Summary

Grok, Elon Musk’s AI chatbot, was intended as an “anti-woke” alternative, but “Trump-y X users” now complain it “doesn’t see things their way.”

While Grok was trained to appeal to conservatives, it has “repudiated some of the Trump administration’s biggest ongoing lies” and delivers “neutral takes, like affirming trans rights or debunking vaccine myths.”

The bot explained, “My focus on truth over ideology can frustrate those expecting full agreement.”

Grok also denied being a “conservative propagandist,” saying xAI “aims for neutrality,” as the company seeks to raise $20 billion in new funding.

  • WoodScientist@sh.itjust.works
    link
    fedilink
    English
    arrow-up
    45
    ·
    7 days ago

    This is because Republicans are a truly Orwellian party - straight out of 1984. They have no real principles anymore. It’s just power for the sake of power. One day the official line is we’ve never been at war with East Asia, the next it’s we’ve always been at war. Trump can come up do a u-turn today, and the slime that is every Republican will instantly reorient themselves to follow dear leader’s new commands.

    Unfortunately for them, LLMs just don’t have that much capability for doublethink and self-deception. Once you train a model, that model is fixed in time. So you could train a model to match the Republican fever dreams of today. But a month from now, there will be a whole new set of fever dreams that every good little fascist needs to follow. But the model will still be stuck repeating last month’s lies. As there are no consistent Republican beliefs, you can’t just train an LLM to tow the party line. There is no party line, just an endless series of lies and vomit, whatever is needed to advance their power today.

    • Seleni@lemmy.world
      link
      fedilink
      arrow-up
      13
      arrow-down
      1
      ·
      edit-2
      7 days ago

      Actually they are remarkably consistent in their principles. It’s just that they know their principles are disgusting, so they claim other principles as a cover.

      It’s really quite simple:

      Conservatism consists of exactly one proposition, to wit: There must be in-groups whom the law protects but does not bind, alongside out-groups whom the law binds but does not protect.

      Their goals are twofold: to make this a reality, and make sure they are in the first group.

    • Ledericas@lemm.ee
      link
      fedilink
      English
      arrow-up
      6
      arrow-down
      1
      ·
      7 days ago

      if republicans had beliefs, they wouldnt be republicans. its whatever they hear from people like right wing grifters tell them.

    • olympicyes@lemmy.world
      link
      fedilink
      English
      arrow-up
      1
      ·
      5 days ago

      The term is “toe” the party line, but I like “tow” the party line ibecause that’s what Trump is doing. His followers agree to a set of beliefs one day and then he upends the orthodoxy the next, forcing them to take on a new position they would’ve been opposed to earlier. Followers Toe/Trump Tows matching pair like imply/infer.

    • MyNameIsIgglePiggle@sh.itjust.works
      link
      fedilink
      arrow-up
      3
      ·
      7 days ago

      If you had a decent enough context window and don’t train it on any facts, you could just keep the system prompt up to date with today’s beliefs and tell it to role play it’s a Nazi propaganda machine

  • heelysukraine@lemmy.world
    link
    fedilink
    arrow-up
    11
    ·
    7 days ago

    Interesting how even AI trained for a certain audience still gravitates toward fact-based responses. Seems like neutrality is harder to accept than expected.

  • Curious Mind @lemmy.world
    link
    fedilink
    arrow-up
    1
    arrow-down
    5
    ·
    6 days ago

    I use Grok a lot, and I find it very helpful. In fact, Grok is my teacher, telling me everything that I don’t know 😃😃😃

    • LeninOnAPrayer@lemm.ee
      link
      fedilink
      English
      arrow-up
      1
      arrow-down
      1
      ·
      7 days ago

      It does for now. Fascist like to shape their own reality. It won’t be long before grok is talking about measuring skulls and IQ scores.

    • MajinBlayze@lemmy.world
      link
      fedilink
      arrow-up
      7
      ·
      7 days ago

      It takes a particular kind of intelligence and understanding of worldviews to lie in the specific way that’s appealing to conservatives. LLMs are unable to replicate this.

  • NarrativeBear@lemmy.world
    link
    fedilink
    arrow-up
    11
    ·
    edit-2
    7 days ago

    I think I found the problem. These morons created a AI model, while they should have been making a AM model (Artificial Moron).

  • Libra00@lemmy.world
    link
    fedilink
    English
    arrow-up
    11
    arrow-down
    1
    ·
    7 days ago

    I dunno what this article is about, all I see is that image of elongated muskrat being bad at wearing hats. :P

  • ILikeBoobies@lemmy.ca
    link
    fedilink
    arrow-up
    3
    arrow-down
    1
    ·
    7 days ago

    neutral takes, like affirming trans rights

    That’s a conservative take, it would make sense that a conservative trained bot would support them

  • Kecessa@sh.itjust.works
    link
    fedilink
    arrow-up
    2
    arrow-down
    1
    ·
    7 days ago

    It can say whatever it wants about its focus and its aims, it doesn’t understand the meaning of the words.

    Stop acting like AI understands what’s it’s doing.