Her language consistently defines LLMs in negative terms like “synthetic text extruder” but she claims she’s not trying to denigrate it. What’s missing for me are similar terms from her about how humans create sentences and thoughts. Judging by the state of the internet humans are quite capable of making shit up to argue their point (see latest Fox News apology). She talks about sycophantic AI but give me a car battery and some cables and I can train a sycophantic human (no I can’t but there are people who can). She’s pretty much a walking counter argument for her own claims.