OpenAI's latest image-generation update has taken social media by storm, as users are flooding X, Instagram, and Reddit with Studio Ghibli-style images
The funny thing is OpenAI’s image generator didn’t really do a good job with making a Ghibli stylized version of Altman.
That being said, there will be a downstream impact on media quality if there is no novel approach to balancing creative work and AI slop generators. Don’t think there is a simple answer.
In theory they get super rich, but in practice the early adopters of AI seem to be hemoraging money as a result of it. It doesn’t actually make the bare minimun content so they end up hiring humans to fix their bullshit and the end product is worse than just using humans.
Replacing clip art, generic filler from Getty images, and other hand-crafted slop with machine-made slop for things like slideshows, YouTube thumbnails, and other applications where the image isn’t meant to convey something actually existing from the primary content, that I think is fine.
Of course it should be based on free software (such as AGPL) and use only freely provided or public domain inputs.
Of course it shouldn’t be used to misrepresent its outputs as produced by, authorized, or of people that it is not.
But what we have right now is an another sort of enclosure of the cultural commons, blended with plagerism-by-another-name. If there are already terms for this sort of misappropriation, I can’t think of them right now.
So we’re teaching the children that only high level art is worthwhile and they shouldn’t even try to make at themselves because they suck at it and you can just generate it. Cool.
How do you define better? More photrealistic? I’d wager kids could learn as much if not more from your own hand-drawn chicken scratch that has a greater emphasis and less distractions on the points you want to convey. They might relate to the lack of conventional quality that they themselves aren’t able to achieve as well. There is an incredible vapidness to AI art. Also it absolutely blows at trying to make anything diagrammatic for teaching. I’ve tried to use it to convey scientic topics that I’d normally use grant funds (back in the day when there were grants) to hire artists to do, and it was an exercise in purified frustration.
The funny thing is OpenAI’s image generator didn’t really do a good job with making a Ghibli stylized version of Altman.
That being said, there will be a downstream impact on media quality if there is no novel approach to balancing creative work and AI slop generators. Don’t think there is a simple answer.
Replacing amazing creative humans with bland AI generated content is not a good use of AI.
Ironic since the decrease of human made work (art or software) will decrease the quality or diversity of generative AI itself
Which the shareholders couldn’t freaking care less. They only need to get super rich in their lifetime.
In theory they get super rich, but in practice the early adopters of AI seem to be hemoraging money as a result of it. It doesn’t actually make the bare minimun content so they end up hiring humans to fix their bullshit and the end product is worse than just using humans.
deleted by creator
Artist will no longer exist as a species
Mostly true, but…
Replacing clip art, generic filler from Getty images, and other hand-crafted slop with machine-made slop for things like slideshows, YouTube thumbnails, and other applications where the image isn’t meant to convey something actually existing from the primary content, that I think is fine.
Of course it should be based on free software (such as AGPL) and use only freely provided or public domain inputs.
Of course it shouldn’t be used to misrepresent its outputs as produced by, authorized, or of people that it is not.
But what we have right now is an another sort of enclosure of the cultural commons, blended with plagerism-by-another-name. If there are already terms for this sort of misappropriation, I can’t think of them right now.
And despite all of its other problems, it’s still not even profitable.
It’s a good use for me. I work with children and the things I’ve “created” have been significantly better thanks to mid-journey.
Before that it was just generic clip art, now I can make really beautifully themed stuff that was both out of my skill range and price range.
The artists, would never get money from me since I’m not rich enough to afford it but the children benefit.
So we’re teaching the children that only high level art is worthwhile and they shouldn’t even try to make at themselves because they suck at it and you can just generate it. Cool.
Well I don’t teach art so… and they don’t even know
How do you define better? More photrealistic? I’d wager kids could learn as much if not more from your own hand-drawn chicken scratch that has a greater emphasis and less distractions on the points you want to convey. They might relate to the lack of conventional quality that they themselves aren’t able to achieve as well. There is an incredible vapidness to AI art. Also it absolutely blows at trying to make anything diagrammatic for teaching. I’ve tried to use it to convey scientic topics that I’d normally use grant funds (back in the day when there were grants) to hire artists to do, and it was an exercise in purified frustration.