Raw LLM Responses

If I understand what you advocate for: it would be fair to let Image Generators "learn" from large datasets, just like we, human artists, do. Or else, human artists should fund the development of bodies and eyes for those LLMs, so that they are closer to the humans. Thus we wouldn't stand against the data theft since we would empathize with them? If the issue is with copyrights and malicious use of protected datas, why should we adress giving programs limbs and eyes, and not talk about using the exisiting set of copyright laws first. To clean the path for ethical AI programs to emerge, and then we can dream of fellows robot artists... Her argument isn't about so much about the speed at which AI can generate images. I believe it has more to do with the cascade of people and companies seeing opportunities in making quick money using the current loophole that AI image generators represent. The image generator itself doesn't have any advantage, the user however can produce and sell images based on other people's work and time. They can bar the way for juniors entering the industry and finding no starting opportunities. In that sense it is absolutely an unfair advantage, even more so, a form of possession of stolen goods. To quote you: "Better to let the technologists do what they do best and develop tools that will improve artist workflows and help democratize the creative process." To which I would reply, better let the lawyers and law makers do what they do best and fight alongside artists so that the millions of artworks on internet benefit from what is due to them: legal protection. Democratizing the creative process can be done with time and commitment, humans haven't waited 2023 and its AI generators to start producing art...