The first is that all “AI” is not equal. It’s specifically generative AI that most take issue with, mostly due to questionable ethics in training. Image editors have employed techniques marketed as “AI” for many years that are mostly or entirely unrelated to modern generative AI.
The second is that whether something is “AI art” is a spectrum, not binary. On one end you have creations in which generative AI played no role and on the other you have images that were generated off of nothing but a prompt or vague scribbles. In the middle you have things like images where the artist traced over an AI image or used bits and pieces of generated imagery. Probably the closest shorthand for where an image lands on the spectrum is to what degree the creator engaged their artistic skills.
A great many of digital artists would be happy to use Photoshop 7/CS1/CS2, all long predating generative AI, if those ran on modern operating systems. Some prefer modern simplistic (and without AI) tools like Paint Tool SAI.
That might be part of the equation, but for many it’s a strong gut reaction to having the work of themselves and others taken without consent, turned into the visual equivalent of pink slime and press-formed into other shapes, and sold as a service. It just feels wrong. Even I get a little squeamish thinking about it, and I only do art in a minor/hobby capacity — it’s something I’ve put time into, but it doesn’t pay my bills.
If training were purely ethical, the creative community probably still wouldn’t love generative AI, but it probably wouldn’t hate it nearly as much either. It’s the cavalier attitude towards violation of consent for the sake of profit that really seals the deal.
The first is that all “AI” is not equal. It’s specifically generative AI that most take issue with, mostly due to questionable ethics in training. Image editors have employed techniques marketed as “AI” for many years that are mostly or entirely unrelated to modern generative AI.
The second is that whether something is “AI art” is a spectrum, not binary. On one end you have creations in which generative AI played no role and on the other you have images that were generated off of nothing but a prompt or vague scribbles. In the middle you have things like images where the artist traced over an AI image or used bits and pieces of generated imagery. Probably the closest shorthand for where an image lands on the spectrum is to what degree the creator engaged their artistic skills.
A great many of digital artists would be happy to use Photoshop 7/CS1/CS2, all long predating generative AI, if those ran on modern operating systems. Some prefer modern simplistic (and without AI) tools like Paint Tool SAI.