I have been closely observing the development of generative neural networks for the past few years now, and even tried training one myself about 4 years ago. Though without much luck back then, my computers were a lot less powerful than today, and the technology was in its early development then.
I believe really good artists and creators should not worry too much about the emerging trend of artificially generated images. More on this later, but for now, remember: Computationally cheap general-purpose neural net will produce cheap general-looking images.
In the spirit of adding a speck of originality and personal finesse to my own artificially generated artwork, I trained a neural network with my previous works.
I believe artists and creators should not worry about this social trend of making selfies with “AI”. Fine-tuning a neural net to your specific needs, style, and characteristics will augment your creative processes and in turn, make you into a more valuable creator.
This is an oversimplification, but the fact that neural nets were built upon fancy statistics and mathematics probably hurts some of us more deeply than others.
Perhaps it is a very hard pill to swallow when one realizes that everything we know, everything we do, and everything we will ever understand is just quantifiable data. And I think, this is where most people got it wrong. Infinitely complex pieces of data and parameters that make up a person are so much more valuable than any Al models out there.
Every one of us is a black box filled with infinite data points and parameters. Not a single neural network can replace you. Keep your heads up, you're doing fine.
Feel free to challenge my thoughts and spread them around. And to artists who blindly deny and decline this movement before even touching it themselves, be very careful of unknowingly gatekeeping art just as they did in America back in 1913 with the release of The Cubies' ABC. Look it up.