Raw LLM Responses

I find it hilarious that people think that Ai companies that cracked human language and nearly broke the tower of babel problem just as a consequence of their original work can beat AI companies with simple data tricks. Here's a kicker for you, screenshot cloning, and pixel remapping before training to purify the data, which both OpenAI and midjourney can both do because they both have video models. OG training used Pixel Mapping because it was easier on an empty data set. Now that they have benchmarks of sorted data prior to everyone trying to poison them, they can use separate training techniques, they can have the model re-map the original picture for consumption similar in a way when trying to predict the next frame of a video. They can recombine sections then saturate the image. _OK A.I bro, if the A.I companies are doing that why are they all worried / haven't said anything._ Not an A.I bro, I'm a Cyber sec bro, who uses Ai for pattern rec regularly and are aware of tech trends, one being, *you don't tell people you fixed a patch, and let them think it's still exploitable* If everyone thinks nightshade and Glare still work when they *don't* OpenAI can keep moving along. My 10 cents if you want your work online and don't want AI taking it, Watermark it, but it's already to late. Ai has made it to far, and due to multimodal and the dawn of supercomputers, this is a losing fight. My other 10 cent would be to try and convince artists to learn to us the AI tools, create bases, help speed up work time, create backgrounds, or help start a new era of expressionism with art that can really only be done with AI. But what do I know... [disclaimer The situation is constantly evolving, and what's true today might not be tomorrow.]