- Apr 12, 2017
- 626
- 1,094
People who argue against AI art don't do it because hands are deformed. Deformed parts make the result undesirable, period. Whether it is AI generated, 3D model created and rendered, oil painting on canvas, or pencil sketch on a napkin doesn't matter. A bad portion of an image is a bad portion of an image. The general ongoing debate about AI in regards to images is that much of the source material is copyrighted/protected material.
I can't go to facebook or twitter or pull up any random search, pick a person's images, and then start using those images for my own purposes (selling the images outright, using them as "the face of the company" in product literature, or anything else).
The AI basically did do the above. It went online, looked at other people's images, and stored them for later use. A user then goes to the AI and asks for an image. The AI digs around in its storage and returns something. The user then considers the result "their image."
Now the reality is certainly a bit more complicated than that. The AI isn't simply storing, retrieving, and handing out those images. It is fragmenting, slicing, and dissecting the images. The end result is closer to making a collage from the original image sources and even that's a simplification.
Some of the debate is likely due to the potential for technology to replace workers (as has already happened in lots of other jobs and lots of other fields), but the core focused more on potential copyright issues. Technology is going to continue the trend for ages to come of upgrading to do more and more jobs. AI is going to continue to be used for imagery. Arguing that an app has no right to make imagery would be a rather silly stance for anyone to take. But the question of what data (does it owe royalties, does it need permission, is it fair use, etc, etc) is it using from which to create said imagery is a less clear debate.
