Is A.I. Art Stealing from Artists?
The New Yorker – According to the lawyer behind a new class-action suit, every image that a generative tool produces “is an infringing, derivative work.” By Kyle Chayka – “…Last month, McKernan joined a class-action lawsuit with two other artists, Sarah Andersen and Karla Ortiz, filed by the attorneys Matthew Butterick and Joseph Saveri, against Midjourney and two other A.I. imagery generators, Stable Diffusion and DreamUp….
Butterick and Saveri allege that what A.I. generators do falls short of transformative use. There is no transcending of the source material, just a mechanized “blending together,” Butterick said. “We’re not litigating image by image, we’re litigating the whole technique behind the system.” The litigators are not alone. Last week, Getty Images filed a lawsuit against Stable Diffusion alleging that the generator’s use of Getty stock photography amounts to “brazen infringement . . . on a staggering scale.”
Whatever their legal strengths, such claims possess a certain moral weight. A.I. generators could not operate without the labor of humans like McKernan who unwittingly provide source material. As the technology critic and philosopher Jaron Lanier wrote in his 2013 book “Who Owns the Future?,” “Digital information is really just people in disguise.”
(A spokesperson from Stability AI, the studio that developed Stable Diffusion, said in a statement that “the allegations in this suit represent a misunderstanding of how generative A.I. technology works and the law surrounding copyright,” but provided no further detail. Neither DeviantArt, which owns DreamUp, nor Midjourney responded to requests for comment.)”
The Verge: “The chatbots are out in force, but which is better and for what task? We’ve compared Google’s Bard, Microsoft’s Bing, and OpenAI’s ChatGPT models with a range of questions spanning common requests from holiday tips to gaming advice to mortgage calculations.
Naturally, this is far from an exhaustive rundown of these systems’ capabilities (AI language models are, in part, defined by their unknown skills — a quality dubbed “capability overhang” in the AI community) but it does give you some idea about these systems’ relative strengths and weaknesses.