705
Car Wash Test on 53 leading AI models: "I want to wash my car. The car wash is 50 meters away. Should I walk or drive?" (opper.ai)
submitted 8 months ago by to c/technology@lemmy.world
334 comments fedilink hide all child comments
 

Screenshot of this question was making the rounds last week. But this article covers testing against all the well-known models out there.

Also includes outtakes on the 'reasoning' models.

you are viewing a single comment's thread
view the rest of the comments
[–] 1 point 8 months ago (5 children)

I mean, we're watching it happen. I don't think it's hypothetical anymore.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 5 child comments
    •
  • [–] 2 points 8 months ago (4 children)

    I'm sorry but no, models are definitely not collapsing. They still have a million issues and are subject to a variety of local optima, but they are not collapsing in any way. It is not known whether this can even happen in large models, and if it can it would require months of active effort to generate the toxic data and fine-tune models on that data. Nobody is gonna spend that kind of money to shoot themselves in the foot.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 4 child comments
    •
  • [–] 1 point 8 months ago (3 children)

    Then why are newer versions of the major models performing so poorly? For instance, GPT 5.2 is definitely not an improvement over 4.5. What's the root cause?

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] 2 points 8 months ago (2 children)

    The switch you mention (from 4th gen to 5th gen GPT) is when they introduced the model router, which created a lot of friction. Basically this will try to answer your question with as cheap a model as possible, so most of the time you won't be using flagship 5.2 but a 5.2-mini or 5.2-tiny which are seriously dumber. This is done to save money of course, and the only way to guarantee pure 5.2 usage is to go through the API where you pay for every token.

    There's also a ton of affect and personal bias. Humans are notoriously bad at evaluating others intelligence, and this is especially true of chatbots which try to mimic specific personalities that may or may not mesh well with your own. For example, OpenAI's signature "salesman & bootlicker" personality is grating to me and i consistently think it's stupider than it is. I've even done a bit of double blind evaluation on various cognitive tasks to confirm my impression but the data really didn't agree with me. It's smart, roughly as smart as other models of its generation, but it's just fucking insufferable. It's like i see Sam Altman's shit eating grin each time i read a word from ChatGPT, that's why i stopped using it. That's a property of me, the human, not GPT, the machine.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 1 point 8 months ago (1 child)

    It goes beyond the problems introduced by the model router, though. I have to work with GPT 5.2 for my job (along with Claude, Gemini, and a few others), and we have enterprise API access to it. So when I select GPT 5.2 as the model to use, it's spending tokens to actually use it.

    And it's pretty bad. It's noticeably worse than the 4.x series. I find myself having to fix its mistakes far more often.

    I've struggled to reason out an explanation, and model collapse really seems like a contender, especially if you follow information theory and why training these things is so hard.

    As it happens, there's a new talk about exactly this from George D. Montañez. You might find it interesting: https://youtu.be/ShusuVq32hc

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 1 point 8 months ago

    OK that's a fair observation. Honestly my naive guess would be that they simply do not optimize mainline gpt models for the kind of use case you generally have on Api (tool use, multi-step actions, etc...). They need it to be a perky every day assistant not necessarily a reliable worker. Already on gpt-4 i found it extremely mediocre compared to the Claude models of the same time.

    I think that's a more likely explanation than model collapse which is a really drastic phenomenon. A collapsed model will not just fail tasks at a higher rate, it will spit garbled text and go completely off the rails, which would be way more noticeable. It would also be weird that Claude models keep getting better and better while they're probably fed roughly the same diet of synthetic data.

  • permalink
  • fedilink
  • source
  • parent