You knew it was coming: Google begins testing AI-only search results
arstechnica.com
AI Mode could be the future of Google, but it’s currently just an experiment.

Google has become so integral to online navigation that its name became a verb, meaning “to find things on the Internet.” Soon, Google might just tell you what’s on the Internet instead of showing you. The company has announced an expansion of its AI search features, powered by Gemini 2.0. Everyone will soon see more AI Overviews at the top of the results page, but Google is also testing a more substantial change in the form of AI Mode. This version of Google won’t show you the 10 blue links at all—Gemini completely takes over the results in AI Mode.

This marks the debut of Gemini 2.0 in Google search. Google announced the first Gemini 2.0 models in December 2024, beginning with the streamlined Gemini 2.0 Flash. The heavier versions of Gemini 2.0 are still in testing, but Google says it has tuned AI Overviews with this model to offer help with harder questions in the areas of math, coding, and multimodal queries.

With this update, you will begin seeing AI Overviews on more results pages, and minors with Google accounts will see AI results for the first time. In fact, even logged out users will see AI Overviews soon. This is a big change, but it’s only the start of Google’s plans for AI search.

Gemini 2.0 also powers the new AI Mode for search. It’s launching as an opt-in feature via Google’s Search Labs, offering a totally new alternative to search as we know it. This custom version of the Gemini large language model (LLM) skips the standard web links that have been part of every Google search thus far.

    • AbsentBird@lemm.eeEnglish
      1·
      2 months ago

      I don’t get what they gain out of this, doesn’t it remove the sponsored results? Or will people pay to nudge AI weights to recommend their products?

      • werefreeatlast@lemmy.worldBannedEnglish
        4·
        2 months ago

        Example:

        How do you make the best cookies diy?

        …the new Nabisco Cookies can be crushed to a fine powder mixed…

        What are other non-nabisco cookies I can cook?

        …Nestle chocolate chip cookies of course!

  • Rooty@lemmy.worldEnglish
    31·
    2 months ago

    I’m decrepit enough to remember pre-Google web, with competing search engines. Bring back webrings!

  • Majorllama@lemmy.worldBannedEnglish
    882·
    2 months ago

    Would this be the same AI that spits out incorrect summarys or just flat out gets data incorrect constantly? What could possibly go wrong by making that the ONLY way to look for things going forward. Fucking morons.

    • wjrii@lemmy.worldEnglish
      6·
      2 months ago

      I recently had a dream that involved a suburb of Green Bay, Wisconsin. I have never been to Green Bay Wisconsin. I know it as a rather small city that is the home to the Green Bay Packers, an administratively anachronistic NFL team that draws a large plurality of its fan base from the greater Milwaukee area. Off the top of my head, I don’t know if Green Bay has “suburbs” in the usual American sense at all.

      I googled the name of this completely nonexistent community, along with the words “Green Bay,” and the AI very confidently hallucinated it into existence, describing it as a lovely shopping and residential area just over the bridge of the same name.

      • Majorllama@lemmy.worldBannedEnglish
        62·
        2 months ago

        Yup. No problem though. In the future AI can just add the now very real town to you AR eyeballs brought to you by Chipotle Exxon and T mobile.

    • Artyom@lemm.eeEnglish
      2·
      2 months ago

      No, this will be the AI that ignores your content and just spits out sponsored content. AI doesn’t really have to be involved much, but if it is, Google doesn’t have to pay up when you file a class action lawsuit for misleading content because they can say “It’s not our fault, thr AI did it.”

  • ShittyBeatlesFCPres@lemmy.worldEnglish
    51·
    2 months ago

    Alphabet/Google needs to fire their CEO. He’s an obvious idiot, not good with employees, not good with investors, and not good at lobbying. That’s like 99% of a CEO’s job. Just get rid of him and Google’s stock price will probably jump 20%.

  • hark@lemmy.worldEnglish
    25·
    2 months ago

    Stick to SEARCH, not making shit up (euphemistically called “hallucinating”).

  • GaMEChld@lemmy.worldEnglish
    145·
    2 months ago

    I mean their search results have gone to absolute shit anyway, the AI is probably better at this point.

    • Rooty@lemmy.worldEnglish
      51·
      2 months ago

      My coworker who has an absolute hard-on for ChatGPT has tried several times in vain to get it to say the correct answer to a question I knew the answer to. LLMs are bullshit engines and toys for the malicious.

  • Computerchairgeneral@fedia.io
    14·
    2 months ago

    Thanks Google, I hate it. At least the udm=14 trick and website still works, at least until google decides to stop supporting that feature. Definitely going to be using that more and more if this becomes the default google experience.

    • Bloomcole@lemmy.worldEnglish
      123·
      2 months ago

      They did that long before AI.
      Google, Apple and plenty others have regular meetings with defense people to accomodate them.
      No surprise Eric Schmit left Google to form a defense company since some employees there had a consciense and didn’t like working on killer drone programs.

      • andros_rex@lemmy.worldEnglish
        292·
        2 months ago

        To borrow Alex Jones’ term - it is information warfare. Train people to accept the outputs of LLMs, attack media literacy (in conjunction with the fact that public schools in the US spent ten years not teaching children how to read), and hide actual sources.

        You can have your LLM trained to weasel around the fact that the chattel slavery of black Americans was the primary cause of the Civil War, or falsely represent science surrounding gender and sexuality (they’re explicitly tuning Grok on this).

        You don’t have to worry about subversive messages on war or women in your “AI” generated “Ghibhli” film. Have all the fun space faring of Star Wars, without the prequel geopolitics everyone hated until they saw it happen.

          • andros_rex@lemmy.worldEnglish
            4·
            2 months ago

            I’m stoned so:

            The technocrats (Bezos, Thiel, Zuckerberg, etc) are SEELE. Musk is Gendo - ie, at the table but rogue (and I don’t want to imagine what he’d do with a planet full of Yue/Rei’s) Thiel doesn’t trust Musk, so Vance is like his Kaji. (Misato is like literally a Freud caricature of a woman, so we can make her a couch to fit the whole sexual ethos)

            COVID was the second impact. The angels are attendant disasters that go with it and things like climate change etc.

            Luigi is Kaworu, a gorgeous bisexual twink who is being crushed by the system.

            The American people are Shinji, paralyzed under stress and trauma. Asuka is a bunch of other countries, being very angry at Americans for not doing shit while not really coping with their own problems.

            Rei is obviously the internet/technology.

            The globalists have been priming us for this information the entire time! Pre programming! Please buy my protein power.

  • Apoplexy@lemmy.worldEnglish
    22·
    2 months ago

    How many cases of “You should eat at least one small rock per day” AI bullshit does Google think is acceptable for production deployment?

    • Sizbang@lemmy.worldEnglish
      21·
      2 months ago

      Wait…you shouldn’t be doing that? Damnit! What am I going to do with the small rock mine I just bought?!

    • negativeone@lemmy.worldEnglish
      2·
      2 months ago

      I was just dealing with a plumbing problem the other day, and I was like, “I know, I’ll ask AI!”

      Basically everything it told me was completely wrong. And this was Claude Sonnet 3.7, one of the best!

  • Delta_V@lemmy.worldEnglish
    312·
    2 months ago

    this sounds like what Google wanted to be since its inception: ask a question, get an answer

    somewhere along the journey, the reality of needing to make money drove the enshitification of search results: ask a question, get offers to sell you an answer

    this seems like a step in a better direction

    • Baggins [he/him]@lemmy.caEnglish
      10·
      2 months ago

      Right because when I search for factual information what I really want is an LLM to tell me “sorry Dave I can’t answer that question right now.” (See: any election related information. As in it doesn’t just kick you back to the web search results it just straight up refuses to give you the answer.)

      • Thistlewick@lemmynsfw.comEnglish
        9·
        2 months ago

        It baffles me how wholly some people are just accepting the complete erosion of the internet. I have lecturers at university, so theoretically educated people, who tell us point blank to plug any questions we have about weekly topics into ChatGPT. The respect I have for these teachers and their content is less than zero at this point.

        • Chris@feddit.ukEnglish
          4·
          2 months ago

          Then you have to check those answers, so you need to search for an authoritative source anyway… which means you need a regular search engine. At that point you may as well have used the search engine in the first place.

          If Google does this, you’ll need to find an alternative search engine to check Google against… so you may as well just switch to a different search engine in the first place.

          There are things that LLMs are good at, being a search engine isn’t one of them. Although I have asked searchy type questions and got some interesting links back which I probably wouldn’t have found on a normal web search with the terms I was using, so they can be useful as a supplementary search tool. I’d rather that than it just giving the answers, which then need to be fact checked elsewhere.

          • uranibaba@lemmy.worldEnglish
            3·
            2 months ago

            At that point you may as well have used the search engine in the first place.

            I was going to respond to this but I think you did so yourself:

            I have asked searchy type questions and got some interesting links back which I probably wouldn’t have found on a normal web search with the terms I was using

            I think they work as supplements and not replacements. As any tool, they have their use and (for me) can enhance my searching. But I would not replace it with only LLM. (Altough I have never had any great luck with ChatGPT and links, they never work - as in ChatGPT give me an anchor element without any link. It’s better at providing me search terms and concepts to look up for what I need.)

  • Chris@feddit.ukEnglish
    407·
    2 months ago

    Hmm, in theory I don’t have a problem with an AI telling me the answer to my question or whatever I’m searching for - but this isn’t a web search. If I’m actually searching for a particular page or context, then I want to be able to do that.

    These are two entirely different things, and if Google goes down this route they aren’t a search engine anymore - they are an LLM provider.

    • Voytrekk@sopuli.xyzEnglish
      233·
      2 months ago

      Couldn’t agree more. I use both tools as they have their place, but when I want a search engine, I don’t want to get an AI answer for the result. Like if I’m searching for documentation on something, what benefit would the AI provide me over just giving the link?