Foreword

While porting the content of my old site to this one, I found a number of blog posts which I never had the courage to finish and post. In retrospect nothing would have changed had I posted them. That said, this one felt worth salvaging, so I'm cleaning up the parts that are useable and breaking it up.

This is a blog post that discusses art and generative artificial intelligence. One problem with discussing ai art is that art is notoriously difficult to quantify the value of. Having written about this previously, let me walk though my previous conclusion, kind of a TLDR for the old post:

AI is good at making art that riffs or builds upon things in its training data, preventing it from doing anything truly original, and it's tuned towards things that look technically proficient as a result of its reward functions. As it floods the market, technical proficiency loses value, and real artists are rewarded for making any art with that bar lowered, similar to how realism fell out of favor in modern art. Essentially, AI art would result in bad human-made art being more valuable, if you focus on expression. We just need to get past our obsession with technical ability.

I wasn't happy with this conclusion at the time, but I think it's a good starting point for a discussion, so it's still been posted here.

Why this feels wrong

I think that approach ignores some of the realities of what making art is like, and of the reality being an artist professionally. Rich and famous people already make bad art which has been inscribed value because they are rich and famous. Art has, for a long time, stopped being about artwork itself - I acknowledge this in the blogpost, so the conclusion obviously doesn't hold weight.

While I had the foresight to realize that AI art would soon become good enough to be indistinguishable from that which is human made, I was shortsighted in thinking that taste would be enough of a differentiator for preventing AI art from ever really getting to being a real problem, and that people would reject it.

So then, here's a bit of a horrifying thought: If AI art is bad because it doesn't have soul, what happens when we can't tell anymore? What happens when AI starts making plunderphonics?

The Problem with Authenticity

To clarify, I do think we're going to move to prioritize physicality a little more. The conclusion was weak for a couple reasons, the first being that authenticity and identity can be faked, and it's not just ai working independently. Not only can authenticity be faked, sometimes faking it is more effective than someone who's not fully authentic. In 2024, I had written out 2 examples.

Kobayashi Yamato

It is possible to fake a persona online for the sake of art, this has been true since the beginning of the internet. People will now go to fairly great lengths to keep up this mirage: in the case of Kobayashi Yamato, creating fake screenshots and content of a cancelled 1992 video game such that they could claim to be the composer of the game's soundtrack. This adds to mystique around the album, which is in the realm of electronic/shoegaze, though no one else seems to think the persona is fake even though the company and project aren't registered, and nowhere else to be seen.

Brat Summer

Authenticity is not something you can control, authenticity is something inscribed onto the author from external sources. Once your work is co-opted by something deemed inauthentic, like marketing or politics, this perception risks being extended to the original work and author itself. In fact, it is easier to fake authenticity than to live a full life under this lens of scrutiny. This is how I viewed Charli XCX's brat being co-opted by the losing Harris 2024 campaign, but I think few fans would agree here.

Velvet Sundown, Breaking Rust

There have already been multiple instances of AI-generated music charting. They go hand in hand with algorithmic music recommendations. Some people have suggested that streaming platforms themselves are generating generic algorithmic music to pad playlists and rack up streams.

TBC

I'm re-examining things under a wider lens. Something that's a bit more about how we interact with art rather than about the art itself, because if it's not about the art, what is it all about? As much as I hate to admit it, I've slowly come to think that artists' say in the matter is very small. I'll continue working this out, in a more bite-size manner this time.