Raw LLM Responses

As a software engineer this amuses me. Thank you for your valiant attempt. Use of AI should be taxed at 99.9% and the proceeds should be used to compensate artists and writers to ensure new data is fed into the models we create. Model pollution is expensive so the tactic is valid (costs me around $1500 to train a new mid sized model set). But in the end I can use another AI to create a filter to filter out polluted data to create a new 'clean' model. Similar to how we use observer models to make an LLM much much more restricted so it doesn't hurt your precious feefees and keep you from having to think as was done with Gemini, CoPilot or any of the other openAI proprietary junk that feeds off of your inputs. As it stands the problem is primarily with how the art is sourced and the necessary attribution or recompense for usage. The licensing model around this needs to be urgently fixed to prevent large amounts of art simply being scraped via a simple one month sub for whatever art site. Keep trying and don't give up. You don't have to beat the tech, you just have to beat the system and licensing concepts around it and the more of a fuzz artists make the less the chance that corpos like Adobe will simply roll over you and scrape your work whether you like it or not. I don't mind losing a model here and there to some clever pollution... keeps me sharp. Good luck out there.