Generative AI datasets could face a reckoning | The AI Beat
August 22, 2023
Over the weekend, a bombshell story from The Atlantic found that Stephen King, Zadie Smith and Michael Pollan are among thousands of authors whose copyrighted works were used to train Meta’s generative AI model, LLaMA, as well as other large language models, using a dataset called “Books3.” The future of AI, the report claimed, is “written with stolen words.”
The truth is, the…