The AI Problem

I usually start a blog with a ‘this is what this blog will have’ kind of posts, but I also usually end up just not writing anything after it. So this time, instead of doing a post like that, I thought, you know what, maybe I’ll jump in with both feet into one of the hottest topics right now – AI, and more specifically, using AI to create content.

Now I’m not just talking about art, I’m also including ChatGPT, but let’s face it, it’s the art that has people most pissed, and you know what, that’s understandable.

Why are people pissed?

The basic claim for people being upset is the fact that AI uses the work of real artists to ‘train’ itself so that when someone like me wants a shiny new profile pic, the only effort I need to put is give the app 10 – 20 selfies.

But how does the AI do it? Well, that’s where the ‘intelligence’ or ‘thievery’ happens, it changes depending on who you ask. The AI takes my selfies and ‘borrows’ from open databases that aren’t run for profit and use this way to get art that usually would have IP / copyright powers to ‘learn’ the way to create a type of design. It’ll scan Monet’s art and use the various pieces to create what is effectively a collage of Monet’s style by using your face (or your prompt) as the base. If Monet was alive, he would be pissed as well. And that brings us to the artists outright boycotting and fighting against the AI art agenda, and you know what, I get it. If my art was taken without my consent and then sold and I don’t get a cut out of it – that’s just not the right way to go about it.

In fact, Marques Brownlee (MKBHD) did an amazing video on this and it’s one you should definitely watch. Taking inspiration from the video, I asked ChatGPT what it thought of what’s happening with AI art, you can see my question and responses below.

The interesting part about the whole thing is that no matter how smart it gets, for now, it’s a tool.

It’s not something that can create things on its own, it’s only a tool to be used (at least at the moment, SkyNet could change that). If a writer goes to ChatGPT and asks for prompts to write, then proceeds to write a book based on that prompt which becomes a best seller, would that be stealing from all the databases / language libraries that were used? Would there be a way to even track that?

While AI art has AI play a bigger role in the process, it does feel like fighting the future if you simply disown all AI art. It’s not going to go anywhere, all we can do is provide suggestions so it gets better.

Which brings me to the next thing that no one mentions – the majority of people that are using AI art were probably not going to use an artist anyway. I say this knowing there are instances where a commercial entity like a game or a book used AI art, and that to me is definitely not the way to go about it, but I am talking about the majority. I’m talking about people like me, that just wanted an avatar.

If you look at my avatar on this website at the moment(not anymore as this is an older article 😛 that’s a photo, no AI), that’s an AI generated pic, now did I have malicious intent? Of course not. I’ve done graphic design as a career and I know how much artists struggle. I just saw a nice feature on a nice app and it was an easy way to get it. Was I ever going to go pay for a portrait of my own for 50-100$ and above? Hell no, I don’t have the money or admiration for my own face to spend that much on what is literally a picture that I’ll use for 2 months before forgetting all about it. It’s not part of my branding, it’s never going to be.

What AI art obviously needs, and I can’t believe I am going to say this, is some form of structure or authority like DMCA. Yes, it sucks to not to be able to have music that would be absolutely awesome on your videos and streams, but guess what, you don’t own the rights to it. It tags the people that own the rights, and they get to decide if it’s okay to be streamed or not!

If aural media can manage something like that, surely we can use AI itself to create something that at the very least, gives credit to the artist based on the style.

So if I create a mid journey piece, it would come with a license that said ‘created in mid journey by Govind using Monet’s designs’.

Yes, it’s a simplistic approach, but I’m not smart enough to do a complex one a simple guy. Once you start crediting, the apps/services that make money from it, will become liable to pay for the credit, the royalty so to speak, to the artist. I paid Lensa, now it’s on Lensa’s part to pay the artist.

Pretty much like if I pay a service like Alibi (shameless plug of a sponsor, I know :P) then it’s not on me to worry about the people that create the music I get from Alibi, Alibi takes care of that. Simple as that.

Will any of this happen? I would think so, the streaming/content creation industry brought the music industry into a similar situation and now we have DMCA, now some creators are going to bring heavy weights like Disney/Marvel, etc into this and once that happens, you can be sure there will be regulation. And that’s all AI needs, regulation. It doesn’t need to be blasted to the outer rings of Saturn, it doesn’t need to be thrown in to the trash and burned, it just needs to be regulated. People that should get the credit, should get the credit.

It always comes down to that, and it’s a no-brainer. The creator, should get credit for creation. If AI can learn how to copy the images, surely AI can learn how to credit the creator of said images?

The lead character artist of Dead by Daylight, Eric Bourdages, has a lovely thread about AI art and I think you should definitely give it a read – https://twitter.com/EZE3D/status/1602518743749476353

But until all that comes about, AI art will always be seen like ‘stolen’ or ‘intelligent’ depending on who you ask.