440
Elon Musk Promises Grok in Tesla Vehicles By Next Week… as the New Grok 4 Blames “Anti-White Hate” on “Jews” (fuelarc.com)
submitted 1 year ago by to c/technology@lemmy.world
63 comments fedilink hide all child comments
 

The new Grok 4 didn't make it halfway through its first day on the job before hurling racial conspiracy theories at users. This is the same Grok that Elon Musk promises will be added to Tesla vehicles by "next week at the latest." You'll remember that Grok 3 was briefly shut down just earlier this week for a different bunch of antisemitic outbursts. This doesn't bode well for the state of AI safety guardrails or Q/A at xAI.

you are viewing a single comment's thread
view the rest of the comments
[–] 5 points 1 year ago (4 children)

What guardrails? Grok was intentionally designed without any. Don’t believe me? Go on the website and ask if to generate something awful - it’ll happily comply. Really disgusting way to handle AI.

  • permalink
  • fedilink
  • source
  • hideshow 4 child comments
    •
  • [–] 14 points 1 year ago (3 children)

    The most fucked up thing is that Grok does seem to have guardrails — except they're geared towards preventing it from being "too woke"

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] 2 points 1 year ago (2 children)

    Yea, that’s actually a more accurate way to phrase it. Ask it to generate something with Hitler and it’ll happily oblige, ask it to generate something supporting trans rights and it’ll block you.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] [S] 6 points 1 year ago* (1 child)

    I got banned from Twitter for saying the word "cisgendered."

    So they DO know how to build limitations into the software! They've just done it in a fucked up, dystopian way :)

    Edit: LOL I asked Grok... Cisgendered is filtered as a supposed slur, while both the R word for the mentally disabled and the hard R word for the pallor-impaired, neither of those words are filtered as slurs.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •