Entry tags:
"the death of art" vs. the further impoverishment of illustrators
This piece about the apocalyptic consequences of AI art is just one of about 1000 pieces on that theme that I've read lately, which are all 1. basically the same and 2. an understandable feeling. And I'm terrible at predicting the future, so I can't say they're wrong. But I do think it's worth mentioning that "will professional illustrators have even more trouble finding work than they already do" is not the same question as "will the whole field of visual art produced by human beings die, because AI art is just as good and everyone will be happy with nothing but that."
Computer text-to-speech technology is now very good, and as a result, there's much less paid work for human beings to record train announcements, telemarketing scams, etc. It's also not very hard to know that what you're hearing is a computer voice; even though there's also technology that can fake a more expressive result (with a very similar approach to AI visual art, that is, imitating existing human samples), if the only goal is "make this text audible" most companies have no reason to bother with that. So, there is and will continue to be a lot of audio material that sounds pretty good and is also obviously a low-effort computer product— not unlike how it's pretty obvious when articles are using stock photos.
A counter-argument would be that the quality of AI visual art is just so much higher than those other things, and indistinguishable from human art— I mean, look at it! But I think that's strongly colored by the novelty effect. We're not used to seeing this stuff; once we're seeing it all the time, I think it's likely that we'll have a pretty good instinct for "yeah, that's another nice piece of AI art." And since the whole idea of this technology requires a large reference database of existing images from the Internet, this could go one of several ways depending on how that database is updated over time:
1. It gets updated very little, because people feel like the AI is already working great as is. So, AI art keeps on riffing on human-generated art that existed online in 2022, and after we've seen a million such riffs, it becomes an easily identifiable kind of thing. It won't be repetitive in the sense of always having the same ideas, since there is a chaotic element to the process, but its ideas about what things can look like won't really change.
2. It gets updated using newer material, but only human-generated newer material. But, if humans will in fact be making a lot less art because AIs are getting most of the work, this is pretty much the same as 1.
3. It gets updated using all of the material online, a lot of which is now AI art. That has the most potential to evolve into something we might call genuinely creative, but feeding on itself could also make it much more obvious that AI art is its own thing with its own weird ideas. People who are into that will be into it, and people who are into other kinds of things done by humans will be into those. And that's not necessarily either/or: that is, "guiding an AI to produce particular kinds of things it would be unlikely to do otherwise" will be an artistic skill humans can have.
I guess that's all a long-winded way of saying that I'm sure this technology will be a major cultural presence from now on, and it'll be another in a long line of developments that make it harder to make money as an artist, and unfortunately money is a major factor in what people do, but none of that has anything to do with killing the whole concept of people engaging in a kind of thing people have done for its own sake for as long as there have been people.
Computer text-to-speech technology is now very good, and as a result, there's much less paid work for human beings to record train announcements, telemarketing scams, etc. It's also not very hard to know that what you're hearing is a computer voice; even though there's also technology that can fake a more expressive result (with a very similar approach to AI visual art, that is, imitating existing human samples), if the only goal is "make this text audible" most companies have no reason to bother with that. So, there is and will continue to be a lot of audio material that sounds pretty good and is also obviously a low-effort computer product— not unlike how it's pretty obvious when articles are using stock photos.
A counter-argument would be that the quality of AI visual art is just so much higher than those other things, and indistinguishable from human art— I mean, look at it! But I think that's strongly colored by the novelty effect. We're not used to seeing this stuff; once we're seeing it all the time, I think it's likely that we'll have a pretty good instinct for "yeah, that's another nice piece of AI art." And since the whole idea of this technology requires a large reference database of existing images from the Internet, this could go one of several ways depending on how that database is updated over time:
1. It gets updated very little, because people feel like the AI is already working great as is. So, AI art keeps on riffing on human-generated art that existed online in 2022, and after we've seen a million such riffs, it becomes an easily identifiable kind of thing. It won't be repetitive in the sense of always having the same ideas, since there is a chaotic element to the process, but its ideas about what things can look like won't really change.
2. It gets updated using newer material, but only human-generated newer material. But, if humans will in fact be making a lot less art because AIs are getting most of the work, this is pretty much the same as 1.
3. It gets updated using all of the material online, a lot of which is now AI art. That has the most potential to evolve into something we might call genuinely creative, but feeding on itself could also make it much more obvious that AI art is its own thing with its own weird ideas. People who are into that will be into it, and people who are into other kinds of things done by humans will be into those. And that's not necessarily either/or: that is, "guiding an AI to produce particular kinds of things it would be unlikely to do otherwise" will be an artistic skill humans can have.
I guess that's all a long-winded way of saying that I'm sure this technology will be a major cultural presence from now on, and it'll be another in a long line of developments that make it harder to make money as an artist, and unfortunately money is a major factor in what people do, but none of that has anything to do with killing the whole concept of people engaging in a kind of thing people have done for its own sake for as long as there have been people.