I had a couple of experiences where I suspected I was hearing LLM-generated/edited text being read aloud. It was at two different webinars about that were about roadmaps or case studies about some products that I use. It was a bit uncanny because I could detect the stylistic patterns ("It's not X, it's Y" and "No X, no Y, just Z"), but it was kind of jarring to see them spoken by a person on a video call. It makes me think this kind of pattern might be engaging, but for a lot of people, it now sticks out for the wrong reasons.

Once LLM generated speech or content start getting into the live answers of Q&A sessions, that would be sad. I know some people try to get through interviews, but I think that might be a bit harder to not detect.

> It was a bit uncanny because I could detect the stylistic patterns ("It's not X, it's Y" and "No X, no Y, just Z"),

That's just marketing-speak. LLMs sound like that because LLMs were trained on marketing-speak.