Artificial Intelligence created "art" is an abomination for a number of reasons. The biggest of which being that AI gets trained on the works of countless artists who have their works stolen so a computer can mimic what they've created. Without artists, generative AI models couldn't exist.

It's a serious problem because the companies behind generative AI and the people and companies utilizing generative AI are poised to make billions off the technology while the artists who made it possible receive no compensation at all. For this reason, there's been a bit of a backlash when companies use AI "art" for their packaging, advertising, or other materials... but it's not enough of a backlash to stop the trend from trending.

Nope. Generative AI is here to stay.

And now Apple has gotten in on the game with an app called "Image Playground." It's the same type of prompt-driven AI that's been around for a while now, it's just that Apple is attempting to make it easier to use.

And easy it is. You feed Image Playground some parameters that you type or select from suggestions, like so...

And you're presented with a bunch of options that have been generated based on your parameters, such as this one...

Immediately there is a problem in that this looks absolutely nothing like me, despite the fact that I fed it a literal photo of myself. Out of the dozens of friends and family that I put through Apple's generative AI, it usually comes down to the hair. For reasons unknown, Apple is adding tremendous volume and wacky styles that in no way match what was sent to them.

You're always given multiple options, but rarely do the photos of myself look much like me. And, again, it usually comes down to the hair. If they could get the hair drawn better, I think that would be half the battle...

From time to time I get something that kinda-sorta looks like it could be me...

But here's the thing. In all the photos of myself that have been generated, you can't look too close. Because the minute you do, you start to notice some horrific details. Usually with the eyes and teeth...

And once you see it, you go back to look at all the photos that have been generated and everybody looks like serial killers. It's genuinely bizarre, and I'm not really understanding why something so standard as eyes and teeth are so difficult to render.

I did have better luck with photos of my younger self...

Which gets me this...

I can sorta see me in that? It helps to add props...

Even though props are highly problematic. I don't think any of them work as intended. For example, this is me in sunglasses, but AI makes my eyes visible through them, which is freaky as hell...

I put myself in a disco, and suddenly I'm in... suspenders?!?

And when I select "Witches Hat" as an accessory, it's like the AI didn't know what to do, so it put me in a... sombrero?!?

I thought it would be cool to see me as an astronaut, since that's what I wanted to be when I was a kid, but Apple removes the faceplate and has my chin sticking out every time? I don't think that's how spacesuit helmets work...

And it's not just one option... it's 5 out of 6. And sometimes they even give me a cleft in my chin, which is wild...

A lot of times, the variants that you get make it very clear that Apple has run out of ideas as to how to portray a scenario. For example, if I were told to draw somebody as a scientist, I'd be sure to have test tubes and science gear in the shot. But Apple thinks that a lab coat and glasses is all that's needed. Or, as in the first photo, TWO pairs of glasses...

   
Apple has a more specialized version of their Image Playground within their Messages app called "Genmoji" which will generate more simplified versions of stuff. But it's via typing to generate...

I've mainly been concentrating on pictures of me to illustrate the app (at least I can be sure that I own the photo being used as reference!), but you don't have to use photos of people. You can type just about anything.

But with wildly varying results.

When I asked for "A penguin on a polar bear" not a single image generated had the penguin on the polar bear. So I tried again with "A penguin ON TOP OF a polar bear" which gave me multiple options. Only one correct...

Like the last time, the majority of the generated images were just a penguin with a polar bear. And usually the penguin was massively huge like this...

Things get a bit disturbing when the generative AI decided to merge the penguin with the polar bear into a single creature...

But the most disturbing image that I got was a truly gigantic penguin fucking a baby polar bear? I mean, it's definitely on that polar bear, so to speak. Though they are kinda merging at the feet...

So... um... yeah... not quite ready for prime time. There's a lot more work that needs to be done. And though I will likely not use it for anything except being occasionally amused by the results, I can see where this is going to be a very big deal for people who don't want to put the time and effort into creating something themselves having the computer do the work for them.

That's going to do severe harm to a number of different art careers, but artists are getting used to that. They've already had their art stolen a zillion times over to train the AI behemoths that make AI generated images possible in the first place.

I used to take solace in the fact that I'd likely be safely dead by the time AI had fully integrated itself into our lives, but here we are. Mainstream use has not only arrived, but it's progressing so fast that I can't imagine what the future is going to look like.

I just hope it's better than that last penguin on a polar bear image, because yikes.