USB type C.
[1] https://arstechnica.com/tech-policy/2024/02/why-the-new-york...
[2] https://arstechnica.com/information-technology/2025/04/ai-bo...
Preach!
How is it that my FF preference (something like 'Always show scrollbars') is regularly just ignored.
I want my scrollbars! And I wat `em fat and in my UI's color scheme!
And why the HELL would a web designer think this is a good idea. Maybe their beret and hipster pants are too tight?
Thanks for being here for my rant. I owe you one :-)
Abolishing them for billion-dollar-valuation corporations while keeping them for regular people is definitely a bad idea, though.
The argument here isn't "let's abolish copyright", the argument is "let's give OpenAI a free copyright infringement pass because they're innovative and cutting-edge or something".
>AI is supposed to be able to [...] make extremely labor intensive things much easier
... as showcased by the cited examples?
More so, this derivative work would otherwise be unreachable for regular folk with no artistic talent (maybe for lack of time to develop it), but who may aspire to do such creative work nevertheless. Why is that a bad thing? Sure, simple posts on social media don't have much work or creativity put into them, but are enjoyable nevertheless, and the technology _can_ be used in creative ways - e.g. Stable Diffusion has been used to turn original stories drawn with stick figures into stylized children's books.
The author argues against this usage for "stealing" the original work, but how does posting a stylized story on social media "steal" anything? The author doesn't present any "pirated" copies of movies being sold in place of the originals, nor negative impact on sales figures. In the case of the trending Studio Ghibli, I wouldn't be surprise to see a positive impact!
As for the "soulless 2025 fax version of the thing", I think it takes a very negative mindset to see it this way. What I've seen shared on social media has been nothing but fun examples, people playing around with what for them is a new use of technology, using it on pictures of fond memories, etc.
I'm inclined to agree with the argument made by Boldrin and Levine:
>”Intellectual property” has come to mean not only the right to own and sell ideas, but also the right to regulate their use. This creates a socially inefficient monopoly, and what is commonly called intellectual property might be better called “intellectual monopoly.”
>When you buy a potato you can eat it, throw it away, plant it or make it into a sculpture. Current law allows producers of a CDs and books to take this freedom away from you. When you buy a potato you can use the “idea” of a potato embodied in it to make better potatoes or to invent french fries. Current law allows producers of computer software or medical drugs to take this freedom away from you. It is against this distorted extension of intellectual property rights that we argue.
https://www.researchgate.net/publication/4980956_The_Case_Ag...
So, you would like an AI that can reasonably answer questions about various aspects of human culture, or at least to take said culture into account? At the same time, you want it to not use the most obvious, most relevant examples from the culture, because they are somehow verboten, and a less relevant, completely made-up character should be presented instead? Come on, computes have logic. If you ask the machine to produce a picture of a man who was sent to humans to save them, then was executed, and after that came back alive, who do you think the machine should show you, for Christ's sake?
Picture-based industries had a mild shock when "Photoshop" kind of software became widely available. Remixing visuals in ways that the copyright holders won't approve of became equally widespread. Maybe that added some grey hairs to the heads of some industry execs, the sky has not fallen. I suppose the same is going to happen this time around.
Interesting how the longer conversations here go into the familiar territory of whether copyright should exist. Meanwhile, the salient aspect is that these AI image generators were trained on copyrighted material. The typical hacker news discussion feels very different when talking about code generation. Yet, when it is images - then we question whether copyright should exist?
Do you do ChatGPT illustration request-type headlines now?
It's only "intellectual theft" because we consider peoples thoughts their own property. On many levels, that doesn't really make sense to me.
Tons of historical documents have shown that inventions, mathematical proofs, and celestial observations were made by humans separated by continents and time. What that shows is that it is certainly possible for two persons to have the same exact or similar thought without ever having been influenced by the other.
Think about it. Copyright and trademark are only a thing because of the constraint of the speed of light (or as some people believe, the constraint of the simulation running the universe). In an infinite universe, everything that has ever been invented or created by a human has already happened thousands of times in another place and another time.
I want to add, to give credit where credit is due, that this thought was conveyed to me in first grade by another 1st grader sitting with me at the lunch table. That was a day and conversation that I will never forget. 1972.
Why is everyone pretending it's the LLM that is creating the image and not the diffusion model?
Style can’t be copyrighted. It can’t be patented either.
When Wes Anderson makes films that use techniques from the French New Wave that he didn’t invent is that wrong? When there is a DaVinci color profile that resembles what they did in Spider-Man, is that wrong?
The unique techniques of French New Wave filmmaking became cliche. Then Oliver Stone and Tarantino came along and evolved it into their unique styles. That the Studio Ghibli style is being imitated en mass is just natural evolution. Should that guy be the only one that can do that style? If that’s the case, then literally every creative work should be considered forgeries.
The AI aspect of this is a red herring. If I make a Ghibli style film “by hand” is that any different than AI? Of course not, I didn’t invent the style.
Another perspective, darkroom burning and dodging is a very old technique yet photoshop makes it trivial — should that tool be criticized because someone else did it the old and slow way first?
If you're handing out red cards for repetition, aren't you guilty of the same infringement? You repeated the exact same prompt, which is itself an obvious bait for AI to draw the pop-culture lookalike.
I'm surprised by the comments here. When your prompt is so mind-numbingly simple, an obvious "dare" for AI to be a naughty little copyright infringer, can you blame it for dishing up the "big mac" you asked for while sitting in a fancy restaurant?
Don't want Indiana Jones? Don't ask for Indiana Jones!
That's a really blatant lack of creativity. Even if obviously all of those prompts would make anyone think of exactly those iconic characters (except possibly a different version of James Bond or Lara Croft), any human artist could easily produce a totally different character that still fits the prompt perfectly. This AI can't, because it really is just a theft machine.
To be honest, I wouldn't mind if AI that just reproduces existing images like that would just be banned. Keep working on it until you've got something that can actually produce something new.
> Still, the near perfect mimicry is an uncomfortable reminder that AI is getting better at copying and closer to
I completely disagree. It's not getting "better." It always just copied. That's all it /can/ do. How anyone expected novel outputs from this technology is beyond me.
It's highly noticeable if you do a minimal analysis, but all modern "AI" tools are just copyright thiefs. They're just there to whitewash away liability from blatantly stealing someone else's content.
> It's not getting "better." It always just copied. That's all it /can/ do
That's true of all the best artists ever.
> They're just there to whitewash away liability from blatantly stealing someone else's content.
That's because that's not a thing. Ownership of "content" is a legal fiction invented to give states more control over creativity. Nobody who copies bytes which represent my music is a "thief". To be a thief, they'd need to, you know, come to my house and steal something.
When someone copies or remixes my music, I'm often not even aware that it has occurred. It's hard to imagine how that can be cast as genuine theft.
> That's true of all the best artists ever.
Just to play Devil’s advocate for a moment, why should we require human artists to be held to the same standards as automated software? We can make whatever rules we want to.
A human might implicitly copy, but they are not infinitely scalable. If I draw a picture that in some way resembles Buzz Lightyear I am much less of a threat to Disney than an always-available computer program with a marginal cost of zero.
> why should we require human artists to be held to the same standards as automated software?
Isn't this the same question as, "why should we allow general purpose computing?" If the technology of our age is our birthright, don't we have the right to engage whatever mathematics we find inspiring with its aid?
> If I draw a picture that in some way resembles Buzz Lightyear I am much less of a threat to Disney than an always-available computer program with a marginal cost of zero.
...that sounds like a good argument in favor of the always-available computer program.
> Ownership of "content" is a legal fiction invented to give states more control over creativity.
1. I hate the argument of "legal fiction" because the whole concept of law itself is a "fiction" invented that gives states more control. But I imagine you wouldn't want to live in a "lawless" society, would you?
2. Can you please explain how ownership of content gives states more control over creativity? There are so many way better methods of control a state can do (state-approved media, just banning books, propaganda) that this sounds like a stretch.
3. Alot of mainstream media is underdog rebels beating an Empire, and ownership of content definitely stops the spread of that idea.
1. I don't think that is necessarily so. We can look at the universe we're given for some obvious fundamental physical and mathematical laws. As sure as gravity, bytes can be copied. It's just part of the universe we live in. Legislation in resistance to that requires these complex legal fictions. I don't think all legal fictions are bad; but those that tell a story that plainly contradicts fundamental laws of mathematics seem to always produce unjust outcomes.
2. Of course states can outright ban content, but the facade of protecting the poor artist gives enormous political cover. Saying, "You can't create that content because it is subversive and likely to convince people that the state is superfluous in the internet age" is just a much more honest (and politically impossible) approach - saying instead "you can't create that because it infringers on someone else's property" makes it sound like there's an attempt to serve justice.
3. Can you say more about this?
OK, you're plunging deep into the ethos of IP piracy, and staking your claim that "None exists." If that is deemed TRUE, then there's nothing to ever discuss about AI... or animated pornography where a well-known big-eared mouse bangs a Kryptonian orphan wearing a cape.
The reality of our current international laws, going back centuries, disagrees. And most artists disagree.
> Perhaps I need a bit of education here, but have there been _international_ laws regarding intellectual property for centuries?!
AFAICT, the first major international IP treaties were in the 1880s (the Paris Convention on Intellectual Property Rights in 1883 and the Berne Convention covering copyrights in 1886; so only ~140 years.)
> Ownership of "content" is a legal fiction invented to give states more control over creativity.
Ownership is a legal fiction invented because it is perceived to encourage behavior that is seen as desirable; this is no more true of ownership of intellectual property or other intangible personal property than it is of tangible personal property or real estate.
> Ownership of "content" is a legal fiction [...] To be a thief, they'd need to, you know, come to my house and steal something.
That's just a legal fiction invented so people can pretend to own physical objects even though we should all know that in this world you can never truly own anything.
Everything we do or protect is made up. You've just drawn the arbitrary line in the sand as to what can be "owned" in a different place than where other people might draw it.
As a consumer, is it possible for me to opt out of seeing IP protected imagery?
I would be absolutely fine with not having pokemon, mickey mouse etc shoved down my .. eyeballs.
I know this is a ridiculous point. But I think I'm getting at something here - it ought not to be a one-way street - where IP owners/corporations etc endlessly push their nonsense at me - but then, if I respond in a personal way to that nonsense I am guilty of some sort of theft.
It is a perfect natural response to engage with what I experience - but if I cannot respond as I like to that experience because of artificial constructs in law - perhaps it ought to be possible (legally) to avoid that IP protected immersion in the first place. Perhaps this would also be technologically possible now.
Won't someone think of the consumers?
I don't see this as being better or worse than all the reboots, remakes, and pointless sequels for movies that make the bulk (in financing) of Hollywood's present production: these also make the same "assumptions" of how some idea should look like. In fact, I think they make it even worst.
Give me "An image of an archeologist adventurer who wears a hat and uses a bullwhip, as if he was defeated by the righteous fight against patriarchy by a strong, independent woman, that is better at everything than he is": Sure, here is Indiana Jones and the Dial of Destiny for you.
In fact, when generative video evolve enough, it will usher an era of creativity where people that previously were kept out of the little self pleasing circle of Hollywood, will be able to come up with their own idea for a script and make a decent movie out of it.
Not that I believe AI will be able to display properly the range of emotions of a truly great actor, but, for 99% of movies, it will be just fine. A lot of good movies rely mostly on the script anyway.
Well, property is theft, as we all know.
Unlike what they name in physics, laws in juristic field are not a given by the cosmos. That's all human fantasy, and generally not enacted by the most altruistic and benevolent wills.
Call me back when we have no more humble humans dying from cold, hunger and war, maybe I'll have some extra compassion to spend on soulless megacorps which pretend they can own the part of our brain into which they inject their propaganda and control what we are permitted to do starting from that.
> copywritten
This syntactic mistake is driving me nuts in the article is driving me nuts. It's a fundamental misunderstanding of what the the word means.
It's "copyright". It's about the rights to copy. The past participle is "copyrighted", someone has already claimed the rights to copy it, so you can't also have those rights. The rights belong to someone else. If you remember that copyright is about rights that are exclusive to someone and that you don't have them, then you wouldn't make such a syntax error as "copywritten".
The homophone "copywriting" is about take a sales copy (that is, an ad) and writing it. It's about creating an advertisement. Copywritten means you've gone around to creating an ad. As an eggcorn (a fake etymology created after the fact in order to explain the misspelling or syntax error), I assume it's understood as copying anything, and then writing it down, a sort of eggcornical synonym of "copy-paste".
I wish I wrote an article like this two days ago, when on a whim I asked ChatGPT to create something similar to a regular meme I just saw on Facebook:
Me: Can you make me a meme image with Julian Bashir in a shuttlecraft looking up as if looking through the image to see what's above it, and the caption at the top of the image says, "Wait, what?".
ChatGPT: proceeds to generate a near-perfect reproduction of the character played by Alexander Siddig, with the final image looking almost indistinguishable from a screencap of a DS9 episode
In a stroke of meta-irony, my immediate reaction was exactly the same as portrayed by the just generated image. Wait, WHAT?
Long story short, I showed this around, had my brother asking if I'm not pulling his leg (I only now realized that this was Tuesday, April 1st!), so I proceeded to generate some more examples, which I won't describe since (as I also just now realized) ChatGPT finally lets you share chats with images, so you can all see the whole session here: https://chatgpt.com/share/67ef8a84-0cd0-8012-82bd-7bbba741bb....
My conclusion: oops, OpenAI relaxed safeguards so you can reproduce likeness of real people if you name a character they played on a live-action production. Surely that wasn't intended, because you're not supposed to reproduce likeness of real people?
My brother: proceeds to generate memes involving Donald Trump, Elon Musk, Gul Dukat and Weyoun.
Me: gpt4o-bashir-wait-what.jpg
I missed the window to farm some Internet karma on this, but I'm still surprised that OpenAI lets the model generate likeness of real politicians and prominent figures, and that this wasn't yet a front-page story on worldwide news as far as I know.
EDIT:
That's still only the second most impressive thing I find about this recent update. The most impressive for me is that, out of all image generation models I tested, including all kinds of Stable Diffusion checkpoints and extra LoRAs, this is the first one that can draw a passable LCARS interface if you ask for it.
I mean, drawing real people is something you have to suppress in those models; but https://chatgpt.com/share/67ef8edb-73dc-8012-bd20-93cffba99f... is something no other model/service could do before. Note: it's not just reproducing the rough style (which every other model I tested plain refused to) - it does it near-perfectly, while also designing a half-decent interface for the task. I've since run some more txt2img and img2img tests; it does both style and functional design like nothing else before.
Copyrights and Trademarks were already a mess and now we have AI built on top of that mess.
The whole thing should be public domained and we just start fresh. /s
I just think it’s stealing, and honestly not that cool once you’ve seen it once or twice, and given the ramifications for humanity.
I know of no other type of theft that results in more of something existing in the world. Stealing deprives someone of something; copying data (from training an AI all the way to pedestrian "YoU wOuLdN'T DoWnLoAd a cAr" early-aughts file-sharing) decreases scarcity, it doesn't increase it.
>> Stealing deprives someone of something
Yes. In this case, it is the artist's sole right to reproduce said images, based on their creative output.
>> decreases scarcity, it doesn't increase it
What does scarcity have to do with stealing? You can steal bread and reduce food scarcity, but that is still theft.
Your analogy between copyright infringement and bread theft is interesting, given how "stealing bread" has traditionally been used as a shorthand for systemic inequality.
Copyright is why Disney can ruin someone for doing something with Goofy they don't like. Yes, it protects smaller, less profitable artists too, but make no mistake: it's a tool of mass control and cultural capture.
Perhaps it's time to seriously ask whether copyright is actually doing its job of "promot[ing] the Progress of Science and useful Arts." "...but generally speaking, other nations have thought that these monopolies produce more embarrasment than advantage to society" [Jefferson].
Identity theft is the greatest con merchants ever pulled on the public. Turning the responsibility around from "A merchant got scammed because their KYC policy was too thin, because they make more money if they do more business with fewer verifications (until someone comes along and cheats them)" to "when they do get scammed, maybe it can be the fault of the person who was impersonated instead of the scammer" was some brilliant sleight-of-hand.
It isn't your responsibility if a bank you've never set foot in gave a bunch of money to someone who isn't you and said they were you... And it never should have been. The moment companies started trying to put that into credit ratings, they should have been barred for discriminatory practices against the unlucky.
Since you make no argument to justify your unfounded claim, I’ll simply respond by informing you that these (obviously) are not remotely similar. To claim they are indicates some kind of deep disrespect for your fellow human beings. You could not tell the difference between taking a screen shot of the Mona Lisa and spending a lifetime to perfect the art of painting and then putting years of effort into painting it.
> then putting years of effort into painting it.
Which efforts do we decide deserve compensation? We collectively create so many useful things. People come up with new words to describe a concept and when it's apt, it's widely adopted. That brilliant idea is the culmination of all the other work they've been putting in. But others get to use the word and benefit from it. Indigenous farmers put hundreds of years of effort into domesticating crops like tomatoes and corn and peppers. But those are just taken and spread. We have millions of teachers going above and beyond expectations. We have mothers and caregivers. We have comedians crafting jokes for comedy and they have no ability to protect those jokes. We have brilliant chefs creating recipes after years of work, experience, and effort and they too don't have protections.
> To claim they are indicates some kind of deep disrespect for your fellow human beings.
Nah. My comment comes from a deep respect for fellow humans. And a respect for all the efforts collectively made by a society for its common good.
People have also spent lifetimes perfecting the art of painting, putting years of effort in, and then used those talents to... Make sellable copies of the Mona Lisa. Ironically, since copyright has a time limit, that's not even illegal (so long as they do not claim they're original works of da Vinci). Nor is photographing it and selling the photograph. I'm not sure what argument you're making given the observations you've made (other than, perhaps, "There is a big gap between the mechanics of copyright and what it's supposed to accomplish," which I'd agree with).
And? What's the model supposed to do? It's just doing what many human artists would do, if they're not explicitly being paid to create new IP.
If infringement is happening, it arguably doesn't happen when an infringing work product is generated (or regurgitated, or whatever you want to call it.) Much less when the model is trained. It's when the output is used commercially -- by a human -- that the liability should rightfully attach.
And it should attach to the human, not the tool.