• Re: Capabilities of LLMs

    From Johann 'Myrkraverk' Oskarsson@johann@myrkraverk.invalid to alt.usage.english,comp.ai.philosophy,comp.lang.misc on Thu Oct 1 22:10:34 2026
    From Newsgroup: comp.lang.misc

    On 10/1/2026 5:27 PM, Stefan Ram wrote:
    Johann 'Myrkraverk' Oskarsson <johann@myrkraverk.invalid> wrote or quoted:
    Anyway, a human insufferable know-it-all could have corrected my ques-
    tion, and followed up with an /achtually/ stating that we are merely
    talking about four years, from fourteen hundred and ninety seven to fif-
    teen hundred and one [1].

    OTOH, more than once I got answers where the model apparently
    tried to guess my intend, and I had to write in the next prompt,

    |Please answer my question literally.

    Dear Stefan,

    Usually, when that happens to me, it seems to want to make up the quest-
    ion. As in, answer something completely different. Sometimes, when I
    ask a search engine for a definition of a word, the /ai assist/ will de-
    fine a completely different word.

    So I have trained myself never to read the /ai assist/ summary when ask-
    ing a dictionary, and always click merriam-webster, oxford, cambridge,
    or some other well known human-edited dictionary.

    If you think /define a completely different word/ is a valid answer,
    I'll point out the we are humans, and sometimes only read the definition
    and don't always re-read what we wrote; meaning, /what word is the ai
    assistant defining/, we just wrote it. Putting it in a different font
    or enlarge the word-under-definition just makes us skip that part hard-
    er.

    And people who don't understand this, that the ai can simply answer
    something completely different, or use outdated references [1] such as
    an article on the wrong version of Drupal [3] will forever wade in error
    and confusion through life [5]. I mean, that's just as bad as getting a
    quote from Python 2.7 when you're trying to do something extra-Pythonic
    in 3.17 [6].

    So we live in a world, where the media controlled gaslighting is always
    telling us the ai is great, that the ai knows best, and that we should
    all be using ai for everything, when in fact the ai is completely stupid
    and cannot be made to understand it should be able to connect the dots
    and know -- at least the basics -- that the Kalmar Union existed, and
    that it included both Iceland and Sweden at the same time -- here is a Wikipedia article with a map,

    https://en.wikipedia.org/wiki/Kalmar_Union

    which I have not personally verified -- even though I'm not asking it
    about the Kalmar Union by name. You know, completely differently from
    an insufferable human know-it-all.

    Then the insufferable know-it-all humans of Usenet are trying to tell us
    the ai is great? I mean, how insufferable is that? I don't know what
    the people of comp.ai.philosophy are thinking, because this is not OK.
    It is not OK to gaslight regular people who don't know better that they
    should be paying ai tokens to Sam Altman, rather than just read physical
    books themselves, and become the insufferable know-it-all.


    Best wishes, and happy suffering from insufferable know-it-alls of Use-
    net!

    [1] Consider the following Wikipedia article,

    https://en.wikipedia.org/wiki/Bureaucracy

    which has the following quotation from an outdated but publicly availa-
    ble article titled /Scribal Training in Ancient Egypt/,

    Similarly, Ancient Egypt had a hereditary class of scribes that
    administered a civil-service bureaucracy.[18]

    and you can read it here,

    https://www.jstor.org/stable/600648?origin=crossref

    which I haven't, to be honest. On the other hand, this guy,

    https://en.wikipedia.org/wiki/Bob_Brier

    and

    https://drbobbrier.com/about-bob-brier

    taught me in a video course [I won't name the exact episode] that
    any-body with intelligence could become a scribe, and confirmed it in a pri- vate correspondence [2]. I have however decided not to get into an edit
    war with the Wikipedia editors, and have left the article incorrect. It
    would be more correct to leave out the sentence entirely.

    [2] Paraphrasing the librarian from The Mummy (1999), /take that you
    Wikipedia editors/.

    [3] As far as I know, there's no comp.lang.drupal, but I've included comp.lang.misc for a few programmers so they can read this monologue as
    well, and be illuminated about the fact they can't trust the A.I. tuto-
    rials, because it can be teaching the wrong version of the programming
    language [4].

    [4] Which also begs the question, should we try to start comp.lang.php,
    or leave that up to the Bangko Sentral ng Pilipinas?

    [5] Here I am paraphrasing an Icelandic expression, and people from alt. usage.english can suggest a better one.

    [6] I have however decided not to include comp.lang.python in this
    discussion, since all programmers already read comp.lang.misc for Rust
    because comp.lang.rust is still under consideration by news.groups.pro-
    posals.
    --
    Johann | email: invalid -> com | http://www.myrkraverk.com/blog/
    I'm not from the Internet, I just work there. | via Easynews.com https://bsky.app/profile/myrkraverk.bsky.social | for ( ;; ) _:;
    Federated at https://fed.brid.gy/bsky/myrkraverk.bsky.social

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Bob@usenet@bananacorp.nl.invalid to alt.usage.english,comp.ai.philosophy,comp.lang.misc on Sun Oct 25 14:15:16 2026
    From Newsgroup: comp.lang.misc

    Johann 'Myrkraverk' Oskarsson <johann@myrkraverk.invalid> writes:

    On 10/1/2026 5:27 PM, Stefan Ram wrote:
    Johann 'Myrkraverk' Oskarsson <johann@myrkraverk.invalid> wrote or quoted: >>> Anyway, a human insufferable know-it-all could have corrected my ques-
    tion, and followed up with an /achtually/ stating that we are merely
    talking about four years, from fourteen hundred and ninety seven to fif- >>> teen hundred and one [1].

    OTOH, more than once I got answers where the model apparently
    tried to guess my intend, and I had to write in the next prompt,

    |Please answer my question literally.

    Dear Stefan,

    Usually, when that happens to me, it seems to want to make up the quest-
    ion. As in, answer something completely different. Sometimes, when I
    ask a search engine for a definition of a word, the /ai assist/ will de-
    fine a completely different word.

    So I have trained myself never to read the /ai assist/ summary when ask-
    ing a dictionary, and always click merriam-webster, oxford, cambridge,
    or some other well known human-edited dictionary.

    May I add that using "dict", or some convenient graphical interface for
    that, will allow for (ab)using less resources. Rejoice. Check.

    Cheers,

    Bob
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Johann 'Myrkraverk' Oskarsson@johann@myrkraverk.invalid to alt.usage.english,comp.ai.philosophy,comp.lang.misc on Sun Oct 25 23:34:17 2026
    From Newsgroup: comp.lang.misc

    On 10/25/2026 10:15 PM, Bob wrote:
    Johann 'Myrkraverk' Oskarsson <johann@myrkraverk.invalid> writes:

    On 10/1/2026 5:27 PM, Stefan Ram wrote:
    Johann 'Myrkraverk' Oskarsson <johann@myrkraverk.invalid> wrote or quoted: >>>> Anyway, a human insufferable know-it-all could have corrected my ques- >>>> tion, and followed up with an /achtually/ stating that we are merely
    talking about four years, from fourteen hundred and ninety seven to fif- >>>> teen hundred and one [1].

    OTOH, more than once I got answers where the model apparently
    tried to guess my intend, and I had to write in the next prompt,

    |Please answer my question literally.

    Dear Stefan,

    Usually, when that happens to me, it seems to want to make up the quest-
    ion. As in, answer something completely different. Sometimes, when I
    ask a search engine for a definition of a word, the /ai assist/ will de-
    fine a completely different word.

    So I have trained myself never to read the /ai assist/ summary when ask-
    ing a dictionary, and always click merriam-webster, oxford, cambridge,
    or some other well known human-edited dictionary.

    May I add that using "dict", or some convenient graphical interface for
    that, will allow for (ab)using less resources. Rejoice. Check.

    Fair, but since I do have some life challenges that is not a priority.

    I have a still-sealed copy of Webster's Encyclopedia on a CD-ROM for
    Windows 3.1 and 95. I have not yet opened it to find out whether it
    works on Windows 11, but that's an adventure that awaits.

    Or I try it with a virtual machine, and that's a decision I make when
    the time comes for testing out CD-ROM encyclopedias. Though, currently
    I have only one; I might have another by the time I get to this quest.


    Best wishes, and happy encyclopedias!
    --
    Johann | email: invalid -> com | http://www.myrkraverk.com/blog/
    I'm not from the Internet, I just work there. | via Easynews.com https://bsky.app/profile/myrkraverk.bsky.social | for ( ;; ) _:;
    Federated at https://fed.brid.gy/bsky/myrkraverk.bsky.social
    --- Synchronet 3.22a-Linux NewsLink 1.2