AI slop is now a verification problem, not a taste problem
An AI-assisted paper about Z-Library cited a TorrentFreak article that doesn't exist. A teardown of how Block Blast harvests players for ad money had a reader admit they got halfway through before realizing it was AI-written. An Ask HN asked for a flag for AI-generated articles, and a commenter noted that HN's new guideline ("Don't post generated text or AI-edited text") covers comments but is unclear about submissions. The Dessa speech-only conversation model drew a mix of awe and unease: realistic voice, "2024-level" intelligence.
The sharpest comment was about the citation case. A citation is a verifiable claim about another document, so the failure is that the workflow let it pass as verified, not that a model invented it. Others replied that academic publishing was a clown car before AI.
The counterpoint is that detection doesn't work. AI checkers are unreliable, voting can be gamed, and one commenter said a flag could add stigma against people who use LLMs responsibly.
So what?
Trust is becoming the scarce thing. If you ship content, research or citations with AI in the loop, add a verification step that checks sources automatically. That is also a product opportunity for anyone building provenance and fact-checking tools.
Read these
AI-Assisted Paper on Z-Library Cites a TorrentFreak Article That Doesn't Exist
How Block Blast harvests its players for ad money
Ask HN: Add flag for AI-generated articles
Dessa: Conversation Model trained only on speech