I don’t like reading obviously AI-generated content on Twitter. There’s a derogatory term for it: AI “slop”1, which means something like “AI content presenting itself as human”, or even just “unwanted AI content”. But I have no problem reading AI-generated content when I talk to Copilot or ChatGPT. Why is that? Having a good theory of AI slop is important for a few reasons. First, it’s just an interesting question why AI responses feel okay or gross in different contexts. Second, if you’re build...