I've written about "disordered counterpublics" groups that reinforce shared delusions Here, the LLM becomes a 1 person counterpublic. It mirrors assumptions, absorbs contradictions, & provides the sense of a responsive, confirming audience.
It is the architecture of disordered discourse, automated.

I dug into the civil service & Microsoft claims that copilot saves workers 2 weeks a year, and similarly it is based on self-reported estimates. www.wheresrhys.co.uk/2025/06/07/t...

While, rightly, there's a lot of concern about when AI _is_ used without consent, transparency etc, rushing towards the benefits while ignoring the harms... I wonder what these cynical claims of AI-ness might do to people's ability to recognise, believe and embrace benefits when they are real

But I think it's interesting that this AI branding creates associations in people's minds between "cutting edge AI" and "rubbish service delivery", which, whatever you think of AI, is grounded in presence of the "AI" marketing buzzword, rather than presence of the actual technology.

implying that, despite the hype, cutting edge technology is not in fact being used to power them. "No shit, Sherlock," I hear you say, "Didn't you know it's all hype??!"...