336
I'm just trying to search for a website... (thelemmy.club)
submitted 7 months ago by to c/fuck_ai@lemmy.world
81 comments fedilink hide all child comments
 

Been trying to get away from DuckDuckGo because of their support of "AI" stuff, but come on, one of these isn't even in English, and that's what I have my language set to.

Search result from Searx Belgium for "Piefed" that instead shows multiple results for ChatGPT, with one of them appearing to be in Spanish.
you are viewing a single comment's thread
view the rest of the comments
[–] 9 points 7 months ago (7 children)

SearXNG is a mixed bag to be honest...I find that Ononoki's SearXNG delivers decent results...It's a bit mixed at times with PieFed being the third option. Some topics are surfaced better than others, so your mileage will vary with SearXNG.

Using the No AI version of DuckDuckGo sends them a message that their bet on "AI" isn't going to win them any accolades. It's an easy way to push back. As they are tracking those metrics and people can see the numbers of how many users disapprove of AI.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 7 child comments
    •
  • [–] 6 points 7 months ago (6 children)

    Bing and yahoo are VERY bad, it seems they detect that they are being used through proxies, I made another comment for what i use in searxng.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 6 child comments
    •
  • [–] 4 points 7 months ago (4 children)

    DDG's backend is really just Bing, though ...

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 4 child comments
    •
  • [–] 5 points 7 months ago (3 children)

    They give different results, it seems they have different treatment to the data, which is interesting. Startpage gives different results than google, duckduckgo gives different results than bing, but on the bigger picture, a bigger sample gives the statistically optimal result??? (closer to the source truth on the most probable good result) question mark, huge number theory on statistics.

    (i'm drunk)

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] 3 points 7 months ago (1 child)

    Bing in a sandboxed session yielded results identical to DDG in my experiments.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 2 points 7 months ago

    In searxng they don't give the same results, bing brings some completely different stuff both on an selfhosted instance and on an public instance (it just gives random garbage), it seems proxying does some stuff to bings internal algorithm, as google is different than startpage on searxng.

    Edit: a screenshot showing it.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 2 points 7 months ago

    Given the same methodology (search algorithm) a larger sample size gives a truer picture of what that methodology favors. If the methodology isn't neutral or doesn't bias in the same way as your desired outcome then a smaller sample size analyzed by a different methodology may provide better results.

  • permalink
  • fedilink
  • source
  • parent