434
Sam Altman admits OpenAI ‘totally screwed up’ its GPT-5 launch and says the company will spend trillions of dollars on data centers (finance.yahoo.com)
submitted 1 year ago by to c/technology@lemmy.world
163 comments fedilink hide all child comments
 

“I literally lost my only friend overnight with no warning,” one person posted on Reddit, lamenting that the bot now speaks in clipped, utilitarian sentences. “The fact it shifted overnight feels like losing a piece of stability, solace, and love.”

https://www.reddit.com/r/ChatGPT/comments/1mkumyz/i_lost_my_only_friend_overnight/

you are viewing a single comment's thread
view the rest of the comments
[–] 67 points 1 year ago (34 children)

It annoys me that Chat GPT flat out lies to you when it doesn’t know the answer, and doesn’t have any system in place to admit it isn’t sure about something. It just makes it up and tells you like it’s fact.

  • permalink
  • fedilink
  • source
  • hideshow 34 child comments
    •
  • [–] 25 points 1 year ago

    It doesn‘t know that it doesn‘t know because it doesn‘t actually know anything. Most models are trained on posts from the internet like this one where people rarely ever just chime in to admit they don‘t have an answer anyway. If you don‘t know something you either silently search the web for an answer or ask.

    So since users are the ones asking ChatGPT, the LLM mimics the role of a person that knows the answer. It only makes sense AI is a „confidently wrong“ powerhouse.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 41 points 1 year ago (12 children)

    LLMs don't have any awareness of their internal state, so there's no way for them to see something as a gap of knowledge.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 12 child comments
    •
  • [–] 34 points 1 year ago* (1 child)

    Took me ages to understand this. I'd thought "If an AI doesn't know something, why not just say so?“

    The answer is: that wouldn't make sense because an LLM doesn't know ANYTHING

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] -1 points 1 year ago (9 children)

    Wouldn't it make sense for an ai to provide a confidence level though?

    I've got 3 million bits of info on this topic but only 4 of them lead to this solution. Confidence level =1.5%

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 9 child comments
    •
  • [–] 24 points 1 year ago (2 children)

    It doesn't have "3 million bits of info" on a specific topic, or even if it did, it wouldn't be able to directly measure it. It's worth reading a bit about how LLMs work behind the hood, because although somewhat dense if you're new to the concepts, you come out knowing a lot more about what to expect when using them, what the limitations actually are and how to use them better if you decide to go that route.

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

    You could do this with logprobs. The language model itself has basically no real insight into its confidence but there's more that you can get out of the model besides just the text.

    The problem is that those probabilities are really "how confident are you that this text should come next in this conversation" not "how confident are you that this text is true/accurate." It's a fundamental limitation at the moment I think.

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

    It's always funny to me when people do add 'confidence scores' to LLMs, because it always amounts to just adding 'say how confident you are with low, medium or high in your response' to th prompt, and then you have made up confidences for made up replies. And you can tell clients that it's just made up and not actual confidence, but they will insist that they need it anyways…

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

    And you can tell clients that it's just made up and not actual confidence, but they will insist that they need it anyways…

    That doesn’t justify flat out making shit up to everyone else, though. If a client is told information is made up but they use it anyway, that’s on the client. Although I’d argue that an LLM shouldn’t be in the business of making shit up unless specifically instructed to do so by the client.

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

    I'm not really sure I follow.

    Just to be clear, I'm not justifying anything, and I'm not involved in those projects. But the examples I know concern LLMs customized/fine-tuned for clients for specific projects (so not used by others), and those clients asking to have confidence scores, people on our side saying that it's possible but that it wouldn't actually say anything about actual confidence/certainty, since the models don't have any confidence metric beyond "how likely is the next token given these previous tokens" and the clients going "that's fine, we want it anyways".

    And if you ask me, LLMs shouldn't be used for any of the stuff it's used for there. It just cracks me up when the solution to "the lying machine is lying to me" is to ask the lying machine how much it's lying. And when you tell them "it'll lie about that too" they go "yeah, ok, that's fine".

    And making shit up is the whole functionality of LLMs, there's nothing there other than that. It just can make shit up pretty well sometimes.

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

    It doesn’t store bits of information. All it has are neurons that form a weighted network

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 10 points 1 year ago

    In the end it's a word generator that has been trained so much it uses facts often enough to be convincing. That's its basic architecture.

    You can ask it to give a confidence level to have an indication of how sure it is of the answer.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 9 points 1 year ago (2 children)

    It wouldnt finish a lyric for me yesterday because it was copyrighted. I sid it was public domain and it said "You are absolutely right, given its release date it is under copyright protection"

    Wtf

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 7 points 1 year ago

    Someone I know (not close enough to even call an "internet friend") formed a sadistic bond with chatGPT and will force it to apologize and admit being stupid or something like that when he didn't get the answer he's looking for.

    I guess that's better than doing it to a person I suppose.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 7 points 1 year ago (4 children)

    It's a feature. Not a bug of LLMs.

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

    It’s neither. It’s a design flaw. They’re not designed to be able to handle this type of situation correctly

    You out there spreading misinformation, saying they’re a manipulation tool. No, they were never invented for this.

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

    Llm is just next word prediction. The Ai doesn't know whether the output is correct or not. If it's wrong or right. Or fact or a lie.

    So no I'm not spreading misinformation. The only thing that might spread misinformation is the AI here.

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

    That’s actually one thing that got significantly improved with GPT-5, fewer hallucinations. Still not perfect of course

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 1 point 1 year ago* (last edited 1 year ago) (1 child)

    And depending on how OpenAI tweaked it this time it will either realize its mistake after being made aware of it or double down even harder on it.

    I only use it for coding and it once told me my code not working was due to a bug in Webkit, so I asked it which bug specifically. It created links to bug reports but rewrote the titles of them. So initially it looked like it had numerous sources that backed up its statement but when I clicked on them those were bugs about totally different things.

    It would not back down even after I specifically told it "You just made all of this shit up and even rewrote the titles" and got stuck in a loop of "I'm sorry, but you're wrong and I am 100% sure I haven't made a mistake".

    Kinda creepy. Especially when you think about the system rewriting reality when it comes to much more important things. Let's just reinvent some history, that would be a good idea, right?

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

    I sometimes approach this like I do with students. Using your example, I’d ask it to restate the source, then ask it to read the title of that source directly. If it’s correct, I might ask it to briefly summarize what the source article covers. Then I would ask it to restate what it told me about the source earlier, and to explain where the inconsistency lies. Usually by this time, the AI is accurately pointing out flaws in its prior logic. At that point I ask again if it is 100% sure it didn’t make a mistake, and it might actually concede to having been wrong. Then I tell it to remember how and why it was wrong to avoid similar errors in the future. I don’t know if it actually works, but it makes me feel better about it.

  • permalink
  • fedilink
  • source
  • parent
    • load more comments (1 reply)