Some AI papers have a finding so good you just want to quote it everywhere. But many turn out to be trash. They’ve got bad statistics, they’ve got bad methodology, they’re really sales pieces, or t…
  • UnderpantsWeevil@lemmy.worldEnglish
    26·
    3 days ago

    … the search engines became crap because of ai

    I mean, search has always been built on some kind of LLM. That’s how you convert a query into a list of page-results.

    We’ve just started trying to wrap the outputs into a natural language response pattern in order to make them feel more definitive and more engaging. The “AI” part of search is mostly window-dressing.

    Plus, ai just lies.

    It has inaccurate heuristics and often tries to back-fill what it can’t find with an approximation in order to maintain user engagement.

    Idk if I’d even call it lying, so much as bullshitting.

    • self@awful.systemsEnglish
      5·
      2 days ago

      I mean, search has always been built on some kind of LLM. That’s how you convert a query into a list of page-results.

      no it fucking hasn’t. the stemming and page ranking algorithms used in traditional search have absolutely nothing to do with LLMs.

      shit, neither stemming nor PageRank as originally defined even have a machine learning component. here’s postgres’ full text search suite, which literally converts a textual query into a list of results (sans page ranking, which is out of scope for a database) in a manner suitable for a production search engine, utterly without any machine learning or other stochastic crap.