LLM-written content is flooding the web, and people are tired of it
On The Flash C Compiler (50352701), the top comment says the Adafruit post is an LLM recap of the author's blog post, which is an LLM summary of a project vibe coded with Claude, and that 'the signal:noise ratio is dropping off exponentially.' On Anthropic's 'The Components of Agent Memory' (50402212), commenters called it marketing and objected to language implying Claude has experiences. One said it is too much to read, 'emotionally'.
A related thread asks HN to add a flag for AI-generated articles (48886741). The responses show why it's hard: detectors are unreliable, voters often can't tell, and some fear it will just increase the stigma around LLMs. HN's guidelines already ban generated text in comments.
The pattern: output is cheap, so attention and provenance are the scarce things. Readers now look for who actually did the work.
So what?
If you do content marketing, AI-generated posts will cost you credibility with technical buyers. Show the human who did the work and the real artifact, such as code, data or a benchmark. For vendors writing 'research' about their own models, expect hostile reading.