Whose data do we train the algorithms with?
Concerned that increasingly sophisticated algorithms that can create illustrations based on a simple text description and the likelihood of finding themselves redundant, a number of artists have highlighted a small clause in the terms of use of Adobe Creative Cloud, which gives subscribers access to graphic design, video editing, web design and cloud services programs. By agreeing to the clause, common in many such services, users give Adobe permission to analyze their content, specifically through the use of machine learning techniques such as pattern recognition, in order to improve the company’s products and services.
This option, enabled by default, can be disabled by the user, but as is often the case, the few people pay it any attention, partly out of idleness and partly because it generates a certain perception that the product you use may be, thanks to your contributions and those of many other users, improving over time. Sure, but what happens when this “improvement” is helping develop algorithms capable of eventually replacing the artists themselves?
The use of assistants such as DALL·E, Midjourney, Stable Diffusion and others in creation is generating many concerns among artists. On the one hand, experience shows that their creations published on the web have been obtained through web scraping and used in many cases to feed the…
