Not only could the same model no longer be accurate, but furthermore, this polluted content could then be used by the other LLMS for their own training. They, in turn, deposit terabytes of distorted information on the internet (This vicious cycle will eventually ruin both the internet and our ability to use chatbots effectively, I fear). This means... See more