starting out[0] with “I was surprised by the following results” and it just goes further down almost-but-not-quite Getting It Avenue

close, but certainly no cigar

choice quotes:

Why is it impressive that a model trained on internet text full of random facts happens to have a lot of random facts memorized? … why does that in any way indicate intelligence or creativity?

That’s a good point.

you don’t fucking say

I have a website (TrackingAI.org) that already administers a political survey to AIs every day. So I could easily give the AIs a real intelligence test, and track that over time, too.

really, how?

As I started manually giving AIs IQ tests

oh.

Then it proceeds to mis-identify every single one of the 6 answer options, leading it to pick the wrong answer. There seems to be little rhyme or reason to its misidentifications

if this fuckwit had even the slightest fucking understanding of how these things work, it would be glaringly obvious

there’s plenty more, so remember to practice stretching before you start your eyerolls

  • mozz
    link
    fedilink
    18
    edit-2
    4 months ago

    I miss the days when GPT would make an explicit point within a decent fraction of its answers that it was only a large language model, and not a general purpose intelligence, because those are two very very different (if very similar-seeming to initial human perception) things.

    It seems that the inexorable tide of misperception that that was a futile attempt to forestall has come in.

    • @froztbyteOP
      link
      English
      154 months ago

      that probably got “strategically removed” for uhhhhhhh checks notes financial reasons

      (read: almost certainly some execs made the call to get that nuked, because it didn’t fit the narrative they’re tried to sell)

      • Sonori
        link
        fedilink
        English
        134 months ago

        What your telling me that a company run by the crypto shill behind worldcoin might be bending their technology to create the appearance of progress and inflate their own value? Say it ain’t so.

    • @200fifty
      link
      English
      114 months ago

      I feel like it was all over from the moment they made it talk in first person. No one had any illusions that Inferkit or NovelAI were general intelligences, because it was obvious that they were just language models autocompleting a sentence you typed in.

  • @froztbyteOP
    link
    English
    164 months ago

    and of course the domain is fucking maximum truth dot org

    • @froztbyteOP
      link
      English
      114 months ago

      (idly: I didn’t immediately notice whether this is one of the quantified clueless, so not sure if it should be on sneerclub instead. but that domain set me wondering basically immediately)

      • @froztbyteOP
        link
        English
        84 months ago

        oh, they’re mentioned at the bottom, so it is indeed that crowd

  • @swlabr
    link
    English
    164 months ago

    On their substack the author claims they are “doing non-ideological […] reporting”, which means they are definitely doing ideological reporting. Let’s see…

    next most recent post titled “the dawn of woke ai” says pretty much what you’ll guess it does from the title. It also features the AI rendered POC nazis from a bit back as evidence of “woke”ness… Fun!

    • @Evinceo
      link
      English
      174 months ago
      • SubStack
      • “Maximum Truth”

      can I discard someone’s opinion twice?

      • @Soyweiser
        link
        English
        114 months ago

        “The Dawn of Woke AI”

        Can I do it thrice?

    • @froztbyteOP
      link
      English
      10
      edit-2
      4 months ago

      I didn’t even go exploring. on the one hand I feel this is good (in the sense that it didn’t depress me even more), on the other it would’ve made the who clear earlier

  • @Amoeba_Girl
    link
    English
    144 months ago

    Another way of putting it: Out of 196 questions, ChatGPT-4 got about 5 more correct answers than a random guesser would (39 vs 34.23.)

    What are the odds of that?

    I’m too lazy to look through the tests he’s administering, but IQ tests like the WAIS have vocabulary questions, which yes you would expect an LLM to be better at than random chance.

    I’ve surely said it before but when you see the sort of thinking on display by Mr Max Truth here, is it any wonder why rationalists are impressed with ChatGPT’s reasoning faculties.

    • @Amoeba_Girl
      link
      English
      174 months ago

      I asked ChatGPT-4 if cars in roundabouts in Ireland go clockwise or counterclockwise. It got it wrong. When I told it that, it apologized and gave the right answer. But then I trickily called it out on its right answer, and it apologized again and reverted to the wrong answer. Fundamentally, it knows that the Irish drive on the left side of the road, but it doesn’t understand how to apply that to a roundabout to find the circular direction.

      lol you fucking idiot

      • @selfA
        link
        English
        154 months ago

        this coin I’m flipping fundamentally knows everything about how the Irish drive, but it only seems to feel like giving me the right answer approximately half the time

        this reminds me of very early in my programming career, when I discovered that an NPC I programmed to randomly either move forward or turn left every 10 seconds was surprisingly good at solving simple labyrinths. I used to instantiate like 100 of them and see which ones would win (or “fight” by colliding with each other, or escape the labyrinth by stacking on top of other instances). you’re telling me now I was a handful of incredibly stupid blog posts away from being a renowned AI researcher?

        • @swlabr
          link
          English
          134 months ago

          I used to instantiate like 100 of them and see which ones would win (or “fight” by colliding with each other

          The basilisk will not take kindly to your desecration of AGI for sport.

  • @jonhendry
    link
    English
    94 months ago

    “On my Substack I am doing non-ideological, data-driven reporting!”

    I don’t believe the son of pro-gun propagandist John Lott is capable of doing non-ideological reporting.