Meta Claims Feeding Copyrighted Books into AI Is Fine Since They Lack "Economic Value"

Meta has faced accusations of improperly utilizing copyrighted content to educate its artificial intelligence systems, and the company’s argument in response doesn’t hold much weight.

In the ongoing suit Richard Kadrey and others vs. Meta Platforms Led by a team of authors that includes Pulitzer Prize recipient Andrew Sean Greer and National Book Award winner Ta-Nehisi Coates, the corporation backed by Mark Zuckerberg contends that it has been wrongly accused of Scraping more than 7 million books From the pirated library LibGen, this practice was considered "fair use" of the content and hence was deemed legal.

The flawed arguments don't stop there. Vanity Fair spotlights In a fresh filing, Meta's lawyers are additionally contending that the numerous books owned by the corporation Used to educate its multibillion-dollar language models and propel itself into the extremely competitive AI race are essentially useless.

Meta cited an expert witness who downplayed the books' individual importance, averring that a single book adjusted its LLM's performance "by less than 0.06 percent on industry standard benchmarks, a meaningless change no different from noise."

Thus there's no market in paying authors to use their copyrighted works, Meta says, because "for there to be a market, there must be something of value to exchange," as quoted by Vanity Fair — "but none of [the authors'] works has economic value, individually, as training data." Other communications showed that Meta employees stripped the copyright pages from the downloaded books.

This exemplifies the deceptive tactics and hypocritical reasoning employed by Meta, along with the broader AI sector, whenever they face scrutiny over the vast amount of user-generated content they utilize.

In some way, that thing holds little value at the same time, and perhaps we should cease exaggerating over the purity of artistic expression; besides, an AI. composes imaginative prose just as skillfully as a person can now — but it is also completely crucial for constructing our new synthetic deities that will solve climate change , so please don't charge us for utilizing any of it. The final part is exactly what OpenAI said. contested before the British Parliament Last year—it was argued that there aren't sufficient resources in the public domain to enhance its AI models. Therefore, it needs to be permitted access to current copyrighted materials without having to pay royalties.

Apparently, there is an unstated agreement among leading AI firms. When a Meta researcher asked whether the company’s legal department had approved through LibGen; another replied, “Though I didn’t inquire further, this is akin to how OpenAI utilizes GPT-3, Google employs PaLM, and DeepMind works with Chinchilla, hence we’ll proceed similarly.” Vanity Fair from internally cited messages in the lawsuit.

Significantly, the unspoken rule appears to be complete silence on the matter.

"Under no circumstances should we reveal publicly that we have trained using LibGen; nonetheless, there is a realistic possibility that outside entities might infer our utilization of this dataset," stated an internal Meta presentation slide. The document stated that "should media reports indicate that we have utilized datasets known to be pirated, like those from LibGen, this could weaken our bargaining stance with regulatory bodies regarding these matters."

More on AI copyright: OpenAI Warns It’s “Done For” If It Can’t Use All Your Copyrighted Material

The post Meta Claims It's Acceptable to Use Copyrighted Books for Their AI Models Since They Lack "Economic Value" appeared first on .

Berbagi :