Google rolled out AI overviews across the United States this month, exposing its flagship product to the hallucinations of large language models.

  • trollbearpig@lemmy.world
    link
    fedilink
    English
    arrow-up
    16
    arrow-down
    2
    ·
    edit-2
    6 months ago

    I’m probably late, but in this case this is the combinations of 2 things.

    1. The usual capitalistic incentives ruined yet another company. There was a recent article about how Google pushed out the people who builded and maintaned search on favor of MBA growth focused assholes. Like they put the guy that was Yahoo’s CEO while Yahoo search was crumbling, in charge of Google search to get him to increase the amount of searches they serve, and ads obviously. People keep suggesting to use DDG, or Kagi, or some other comercial product. And for now, we must because Google is basically useless right now. But just give time to the other companies to fall in the same trap hahaha.
    2. LLMs are not smart, not even close. They are just a parlor trick that has non technical people fooled. There is a lot of evidence to me, but to me the most obvious one is that they don’t have anything resembling human short term memory. Like the way they make them look like they are having a conversation is by providing the entire conversation up to that point, including their own previous responses lol, as input/context so the bot autocompletes the conversation. It literally can’t remember a single word of what you said on it’s own. But sureee, they are just like humans lol.

    So what we have here is obvious, we have a company trying to grow like cancer by any means necessary. And now they have a technology that allows them to create enough smoke and mirrors to fool non technical people. Sadly, as part of this they are also destroying the last places of the internet not fully controlled by corporations. Let’s hope lemmy survives, but it’s just a matter of time before they flood this place too.