As the world’s most critical chipmaker faces a lawsuit over allegedly training its artificial intelligence model about a work protected by copyrightsaid Nvidia’s deputy general counsel during a press conference chip manufacturer AI technology conference does not see the extension of intellectual property rights to the creation of generative models of artificial intelligence.
During the conference session devoted to ethical technological challenges today (March 18), Nvidia’s Iain Cunningham noted that the US Copyright Office has already defined content generated by AI models cannot be copyrighted.
“Intellectual property law exists to protect human intellectual endeavor,” Cunningham said during the session. “That’s how it’s always been formulated. I think this is probably not going to change in my opinion, and the result is that it will be more arduous to determine what is the intellectual contribution of a human to a particular work and what is the contribution of a machine. “
Robots cannot “own”. [intellectual] property’
Cunningham said he didn’t believe intellectual property rights would be extended to AI models because “it’s not a established human intellectual effort” and that it “doesn’t make sense” for AI models to “own property.”
“More fundamentally, the whole intellectual property regime exists as an incentive for people to make things, and you don’t necessarily have to incentivize machines to make things,” Cunningham said. “That’s why I don’t see intellectual property extending to machine works.”
Cunningham, however, said he sees an “increasingly arduous time” for courts and other decision-makers in determining what parts of an AI-generated work can be protected under intellectual property law and what parts of an AI-generated work will not be protected. Cunningham said users input their own sketches into the AI models, which can complicate matters even further.
“But on the other hand, I think it will be much easier to create novel things,” Cunningham said. “[T]came up with the idea that you have to take someone else’s work and appropriate it for your own exploit… there’s less incentive to do so when you can easily create something yourself.
Nvidia’s copyright problems
Earlier this month, Nvidia was sued by the authors in a class action lawsuit in California for allegedly training artificial intelligence on their copyrighted works. The authors claimed that their books were part of a dataset of nearly 197,000 books used to build Nvidia’s NeMo AI platform. Part of the book collection on which NeMo was trained included a dataset of unlicensed, copyrighted books from the so-called shadow librarythat stores and distributes unlicensed copyrighted material.
In October, the dataset was declared defunct due to alleged copyright infringement. The authors are seeking unspecified damages for U.S. authors whose copyrighted books were used to train NeMo over the past three years.
