Beehaw
  • Communities
  • Create Post
  • search
    Search
  • Login
  • Sign Up
David Gerard ( dgerard@awful.systems ) M to TechTakes@awful.systemsEnglish · 2 years ago

LLMs still don't understand the word "no", much like their creators

www.quantamagazine.org

external-link
message-square
22
link
fedilink
115
external-link

LLMs still don't understand the word "no", much like their creators

www.quantamagazine.org

David Gerard ( dgerard@awful.systems ) M to TechTakes@awful.systemsEnglish · 2 years ago
message-square
22
link
fedilink
AI Like ChatGPT Are No Good at ‘Not’ | Quanta Magazine
www.quantamagazine.org
external-link
Today’s language models are more sophisticated than ever, but they still struggle with the concept of negation. That’s unlikely to change anytime soon.
alert-triangle
You must log in or register to comment.
  • froztbyte ( froztbyte@awful.systems ) 
    link
    fedilink
    English
    arrow-up
    29
    ·
    2 years ago

    it’s almost like this thing has no internal conceptual representation! I know this can’t possibly be, millions of promptfans and prompfondlers have told me it can’t be so, but it sure does look that way! wild!

    • Kogasa ( kogasa@programming.dev ) 
      link
      fedilink
      English
      arrow-up
      2
      ·
      2 years ago

      It must have some internal models of some things, or else it wouldn’t be possible to consistently make coherent and mostly reasonable statements. But the fact that it has a reasonable model of things like grammar and conversation doesn’t imply that it has a good model of literally anything else, which is unlike a human for whom a basic set of cognitive skills is presumably transferable. Still, the success of LLMs in their actual language-modeling objective is a promising indication that it’s feasible for a ML model to learn complex abstractions.

      • sc_griffith ( sc_griffith@awful.systems ) 
        link
        fedilink
        English
        arrow-up
        21
        ·
        2 years ago

        if I copy a coherent sentence into my clipboard, my clipboard becomes capable of consistently making coherent statements

        • Kogasa ( kogasa@programming.dev ) 
          link
          fedilink
          English
          arrow-up
          3
          ·
          2 years ago

          Yes, but that’s not how LLMs work. My statement depends heavily on the fact that a LLM like GPT is coaxed into coherence by unsupervised or semi-supervised training. That the training process works is the evidence of an internal model (of language/related concepts), not just the fact that something outputs coherent statements.

          • sc_griffith ( sc_griffith@awful.systems ) 
            link
            fedilink
            English
            arrow-up
            12
            ·
            2 years ago

            if I have a bot pick a random book and copy the first sentence into my clipboard, my clipboard becomes capable of consistently making coherent statements. unsupervised training 👍

          • adderaline ( ondoyant@beehaw.org ) 
            link
            fedilink
            English
            arrow-up
            12
            ·
            2 years ago

            this isn’t necessarily true. patterns in data aren’t by nature proof of an underlying system of logic. if you run the line-fitting machine on any kind of data, its going to output a line. considering just how much data is encoded into these transformers, i don’t think we can conclusively say that it has a underlying conception of how language works, much less an understanding of the concepts that language represents. it could really just be using the vast quantities of data it has to output approximately correct statements. there’s absolutely structure there, but it doesn’t have to have the kind of structured understanding humans have about language to produce language, in the same way a less sophisticated machine learning model doesn’t have to know what kind of data its fitting a line to to make a line.

          • self ( self@awful.systems ) 
            link
            fedilink
            English
            arrow-up
            9
            ·
            2 years ago

            let me free up some of your time so you can go figure out how LLMs actually work

      • flere-imsaho ( mawhrin@awful.systems ) 
        link
        fedilink
        English
        arrow-up
        13
        ·
        2 years ago

        it doesn’t. that’s why we’re calling it “spicy autocompletion” .

        • Kogasa ( kogasa@programming.dev ) 
          link
          fedilink
          English
          arrow-up
          1
          ·
          2 years ago

          It does, which is why it’s autocompletion and not auto-gibberish.

      • slopjockey ( slopjockey@awful.systems ) 
        link
        fedilink
        English
        arrow-up
        13
        ·
        2 years ago

        It must have some internal models of some things, or else it wouldn’t be possible to consistently make coherent and mostly reasonable statements.

        Talk about begging the question

  • Backspacecentury ( Backspacecentury@kbin.social ) 
    link
    fedilink
    arrow-up
    21
    ·
    2 years ago

    Ha, I love the sauce on that headline.

    • iAmTheTot ( iAmTheTot@kbin.social ) 
      link
      fedilink
      arrow-up
      6
      ·
      2 years ago

      It’s not the headline used by the publication.

      • David Gerard ( dgerard@awful.systems ) OP
        link
        fedilink
        English
        arrow-up
        17
        ·
        2 years ago

        yes, this is the anti-HN

  • i_love_FFT ( i_love_FFT@lemmy.ml ) 
    link
    fedilink
    English
    arrow-up
    12
    ·
    2 years ago

    it seems like it’s not the worst way to write text if I don’t want to allow an ai to parse my messages…

    • David Gerard ( dgerard@awful.systems ) OP
      link
      fedilink
      English
      arrow-up
      12
      ·
      2 years ago

      not being not sure to fail to not write like this could become the opposite of interesting after a time that isn’t long, though

      • i_love_FFT ( i_love_FFT@lemmy.ml ) 
        link
        fedilink
        English
        arrow-up
        8
        ·
        2 years ago

        Wow… It’s not easy trying not to misunderstand sentences…

  • Omniraptor ( Omniraptor@lemm.ee ) deleted by creator
    link
    fedilink
    English
    arrow-up
    3
    ·
    edit-2
    2 years ago

    deleted by creator

    • self ( self@awful.systems ) 
      link
      fedilink
      English
      arrow-up
      9
      ·
      2 years ago

      we don’t care that your instance of a nondeterministic, unreliable system can’t replicate someone else’s results, and we don’t take marching orders from SSC readers.

  • Kogasa ( kogasa@programming.dev ) 
    link
    fedilink
    English
    arrow-up
    1
    ·
    2 years ago

    deleted by creator

  • intensely_human ( intensely_human@lemm.ee ) BannedBanned
    link
    fedilink
    English
    arrow-up
    1
    ·
    2 years ago

    I challenge anyone here to define the word “no” without looking it up

    • self ( self@awful.systems ) 
      link
      fedilink
      English
      arrow-up
      13
      ·
      2 years ago

      oh goody, the containment thread is leaking

      I won’t even try to guess what the point of this horseshit was but no, fuck off

      • froztbyte ( froztbyte@awful.systems ) 
        link
        fedilink
        English
        arrow-up
        5
        ·
        2 years ago

        kinda hoping it isn’t my fault…

        • self ( self@awful.systems ) 
          link
          fedilink
          English
          arrow-up
          7
          ·
          2 years ago

          you have cursed my mansion with ghosts but here I am with a vacuum and a flashlight sucking up both spirits and treasure

          I may or may not have Luigi’s Mansion on the brain. how much is a gamecube flash conversion again?

          • froztbyte ( froztbyte@awful.systems ) 
            link
            fedilink
            English
            arrow-up
            4
            ·
            2 years ago

            You know, that first paragraph has magnificent haiku potential

TechTakes@awful.systems

techtakes@awful.systems

Subscribe from Remote Instance

Create a post
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: !techtakes@awful.systems

Big brain tech dude got yet another clueless take over at HackerNews etc? Here’s the place to vent. Orange site, VC foolishness, all welcome.

This is not debate club. Unless it’s amusing debate.

For actually-good tech, you want our NotAwfulTech community

Visibility: Public
globe

This community can be federated to other instances and be posted/commented in by their users.

  • 75 users / day
  • 354 users / week
  • 945 users / month
  • 2.06K users / 6 months
  • 56 local subscribers
  • 2.79K subscribers
  • 1.61K Posts
  • 37.2K Comments
  • Modlog
  • mods:
  • David Gerard ( dgerard@awful.systems ) 
  • BE: 0.19.21
  • Modlog
  • Legal
  • Instances
  • Docs
  • Code