ShopDreamUp AI ArtDreamUp
Deviation Actions
Hey, I just got back from... just not being on Deviantart. No trips, no illness, no excuses. I just decided not to come here.
Anyways, I have all these messages to read. I will reply to notes, but not comments. If you sent a comment and need an answer, send me a note.
For the moment, any art I promised to make is postponed indefinitely. Give me some time to think of what to do. I haven't worked on any art at all in all this time, so don't expect any uploads yet. I have gotten a scanner, so I'll be updating all the poor quality sketches I made since September.
Thanks for reading, I guess.
Have fun.
Anyways, I have all these messages to read. I will reply to notes, but not comments. If you sent a comment and need an answer, send me a note.
For the moment, any art I promised to make is postponed indefinitely. Give me some time to think of what to do. I haven't worked on any art at all in all this time, so don't expect any uploads yet. I have gotten a scanner, so I'll be updating all the poor quality sketches I made since September.
Thanks for reading, I guess.
Have fun.
AI is Already Dead
Before you correct me: No, AI is not dead right this second, but its future has been sealed. I got most of the information for this journal from Ed Zitron. Links at the bottom. Reminding you what AI is First of all, it is critically important to get this out of the way: When I say AI, I mean the current generation of Large Language Models (LLM) and Generative Neural Networks: GPT, Google Gemini, Stable Diffusion and Dall-E. I'll be calling them LLMs from now on. If you're going to tell me "AI" is doing cool and valuable stuff, make sure you aren't lumping those LLMs in with actually useful AI technologies, like analytical neural networks or heuristic search trees. AI works well when used in the right place, time and way. With that out of the way, let's get to killing the AI. Hallucinations First of all, we can't ignore the obvious problem with these machines: The LLMs make mistakes. Image generators keep failing to count how many fingers a hand has, or how hair works. Text generators
The story of Google and the Third Party Cookies
I'd love to make this into an interesting story, but I'm not a writer, so I'll just put it straight. So... Third party cookies Cookies are bits of data your browser keeps so that websites can remember what you did while you were on them. Then web developers figured out they could read cookies from other websites if the other sites gave them the right permissions. That's why you can log into Google, then watch Youtube videos on the same account without having to log in again. Then companies figured out a cool new way to use third party cookies: What if large advertising platforms could convince a whole bunch of different websites to use the same cookie everywhere? They could just follow users from one site to another and have a near complete map of each person's browsing history. This gave them the ability to understand a person's interests and daily habits better than most of their own friends did. The decline Unsurprisingly, some governments determined that this was bad, so some
A fun game to play with AI
There's a way you can attack text-to-image generative neural networks in the long run without resorting to using Glaze as a shield and Nightshade as a sword. (By the way, feel free to use Glaze and Nightshade to protect your artwork, if you haven't heard of them yet) If you normally draw highly detailed artwork, make a few (maybe one in every five or ten pictures) with the usual subtle mistakes you'd expect to find in some of the lower quality AI images: deformations in eyes or fingers, strands of hair with no obvious beginning or end, smears or smudges in certain corners, cloth folds on materials that should be pretty rigid... or make up some new minor mistakes of your own, as long as they're barely noticeable at a glance. Mark the images as available for AI training (Or don't. Everyone ignores the label anyways) and don't tell anyone you made them that way. If someone ever makes detection algorithms for Glaze, Nightshade or AI images, your pictures will bypass all of them and get
Doom the Wolf's Unified License
I have decided to create a new license for use of my work and characters. I posted the official version on my website (https://www.doom-the-wolf.com/license), but I'm putting it here so that the license might outlive the website. The old licenses are the following: Old Character License: https://www.deviantart.com/doom-the-wolf/journal/All-Characters-Open-609086031 Old Copyright License: https://www.deviantart.com/doom-the-wolf/journal/Open-access-to-all-of-my-art-770427625 Everything that follows this sentence is the license. Disclaimer This is not a legal document. This license may resemble a legal document but is not written in legalese and will probably not hold up in a court of law. However, I reserve the right to pursue legal action if I believe the license has been violated. I'm just trying to give as many freedoms to my viewers as possible without giving up my ability to prevent use of my works in ways that I consider highly unethical. This document has no loopholes. If the
© 2016 - 2024 Doom-the-wolf
Comments7
Join the community to add your comment. Already a deviant? Log In
Even thought you weren't anywhere, Welcome Back!