I'm not entirely disagreeing, but a couple of thoughts have been taken a bit out of context.
- For "Studio Ghibli" style using 4o, even a simple prompt like 'make this anime style' was working in the same fashion. Plenty of other studios have copied the Ghibli style to a point where it has become generically anime.
- The instance where Miyazaki mentioned that AI was an 'insult to life itself' and that he would never use the tech in practice comes from a brief tech demo where he was shown a nightmarish 3d AI model demo for a zombie nearly a decade ago. Not surprisingly, he hated it. A lot has changed in a decade. I'm sure people will be quoting it a decade from now once we have even better image diffusion.
I DO agree that it reshapes meaning. If everyone has as Ghiblified profile picture, it really does take the charm away. Something that would normally have taken a talented artist a week to churn out for you takes a minute. This still calls into question how we're going to handle copyright law in the future. But generic art styles cannot be copyrighted.
I still want to see artists fairly compensated and live in a world where art is cherished. At the same time, this tech will continue to march forward. If it wasn't OpenAI, it would have been DeepSeek.
No this really isn't something to pass on blame to the consumer. This is an AI policy and trust decision being trumped by commercial priorities. I understand OpenAI is desperate for revenue growth and there's no limit now to what they are capable of doing to that end.
At the same time, if AI is a tool, a tool can be used for good or evil depending on who wields it. A tool can be misused as well. These are early tools. Historically, when these claims of job loss and dehumanization have been made (and they have been made every time a technology has transformed productivity) we discover that in fact they lead to new forms of production and new forms of art and expression. This happened with the phonograph, the camera, radio, etc etc etc. the question now is “is it different this time?” The jury is out in this, and a multiplicity of smaller, more targeted, or even personalized LLMs are being produced. Are we arguing that the ability of AI to provide better diagnosis for medical imaging analysis, or capable of snarking out subtle patterns in DNA replication? This is the same underlying system in use. To me the danger is in “who owns AI?” We have an opportunity to continue the path of increasing productivity to the point where we could provide a universal basic income to everyone and then allow them to pursue whatever endeavors they desire. Or, we could have a world where all AI is owned by three trillionaires. This is sobering and what we need to be most aware of: it is not AI but humans using AI where the danger is. We already know that evil agents will be using this. What we need is more interaction of people within physical space: this is where real humanity lies. And yes, that includes this discussion, done asynchronously in a symbolic text modality and about as far removed from “natural human expression” as you can get.
I am not saying all is fun and games, but this question is more nuanced than the writer may suggest.
OpenAI's endless copyright infringement continues: https://www.theregister.com/2025/04/03/openai_copyright_bypass/
I'm not entirely disagreeing, but a couple of thoughts have been taken a bit out of context.
- For "Studio Ghibli" style using 4o, even a simple prompt like 'make this anime style' was working in the same fashion. Plenty of other studios have copied the Ghibli style to a point where it has become generically anime.
- The instance where Miyazaki mentioned that AI was an 'insult to life itself' and that he would never use the tech in practice comes from a brief tech demo where he was shown a nightmarish 3d AI model demo for a zombie nearly a decade ago. Not surprisingly, he hated it. A lot has changed in a decade. I'm sure people will be quoting it a decade from now once we have even better image diffusion.
I DO agree that it reshapes meaning. If everyone has as Ghiblified profile picture, it really does take the charm away. Something that would normally have taken a talented artist a week to churn out for you takes a minute. This still calls into question how we're going to handle copyright law in the future. But generic art styles cannot be copyrighted.
I still want to see artists fairly compensated and live in a world where art is cherished. At the same time, this tech will continue to march forward. If it wasn't OpenAI, it would have been DeepSeek.
Hey Michael. Very well said! Fuck @sama's at-scale-thievery. You might be interested to see this Substack i posted yesterday about "AGI as Infomatic Delusion." https://margaretwertheim.substack.com/p/agi-as-infomatic-delusion
Well said.
It's a case of even though you can do something - should you?
OpenAI isn't entirely to blame here. Everyone had a choice whether to use chatgpt or not.
To create an anime image using Ghibli style or not....
No this really isn't something to pass on blame to the consumer. This is an AI policy and trust decision being trumped by commercial priorities. I understand OpenAI is desperate for revenue growth and there's no limit now to what they are capable of doing to that end.
At the same time, if AI is a tool, a tool can be used for good or evil depending on who wields it. A tool can be misused as well. These are early tools. Historically, when these claims of job loss and dehumanization have been made (and they have been made every time a technology has transformed productivity) we discover that in fact they lead to new forms of production and new forms of art and expression. This happened with the phonograph, the camera, radio, etc etc etc. the question now is “is it different this time?” The jury is out in this, and a multiplicity of smaller, more targeted, or even personalized LLMs are being produced. Are we arguing that the ability of AI to provide better diagnosis for medical imaging analysis, or capable of snarking out subtle patterns in DNA replication? This is the same underlying system in use. To me the danger is in “who owns AI?” We have an opportunity to continue the path of increasing productivity to the point where we could provide a universal basic income to everyone and then allow them to pursue whatever endeavors they desire. Or, we could have a world where all AI is owned by three trillionaires. This is sobering and what we need to be most aware of: it is not AI but humans using AI where the danger is. We already know that evil agents will be using this. What we need is more interaction of people within physical space: this is where real humanity lies. And yes, that includes this discussion, done asynchronously in a symbolic text modality and about as far removed from “natural human expression” as you can get.
I am not saying all is fun and games, but this question is more nuanced than the writer may suggest.