• kingthrillgore@lemmy.mlEnglish
    991·
    il y a 5 mois

    I spent most of today looking at places to rent in Denver and I come home to Google having killed it’s fucking search engine. What the hell is going on

  • RGB3x3@lemmy.worldEnglish
    75·
    il y a 5 mois

    Google has been bad for a long time, but they’ve shut the bed so hard lately. Seriously, look at this:

    I actually run out of screenshot space before I can get to an actual regular search result!

    • voracitude@lemmy.worldEnglish
      1936·
      il y a 5 mois

      On the one hand, generative AI doesn’t have to give deterministic answers i.e. it won’t necessarily generate the same answer even when asked the same question in the same way.

      But on the other hand, editing the HTML of any page to say whatever you want and then taking a screenshot of it is very easy.

      • Otter@lemmy.caEnglish
        421·
        il y a 5 mois

        It could also be A/B testing, so not everyone will have the AI running in general

          • Otter@lemmy.caEnglish
            331·
            il y a 5 mois

            Wouldn’t they be? They could measure how likely it is that someone clicks on the generated link/text

            • credo@lemmy.worldEnglish
              219·
              il y a 5 mois

              Just because you click on it that doesn’t make it accurate. More importantly, that text isn’t “clickable”, so they can’t be measuring raw engagement either.

              • IllNess@infosec.pubEnglish
                31·
                il y a 5 mois

                What this would measure is how long you would stay on the page without scrolling. Less scrolling means more time looking at ads.

                This is the influence of Prabhakar Raghavan.

              • RvTV95XBeo@sh.itjust.worksEnglish
                2·
                il y a 5 mois

                Just because you click on it that doesn’t make it accurate.

                Given the choice between clicks/engagement and accuracy, is pretty clear Google’s for the former is what got us into this hell hole.

                • sugar_in_your_tea@sh.itjust.worksEnglish
                  1·
                  il y a 5 mois

                  Yup, if you have to repeat your search 3 times, you’re seeing 3x the ads. If you control most of the market, where are your customers going to go? Most will just deal with it and search more.

          • halcyoncmdr@lemmy.worldEnglish
            24·
            il y a 5 mois

            Google runs passive A/B testing all the time.

            If you’re using a Google service there’s a 99% chance you’re part of some sort of internal test of changes.

      • QuadratureSurfer@lemmy.worldEnglish
        21·
        il y a 5 mois

        Technically, generative AI will always give the same answer when given the same input. But, what happens is a “seed” is mixed in to help randomize things, that way it can give different answers every time even if you ask it the same question.

        • jyte@lemmy.worldEnglish
          8·
          il y a 5 mois

          What happened to my computers being reliable, predictable, idempotent ? :'(

            • jyte@lemmy.worldEnglish
              1·
              il y a 5 mois

              Technically they still are, but since you don’t have a hand on the seed, practically they are not.

              • QuadratureSurfer@lemmy.worldEnglish
                1·
                il y a 5 mois

                OK, but we’re discussing whether computers are “reliable, predictable, idempotent”. Statements like this about computers are generally made when discussing the internal workings of a computer among developers or at even lower levels among computer engineers and such.

                This isn’t something you would say at a higher level for end-users because there are any number of reasons why an application can spit out different outputs even when seemingly given the “same input”.

                And while I could point out that Llama.cpp is open source (so you could just go in and test this by forcing the same seed every time…) it doesn’t matter because your statement effectively boils down to something like this:

                “I clicked the button (input) for the random number generator and got a different number (output) every time, thus computers are not reliable or predictable!”

                If you wanted to make a better argument about computers not always being reliable/predictable, you’re better off pointing at how radiation can flip bits in our electronics (which is one reason why we have implemented checksums and other tools to verify that information hasn’t been altered over time or in transition). Take, for instance, the example of what happened to some voting machines in Belgium in 2003: https://www.businessinsider.com/cosmic-rays-harm-computers-smartphones-2019-7

                Anyway, thanks if you read this far, I enjoy discussing things like this.

                • jyte@lemmy.worldEnglish
                  1·
                  il y a 5 mois

                  You are taking all my words way too strictly as to what I intended :)

                  It was more along the line : Me, a computer user, up until now, I could (more or less) expect the tool (software/website) I use in a relative consistant maner (be it reproducing a crash following some actions). Doing the same thing twice would (mostly) get me the same result/behaviour. For instance, an Excel feature applied on a given data should behave the same next time I show it to a friend. Or I found a result on Google by typing a given query, I hopefully will find that website again easily enough with that same query (even though it might have ranked up or down a little).

                  It’s not strictly “reliable, predictable, idempotent”, but consistent enough that people (users) will say it is.

                  But with those tools (ie: chatGPT), you get an answer, but are unable to get back that initial answer with the same initial query, and it basically makes it impossible to get that same* output because you have no hand on the seed.

                  The random generator is a bit streached, you expect it to be different, it’s by design. As a user, you expect the LLM to give you the correct answer, but it’s actually never the same* answer.

                  *and here I mean same as “it might be worded differently, but the meaning is close to similar as previous answer”. Just like if you ask a question twice to someone, he won’t use the exact same wording, but will essentially says the same thing. Which is something those tools (or rather “end users services”) do not give me. Which is what I wanted to point out in much fewer words :)

        • lucas@fitt.au
          10·
          il y a 5 mois

          @RecursiveParadox @voracitude it absolutely has become a meme, there are (or were) a bunch of repeatable results.

          Google is probably whack-a-mole’ing them now, because “google’s AI search results are trying to kill people” has entered the collective consciousness.

          • vimdiesel@lemmy.worldEnglish
            3·
            il y a 5 mois

            I have no doubt some of their AI answers have antivax and injecting bleach recommendations from all over the web as part of their training regime.

        • thegreatgarbo@lemmy.worldEnglish
          2·
          il y a 5 mois

          If you read the arstechnica article Google is correcting these errors on the fly so the search results can change rapidly.

    • dutchkimble@lemy.lolEnglish
      7·
      il y a 5 mois

      But the real question is, is the colour blue that you see, the colour blue that I see?

      • markon@lemmy.worldEnglish
        32·
        il y a 5 mois

        I don’t get the doom but idk I have been watching this stuff closely for over a decade. I think it’s exciting and people are having All these strange expectations out of these systems all the sudden just because they’re smart. Well they were smart before any of this generative AI stuff. Also scientific breakthroughs in medicine, blind people have something that can assist them. As someone with some disabilities, and knowing a lot of people who also have disabilities, it seems to be the privilege of the healthy and comfortable to keep the status quo.

        Also if we want to play that game we were so fucked by climate change already that I had no hope. Now I have a little. It’s not going away so let’s push for open open open free software. (And model weights)

        • vimdiesel@lemmy.worldEnglish
          3·
          il y a 5 mois

          we aren’t talking about using it to -benefit- human society like discovering new proteins or vaccines. We’re talking about it fucking up search results on google and generating billions of new sites with fucking spam. It’s a tool, but it’s being completely misused and ruining the internet.

  • weew@lemmy.caEnglish
    39·
    il y a 5 mois

    Well, we know Google won’t get rid of this.

    They’ll only cancel it after it actually works and becomes useful

  • M0oP0o@mander.xyzEnglish
    38·
    il y a 5 mois

    We really need a whole community just for the very funny AI errors like this. I could spend all day reading about leaving a dog in a hot car, jumping off a bridge and eating at least one rock a day.

  • Infynis@midwest.socialEnglish
    35·
    il y a 5 mois

    I asked Google for the release date of the new Final Fantasy XIV expansion today, which comes out June 28th. It told me March 26th

      • andrew@lemmy.stuart.funEnglish
        4·
        il y a 5 mois

        But not from a knowledge engine. It makes sense if some rando just spouted off a date from the top of their head but this is the former world leader in knowledge capture and search.

  • Lad@reddthat.comEnglish
    34·
    il y a 5 mois

    What the hell is going on with Google search? Has it completely shit itself after the AI implementation? I know its been bad for a while but this is another level.

    • lemmyvore@feddit.nlEnglish
      27·
      il y a 5 mois

      Short answer, yes. The ratio of LLM generated noise to actual content is increasing exponentially as we speak. To us it seems overnight because the increase is so steep but it’s been happening for several years. And it’s going to get a lot worse.

      Honestly, I think we’ll have to go back to 90s methods like web rings and human curated link directories.

      • masterofn001@lemmy.caEnglish
        5·
        il y a 5 mois

        I remember having to buy a book. A book with URLs. Before search engines existed.

        Good times.

        On Google: I’ll never forgive them for getting rid of their pseudo secret government search google.com/unclesam where I found a lot of .mil docs I probably shouldn’t have been able to.

      • PenisWenisGenius@lemmynsfw.comEnglish
        5·
        il y a 5 mois

        I do random hobby tinkering and search results have become so useless that I’m having to read a lot more books. Everything takes longer this way.

  • dependencyinjection@discuss.tchncs.deEnglish
    25·
    il y a 5 mois

    I stopped using Google years ago. I started using Bing but had to stop that as it would divert me to MSN to sign in when clicking a link for a news article. Like a news article for The Independent or The Times or any other.

    I then started using DuckDuckGo which is powered by Bing, but found it wasn’t great at many searches.

    I now use Arc Search most of the time and click browse for me to get the information I want without the bullshit. Search is essentially dead due to greed.

      • drspod@lemmy.mlEnglish
        5·
        il y a 5 mois

        I used Qwant for a few days and then it popped up a modal dialog asking me to turn off my ad-blocker. Never used it again after that.

        • Buffalox@lemmy.worldEnglish
          1·
          il y a 5 mois

          Funny I’ve never seen that, but I switch around from time to time. Because none are perfect unfortunately.

          • vimdiesel@lemmy.worldEnglish
            1·
            il y a 5 mois

            kagi is the best but it will cost you $10 a month. It’s been worth it to me, but probably not to everyone.

            • Buffalox@lemmy.worldEnglish
              1·
              il y a 5 mois

              Interesting concept, but it’s a bit expensive IMO, considering the huge amount of “free” options.

              The pricing is only in USD without taxes. Listing the price excluding tax is illegal here (Denmark and I think the rest of EU), so apparently not a service meant for use outside USA.
              Ad free is not a problem for me, i use Firefox with µBlock Origin.

        • Buffalox@lemmy.worldEnglish
          3·
          il y a 5 mois

          Startpage uses outside search results, but should be very secure like DuckDuckGo, and better certified.
          Qwant is AFAIK more independent, and I like the layout better.

          Both give pretty good search results IMO, but are somewhat lacking in map/geographic searches. For instance searching “Angola” could result in a restaurant in London. Just as a hypothetical example.
          So I do use Google maps too.

        • vimdiesel@lemmy.worldEnglish
          2·
          il y a 5 mois

          qwant = 95% bing results, startpage = 98% google results. They are slimmer. They don’t keep your search history or ip address. better for privacy but not much better for search results.

      • vimdiesel@lemmy.worldEnglish
        2·
        il y a 5 mois

        qwant uses bing and is mostly a proxy for it. Startpage is a proxy for google. the only thing they really do is protect your privacy, they don’t give you better search results.

        • Buffalox@lemmy.worldEnglish
          1·
          il y a 5 mois

          That’s a bit disappointing, I thought Qwant had their own search. I know startpage used Google originally, but I wasn’t sure if they still did.

      • exanime@lemmy.todayEnglish
        21·
        il y a 5 mois

        Tried start page after DDG outage on Thursday and I’m liking it a lot

    • psycho_driver@lemmy.worldEnglish
      11·
      il y a 5 mois

      I’m finding SearXNG to be very good. It operates like dogpile used to but is actually functional and it pretty much entirely squelches product placement results. I actually have to manually go to google if I want to get product listings for something.

      • dependencyinjection@discuss.tchncs.deEnglish
        3·
        il y a 5 mois

        Seems to only have an iOS app right now. There is a desktop version for Windows and MacOS Here.

        It is annoying that it wants to send you the download link by email though.

        Edit: Here is an example of how the results look when asking for opening times of a book store in my city.

  • vimdiesel@lemmy.worldEnglish
    151·
    il y a 5 mois

    how do you guys get these things? I never see any summaries like that. I wonder if one of my adblockers is killing google AI lmao. Do you have to be logged into your google account? I never log into google any more.

  • NutWrench@lemmy.worldEnglish
    13·
    il y a 5 mois

    Behold the wonders of AI! Now, we don’t have to pay human beings to edit webpages for us! Thanks to AI, you can just sit back and watch the money roll in!