Hot take but the main issue with ai is that it feeds off of tons of repetitive slop, it sees a picture of an eye ball, and copy and pastes it onto a face it copy and pasted, tracing pretty much. it's easy for it to steal/copy what it sees a consistent pattern of, it works like a mix of a ransom note and a snapchat filter. That's why it still has trouble with constant movement and hands, movement is hard for to keep track of individual parts consistently, especially if you factor in different styles, frame rates and general technique, and hands and dynamic poses require posing and so many varying different angles and perspective, that sometimes it doesn't even read as body parts to ai. What I'm saying is, we need to stop freaking out and simply strive as creatives to do what a machine can not instead dismissing what it can do entirely, if we break away from receptiveness, we won't really uninvent ai but, the machine will struggle to recognize patterns if we keep changing out patterns and strive to be more unique and efficient ourselves, not only rendering the tech obsolete but constantly growing beyond to the point ai will forever just be playing ketchup. and y'know maybe kick normies off the internet too but, that's a blog for another day
Creachanoid
My favorite little quirk about AI gen images is that they produce what is essentially the average of a given input based on the entirety of their available data set. Meaning that what you see is the most boiled down common denominator. Like those old memes about taking the average photo of 100s of pro baseball players faces and the result being "Joe Baseball". A prime example of the most average output is that if you ask an AI to generate an image of a clock face at any other time other than 10:10, the hands on the clock face will always display that time no matter what time you ask for. This is because when selling watches or taking photos of clocks, this time specifically and a few chosen others are the most aesthetically pleasing. Same goes for generating images of a full glass of wine spilling over. Images of half full or even less full glasses are more prevalent due to them looking the most appealing. It can't adapt into what it doesn't know, only what it has been fed. Things like this makes AI extremely easy to spot since the AI results will always be the most average and look just the way they are expected to look. Then on top of this, there are some AI bros that will feed AI to AI in the hopes of producing "Better" results. Garbage in only results in garbage out. Not too worried about AI, everything involved with it is sort of just eating each other already.