Arts Commentary: AI Implementation and the Arts — Welcome to Dystopia
By Steve Provizer
There will be no winners in this battle for control, whatever the judicial outcomes. Once the AI bubble bursts, many people will be hurt.
There should be no sympathy for the AI juggernaut. Photo: Artificial Paintings
For some time, I’ve followed AI implementation and its effects on artists (see links below). A recent judgment initially seemed to signal progress for those fighting the wholesale consumption of texts that build the large language models (LLMs) powering AI.
Last year, a suit was brought by novelist Andrea Bartz and nonfiction writers Charles Graeber and Kirk Wallace Johnson against Anthropic (founded in 2021 by former members of OpenAI). The case later represented a larger group of writers and publishers whose books Anthropic downloaded to train its chatbot, Claude. The case, originally headed for a December trial, resulted in a settlement: a pool of authors or publishers will receive about $3,000 each for an estimated 500,000 books in the suit.
At least a dozen other major lawsuits are pending, claiming copyright infringement against OpenAI, Microsoft, Anthropic, Midjourney, Stability AI, Perplexity AI, DeviantArt, Nvidia, and Intel. The Anthropic suit is the first to settle in favor of creators.
The key issue: Anthropic accessed those texts by downloading over 7 million digitized books from illegal pirate websites, rather than buying them. Anthropic has valued itself at $183 billion following another $13 billion in investment. One might assume investors would scrutinize how the data powering Anthropic’s AI was acquired—but apparently not.
Importantly, this case does not address Fair Use, the doctrine behind which AI companies have sought shelter. That question is far thornier and destined for the Supreme Court. Some courts have ruled that certain types of AI training do qualify as fair use, but the decisions are complex and may have limited precedent.
If any AI company has built its LLM’s or image creation models (Disney is one litigant) using pirated materials, major court-ordered damages are likely. Yet, the Supreme Court’s eventual Fair Use ruling will have an even greater impact. The outcome remains uncertain; if Fair Use is deemed inapplicable, it would mean major trouble for the industry.
Be prepared to hear “too big to fail” as the AI industry mobilizes powerful lobbying efforts to preserve Fair Use and protect its trillions in investments. This is a high-stakes game—the stock market depends heavily on tech stock growth, and Big Tech is all in on AI. No one is profiting from AI—except Nvidia, the chipmaker.
Artists should expect no help from Congress, as legislation like Dodd-Frank or agencies like the Consumer Financial Protection Bureau are in the process of being rendered toothless or gutted. Government regulation becomes an oxymoron as corporations are given free rein to swallow competitors and operate unchecked.
The Anthropic decision brings some measure of fairness for creators—and that’s worth celebrating. But I’m not optimistic about the endgame of the struggle between the creative arts and technology. The hope was that dazzling new technology would lead us to utopia; instead, it’s steering us in a decidedly dystopian direction.
LINKs
https://artsfuse.org/254081/music-commentary-the-streaming-cesspool/
https://artsfuse.org/294760/arts-commentary-record-companies-suing-ai-who-owns-input
https://artsfuse.org/303672/book-review-mood-machine-in-the-mood-for-manipulation/
https://artsfuse.org/300101/arts-advocacy-sign-our-petition-an-open-letter-to-generative-ai-leaders/
https://horizonmass.news/2024/12/12/ai-blessing-and-curse/
Steve Provizer writes on a range of subjects, most often the arts. He is a musician and blogs about jazz here.
Tagged: AI, AI implementation, Anthropic, DeviantArt, Microsoft, Midjourney, OpenAI (maker of ChatGPT), Perplexity AI, Stability AI
I agree with Steve that we need to be very skeptical of technology built on piracy not to mention AI’s hoovering of electricity in defiance of its environmental impact. In the meantime, though, something creators can do is to support the organizations fighting this piracy. I’m a member of the Authors’ Guild https://authorsguild.org/advocacy/artificial-intelligence/ adding my modest dues to their collective clout.
I believe Cloudfare offers server space and domains which include the ability to prevent site scraping!
Sandrine-Thanks for your reply. I looked over the Cloudfare site and they seem to be about security, like blocking bots, phishing, etc. But I couldn’t find anything about preventing scraping and, in fact, they’re all in on helping implement AI.
This page has a lot of DIY info on preventing scraping: https://stytch.com/blog/how-to-block-ai-web-crawlers. And there are commercial services like DataDome.
BTW, Spotify has new Terms of Service. They’re proud to say that outside AI can’t use your material, but Spotify can train its OWN AI on your music and press material, not to mention that it can use or change your work as they want, or even assign your rights to someone else.