Recently I have been seeing examples of people elsewhere oursourcing posts, apologies, and even entire books to artificial intelligence. Needless to say, I am appalled that there are adults doing this. No, writing is not always easy, but resorting to quick-fix solutions like these only implies carelessness as well as incompetence, neither whereof you deserve.

Because of this, I want to interrupt my schedule to spell this out unambiguously: I will never explain history to you through A.I. At most, I may occasionally consult A.I. if I want to get started on researching a subject, and I’ll openly declare that I am relying on a machine translation in the rare instances where I have little choice, but I’ll never use A.I. to write my own topics and I’ll be avoiding A.I.-written works as much as possible. That is my promise to you.

If my own writing seems a little odd — with em dashes and whatnot — that is just an old tendency of mine to formalize my writing (because you deserve only the best). I used to generate those characters with Microsoft Keyboard Layout Creator, and now I use the Compose key on Bazzite, and that is how I can easily type these fancy characters. I’ll understand if my tendency now looks suspicious, but I can assure you that it is nothing more than a consequence of my perfectionism, which in turn is a consequence of my obsessive–compulsive disorder. Likewise, I have a tendency to sometimes show off fancy vocabulary and orthography that I recently learned, but you can understand that as either my creativity or my pedantry, depending on your point of view.

You may be wondering how I can tell if somebody used A.I. to generate a book. Doing research on the Kriegsmarine, I came across a certain work (which I can no longer find) that had plausible yet unsourced claims. That was strike one. Looking at the author, I noticed that he generated a lot of books entirely unrelated to history and that he apparently had no background in the field. That was strike two. The smoking gun was that he admitted at either the beginning or the end that he had a little help from artificial intelligence when he composed his work. That was strike three. In short, I scrutinize: verifiable academics with years of experience are highly unlikely to suddenly rely on artificial intelligence.

I care about the quality of my contributions, and if you have evidence to suggest that I unwittingly gave you an A.I.-generated source, let me know and I’ll remove it as quickly as possible. I’ll even delete the topic if it happened to be my only source. Exposing our enemies is too important a task to be trusted in the hands of automata.

  • davel
    link
    fedilink
    English
    arrow-up
    13
    ·
    3 days ago

    LLMs didn’t get their em dash habit from nowhere—they got it aping us.

    • La Dame d'Azur
      link
      fedilink
      English
      arrow-up
      9
      ·
      3 days ago

      I’m preemptively angry because I’ve been using em dashes for ages before chatbots were a big thing and have no plan to stop even though I know some prick is going to see them and accuse me of using AI.

      Whoever told people AI is the only one to use emdashes - I wish you 10,000 years of misery.

      • davel
        link
        fedilink
        English
        arrow-up
        5
        ·
        3 days ago

        I get it: me too, and I’m not about to change because of somebotty.

    • CriticalResist8A
      link
      fedilink
      English
      arrow-up
      7
      ·
      3 days ago

      I read a blog post the other day that felt completely lifted off a first chatGPT pass, but it was from 2018. 4 years before we even heard the word LLM.

      As for AI writing you don’t have to offload everything to AI. there’s tiers and methods to it. Perplexity is a very good search engine to find niche stuff though you still need to go through the sources and confirm it says what the LLM thinks it says. But that’s one way to get sources, I’ve used it extensively. That’s actually second tier, first tier is just writing your own way but redoing some parts with an LLM because you feel the sentence/paragraph is too complicated.

      Tier 3 is the big brain move, ask the LLM to write in the style of a forum or reddit comment. If you can get it away from its guardrails it can produce pretty interesting text at times. Models on APIs usually have fewer restrictions placed on them and you get a feel for what these things are actually capable of generating. Not all the time, but sometimes it aligns just right and you wonder what the guardrails are hiding from us lol.

  • NotMushroomForDebate
    link
    fedilink
    English
    arrow-up
    7
    ·
    3 days ago

    I think I’m a little bit out of the loop here. Who is “we” referring to in the title? This community specifically, Lemmygrad in general, or a joint mod/admin statement?

    I’m also aware that CriticalResist has a blog, but I’m not familiar with where you publish your writing. Is it mainly here on this community or do you also have a blog or edit articles on ProleWiki, etc.?

    • Anarcho-BolshevikOPM
      link
      fedilink
      English
      arrow-up
      7
      ·
      edit-2
      3 days ago

      Who is “we” referring to in the title?

      This subcommunity specifically.

      I’m not familiar with where you publish your writing. Is it mainly here on this community

      Yes.

      or do you also have a blog

      No.

      edit articles on ProleWiki

      Occasionally.

    • Nondiegetic (any)
      link
      fedilink
      English
      arrow-up
      6
      ·
      3 days ago

      Anarcho does a bunch of really in-depth writing, mostly in this comm, usually very well sourced essays about history.