5 ms·

Same experience. ChatGPT hallucinated library features which are not there and I believe to have found the source for that, an E-Mail in a mailing list misunder

Same experience. ChatGPT hallucinated library features which are not there and I believe to have found the source for that, an E-Mail in a mailing list misunderstanding a feature and asking for clarification. Furthermore the gas lighting is real and my biggest frustration. Asking about a feature in a program, ChatGPT assuring me it is there, me following the instruction, not finding said feature in said menu and ChatGPT assuring me "Ohh, you must be using a newer version, where the feature was removed". Only it was never there. Happened like 3 times already.

opinionsonly01 4y ago
[flagged]
ethbr0 4y ago
> "Ohh, you must be using a newer version, where the feature was removed". Only it was never there. To be fair, that's a decent approximation of the response humans give when confronted with evidence their memory doesn't match reality.
- layer8 4y ago
  It may be fair, but still frustratingly inadequate for what we would want to use it for.
  - midoridensha 4y ago
    They're trying to create bots that act like humans. Well, it seems they're succeeding beyond anyone's wildest dreams. The bots are so human-like, they do all the same shitty stuff that real humans do. Computers were better in the old days, when they were simple and logical and not human-like at all.
    - zmmmmm 4y ago
      this has always been my theory. We will never achieve or acknowledge true AI until it has replicated all of our flaws, at which people everyone will say "that's totally useless, why would anybody want that?"
      - prox 4y ago
        At which point Q steps in to pass a verdict on us.
        midoridensha 4y ago
        I think Q was right about us all along. That episode's conclusion only made sense because the humans in that fantasy timeline are nothing like real-life humans. And of course, that's why ST:TNG is such a pleasure to watch (except for a few stinker episodes): it shows humans you'd really want to hang out with and work with, instead of the backstabbing rats that exist in real life (as you see on much more realistic shows like "Game of Thrones").
        prox 4y ago
        Yeah, it seems idealism is not in vogue, especially the kind of idealism that wants to see people as professional, in command of their faculties, yet not distant or aloof. It seems people have trouble outgrowing their tribal egotistical manners and behaviors. It’s not at all bad, in general many people work together in a global complex society, but, oh boy, lots of evolution to be had if we want a better, more kinder world. My current pet peeve is the emotional reactivity it elicits of any social media in its worst forms.
        midoridensha 4y ago
        > in command of their faculties As great and optimistic as ST:TNG was, don't forget that in the later seasons, they basically resorted to "which crew member goes crazy in this episode?", I guess because the writers ran out of ideas.
MBCook 4y ago
I’m truly curious: why would you ask ChatGPT such a thing in the first place? I don’t understand why some people seem eager to use it as a search engine or replacement for a Stack Overflow or Quora type site.
- ianmcgowan 4y ago
  I’ll bite - because 80% of my searches it returns a convenient answer that matches my expectations of what the answer will look like. I still have to validate (there’s still an objective reality to be satisfied) but that’s pretty good. 20% of the time I have to google, but that’s made more efficient by using specific terms mentioned by chatGPT - getting the right terms in a google search is half the battle. It does seem archaic already for google to give me a list of links and I have to go through them all manually deciding if they answer the query.
  - MBCook 4y ago
    Huh. Thanks.
  - vba616 4y ago
    Could you give any concrete examples?
  - testrun 4y ago
    How do you know what part of the answer is wrong?
    - dizhn 4y ago
      I asked chatgpt to give me the equivalent "ss" command to "netstat -nap". It did and it was correct because I already knew the answer and tried it anyway. Then I asked the equivalent of "netstat -rn" (if any), it gave me "ss -r" something, which is wrong. So you either know enough to spot the error or you verify all of it. Another example. I had a long session with it where I was trying to get it to write a gitlab ci config file where the artifacts to be deployed to Netlify were already present at the root of the repo. As an addition it had a requirement to use 'sed' to rewrite a domain name recursively. This is the first and only time chatgpt was useful to me although it made a lot of mistakes that needed correcting.It was more useful than a google search because I wasn't able to find the correct format with it as most examples assumed a build step. One thing that was frustrating is how I would give it requirements and it would incorporate them only to lose a prior requirement when fixing a latter one. I ended up hand fixing one of the later configs rather than using the final one it gave me. For example I needed to use a node image because netlify-cli was required (the only custom thing necessary from the image) but it would revert to using an Alpine image and point to a now archived repo with a nonexistent release url for gitlab's cli. When it fixed something else it would break the find .. sed combination and add nonsensical things like *. html to it rather than working on all files recursively. In the end, through a combination of AI helping and the user (me) chiming in and trying to reason about it I was able to get somewhere. Not ideal but not useless.

opinionsonly01 4y ago

[flagged]

ethbr0 4y ago

> "Ohh, you must be using a newer version, where the feature was removed". Only it was never there. To be fair, that's a decent approximation of the response humans give when confronted with evidence their memory doesn't match reality.

layer8 4y ago
It may be fair, but still frustratingly inadequate for what we would want to use it for.
- midoridensha 4y ago
  They're trying to create bots that act like humans. Well, it seems they're succeeding beyond anyone's wildest dreams. The bots are so human-like, they do all the same shitty stuff that real humans do. Computers were better in the old days, when they were simple and logical and not human-like at all.
  - zmmmmm 4y ago
    this has always been my theory. We will never achieve or acknowledge true AI until it has replicated all of our flaws, at which people everyone will say "that's totally useless, why would anybody want that?"
    - prox 4y ago
      At which point Q steps in to pass a verdict on us.
      - midoridensha 4y ago
        I think Q was right about us all along. That episode's conclusion only made sense because the humans in that fantasy timeline are nothing like real-life humans. And of course, that's why ST:TNG is such a pleasure to watch (except for a few stinker episodes): it shows humans you'd really want to hang out with and work with, instead of the backstabbing rats that exist in real life (as you see on much more realistic shows like "Game of Thrones").
        prox 4y ago
        Yeah, it seems idealism is not in vogue, especially the kind of idealism that wants to see people as professional, in command of their faculties, yet not distant or aloof. It seems people have trouble outgrowing their tribal egotistical manners and behaviors. It’s not at all bad, in general many people work together in a global complex society, but, oh boy, lots of evolution to be had if we want a better, more kinder world. My current pet peeve is the emotional reactivity it elicits of any social media in its worst forms.
        midoridensha 4y ago
        > in command of their faculties As great and optimistic as ST:TNG was, don't forget that in the later seasons, they basically resorted to "which crew member goes crazy in this episode?", I guess because the writers ran out of ideas.

MBCook 4y ago

I’m truly curious: why would you ask ChatGPT such a thing in the first place? I don’t understand why some people seem eager to use it as a search engine or replacement for a Stack Overflow or Quora type site.

ianmcgowan 4y ago
I’ll bite - because 80% of my searches it returns a convenient answer that matches my expectations of what the answer will look like. I still have to validate (there’s still an objective reality to be satisfied) but that’s pretty good. 20% of the time I have to google, but that’s made more efficient by using specific terms mentioned by chatGPT - getting the right terms in a google search is half the battle. It does seem archaic already for google to give me a list of links and I have to go through them all manually deciding if they answer the query.
- MBCook 4y ago
  Huh. Thanks.
- vba616 4y ago
  Could you give any concrete examples?
- testrun 4y ago
  How do you know what part of the answer is wrong?
  - dizhn 4y ago
    I asked chatgpt to give me the equivalent "ss" command to "netstat -nap". It did and it was correct because I already knew the answer and tried it anyway. Then I asked the equivalent of "netstat -rn" (if any), it gave me "ss -r" something, which is wrong. So you either know enough to spot the error or you verify all of it. Another example. I had a long session with it where I was trying to get it to write a gitlab ci config file where the artifacts to be deployed to Netlify were already present at the root of the repo. As an addition it had a requirement to use 'sed' to rewrite a domain name recursively. This is the first and only time chatgpt was useful to me although it made a lot of mistakes that needed correcting.It was more useful than a google search because I wasn't able to find the correct format with it as most examples assumed a build step. One thing that was frustrating is how I would give it requirements and it would incorporate them only to lose a prior requirement when fixing a latter one. I ended up hand fixing one of the later configs rather than using the final one it gave me. For example I needed to use a node image because netlify-cli was required (the only custom thing necessary from the image) but it would revert to using an Alpine image and point to a now archived repo with a nonexistent release url for gitlab's cli. When it fixed something else it would break the find .. sed combination and add nonsensical things like *. html to it rather than working on all files recursively. In the end, through a combination of AI helping and the user (me) chiming in and trying to reason about it I was able to get somewhere. Not ideal but not useless.