Hallucinations are the best evidence that AI is not conscious (outside of basic logic)--that every word output is simply mindless pattern matching with no intent or awareness attached. They arent typos, they are insights into the foundational process of simply matching what's appropriate and demonstrating the inevitable flaws of that process where the content isnt generated so much as approximated to a particular degree of certainty.
perhaps the best distinction for now could be:
generative vs processing, point of origin vs simply being a conduit
[link] [comments]