Raw LLM Responses

Hello, I did enjoy your video, and as a fellow artist, I would like to fact check you. (I draw <poorly, still practicing>. Stating this now so I don't get called a fake artist for just using AI). Stolen artwork? Many of them are trained off of public domain images (see Stable diffusion). Also, they don't steal any more than a human glancing around a museam steals art. For it to learn what tree looks like, it would be presented with millions of images of trees, and then it would "think" to itself "oikay, so that's what a tree kinda looks like." It never copies directly. on the other hand, Fanart take direct copies of artwork, and are published around as if new, often with no credit to the original artist. Almost all art uploaded to the internet is unliceneced, giving anybody, including creators of Gen AI, full rights to copy from it. Energy use? Well, there are 2 things to say here. 1: Training a very high-quality Gen AI uses about the same amount of energy as creating a high-budget film, while generating an image uses about the same amount of energy as charging a smartphone. I've never once seen an artist speak out about these. 2: How many times does a digital artist charge their device while doing a commission? 5 times? 10? An iPad has about 2.15x the battery capacity of a smartphone, so that would be equivalent of 10-20 AI images generated, for 1 conventionally made image. If they're using a PC, this stat gets even worse. You are not at all incorrect that AI is being used to create deepfakes of people. These are indeed vile and should not be entertained. Though almost everybody agrees on that, including the governemts of the world and companies that create Gen AI. There are preventative measures in place, and they are growing stronger over time. Eventually, no amount of jailbreaking (not including an amount that crosses the tipping point at which it would just be easier to make it with photoshop) will allow deepfakes of people. AI is a new technology, and regulations are still being put in place and implemented. Responding to the Byte article you put up (3:56), this is very misleading. OpenAI argues that it is impossible to train advanced AI models like ChatGPT without using copyrighted materials. nothing was said about money. This is true, as copyright applies to very many things across the internet. They do not, however violate any copyright law, as its use of copyrighted content qualifies as fair use under copyright law. The purpose of their models is to generate new content based on an understanding of language and reasoning, rather than reproducing existing material. Trying to claim this violates law is like trying to claim that by taking a free class online on how to draw a chicken, the student is violating the copyright of the example images the teacher used every time the student draws one. The comment you put up about how AI art saves time is slightly misleading. Yes, it can save time if you just need a really general image, but it does take a lot of effort to actually learn how to speak to one and get it to give you exactly what you want. Prompts can quite easily get incredibly complex. Artists who use AI will also often have to post-process the art, cleaning it up and fixing any mistakes. Yes, it may take less time in general to make AI art, as opposed to conventional art, but so does digital art. Creating a drawing is (almost always) much, MUCH faster than writing a book. The AI summaries are there so you can quicky glance at your phone and decide if the discussion takes priority over the current situation they're in. They are not meant to be a replacement for actually reading messages. Almost no AI is made with any intent to fully replace anything good (notice my use of "good here". There are a few notable exceptions, such as dangerous jobs) You say at the end that they won't be hurt if they arent doing anything bad, but them learning what a tree is from images is benign. No matter where those images came from. The bigger risk being those who try to actually steal or copy the art and generate funds from it. (I'll reiterate, Gen AI simply learns what things look like based on hundreds of millions of images. They do not copy or steal anything.) Have a good day/night, and I hope you continue to explore artistry! :) Edit: quick note on the last portion of the Q/A, NFTS fell because anyone could get it just by screenshotting. They were stupid idea in the first place, and people only thought the would get big because they were tied to Crypto currencies, which are growing significantly. AI, however, has many real-world applications, is actually useful, and has been a growing technology for almost t80 years now. NFTs lasted but a year.