Previous discussion of the quoted post: https://news.ycombinator.com/item?id=44052246 https://news.ycombinator.com/item?id=44052246 > [...] LLMs are going to

Previous discussion of the quoted post: https://news.ycombinator.com/item?id=44052246 https://news.ycombinator.com/item?id=44052246 > [...] LLMs are going to open a whole new world of intimate surveillance I think that's right. I'm wondering what the main driving factors are. Users tend to be much more personal with LLM assistants. And LLMs which are not trained to be assistants also also good at cross-linking and understanding human writing outside the prompt. So, while the post is not about this, I think that LLMs probably increase the power of AI algorithms that process human writing outside of assistant conversations. Maybe even the ones used for de-anonymization and building interest/personality cohorts?