Twelve copyright lawsuits pitting authors and publishers against OpenAI, creator of ChatGPT, and Meta (formerly Facebook) have been consolidated to one. The purpose is to avoid duplication of trial preparation by the defendants who created artificial intelligence (AI) programs, OpenAI and Meta, and conflicting verdicts. The combined suits will be tried in federal court for the Southern District of New York.
The issue concerns the creation of Large Language Models, search programs that use material from copyrighted works that provide answers to questions through the use of artificial intelligence. You may have asked questions or made requests of OpenAI's ChatGPT or Meta's Llama. Within seconds, they will give you often fairly detailed answers to those questions. How do they know those answers? The answer is they do a crash course (a few seconds) to review compendious amounts of writing and then give you their response. It is all amazing, but that still brings us back to the question of where do they get all that information? The answer is often from books or other written material that is copyrighted. What really peeves the authors and content creators is they pay nothing for the works they have used.
It's no surprise that the content creators have gone to court to stop the practice. They want royalties. However, the resources these AI searchers use are enormous. The cost and complications to paying the appropriate royalties to each source in proportion to their use would be overwhelming. We might lose access to these incredible tools if copyright law was applied, but that doesn't exactly justify using someone's copyrighted work without payment. This is a case where rules developed in earlier times don't work that well today, but a practical way of updating them to fit today's times is not easy to fashion.
For now we have a multitude of lawsuits designed to protect the financial interests of authors and publishers in the courts. This latest ruling will at least prevent having numerous suits filed in different courts which may reach conflicting verdicts. That would be even messier. Now there is one, but whatever the verdict, it may be less than perfect from the standpoint of protecting both the writers legitimate financial interests and the public interest in making information available.
