The Myth of AI Democratization
Some still believe that training large language models (LLMs) on copyrighted content is a form of “democratizing knowledge.” But when you look closely at how these models actually handle the material they ingest, the picture looks a lot less heroic – and a lot more extractive.
