I'm still not convinced that the current state of large language models is a net positive for society. We're sacrificing depth for breadth and nuance is getting lost in the process.