Login
You're viewing the social.afront.org public feed.
  • Apr 23, 2026, 12:29 AM

    Style guide question: how to consistently refer to the umbrella of generative AI, including LLMs and diffusion image/audio generators?

    💬 16🔄 10⭐ 2

Replies

  • 💬 3🔄 1⭐ 1
  • Apr 23, 2026, 1:10 AM

    I know y'all are nerds (non-derogatory) but please do keep in mind that one of my goals with converging on a term to use consistently is making my writing easily comprehensible by a large audience. I am talking about the crud that we all have to talk about all the time now, and I want to be as congruent as possible with common usage without unnecessarily contributing to hype

    💬 3🔄 0⭐ 0
  • Apr 23, 2026, 12:31 AM

    @glyph I voted "generative AI" but I make a concerted effort to avoid using "AI" or at least wrap it in scare quotes. I would likely substitute "models"

    But I am also fully cognizant of what a grouch I am and the enumerated choice I selected is an understandable compromise for "being understood by normal people"

    💬 1🔄 0⭐ 0
  • Apr 23, 2026, 12:33 AM

    @SnoopJ being understood is a big part of the problem but also just … not having to type a fifty-syllable phrase every goddamn time

    💬 1🔄 0⭐ 0
  • Apr 23, 2026, 12:35 AM

    @glyph yea, I definitely write "AI" way more often, myself so I guess I'm not even being particularly consistent with that vote?

    But I think I also read it with the presumptive context of what you've been working on and assuming that this would be too terse for… some reason?

    💬 1🔄 0⭐ 0
  • Apr 23, 2026, 12:42 AM

    @SnoopJ @glyph I personally use LLMs as my more explicit term, even though that *technically* excludes some of the comparably problematic generative AI techniques.

    💬 1🔄 0⭐ 0
  • Apr 23, 2026, 3:58 AM

    @ancoghlan @SnoopJ @glyph I tend to think that what you choose matters less than it seems particularly because it is such a garbage term at this point.

    What I feel is more useful to me is when a writer, either pro or against, makes clear at the onset what term they are using and what encompasses it.

    "For the sake of brevity, when I use the term 'AI' from here on out, I am including xxxxx, and excluding the xxxxx aspects."

    💬 0🔄 0⭐ 0
  • Apr 23, 2026, 12:40 AM

    @glyph I'm also in the "stop calling it AI" camp. If I may offer a possible alternative that is equally concise of "generative models"

    💬 0🔄 0⭐ 0
  • Apr 23, 2026, 12:43 AM

    @glyph Voted genAI, but depending on context I might prefer plain AI with a footnote or parenthetical on the first instance

    💬 0🔄 0⭐ 0
  • Apr 23, 2026, 1:18 AM

    @glyph for what it's worth, my talk on Saturday uses "agents" almost exclusively, except where referring to LLMs as a technology are the truly relevant factor. I've done that to refer solely to the products that people engage with, regardless of what technology underlies it.

    (This is relevant because ChatGPT, for instance, is an LLM-powered chatbot thatcan call out to both diffusion models and symbolic reasoning models where necessary. The shape of the product is more important than the first line of tech you encounter.)

    💬 0🔄 0⭐ 0
  • Apr 23, 2026, 1:57 AM

    @glyph in informal contexts I tend to write [“AI” (🙄)] without the square brackets as it captures both the scare quotes and my general attitude to the use of the term in the current hype cycle – BUT scare quotes don’t translate to spoken contexts like talks & instead it will just sound like we’re agreeing that it’s accurate &/or reasonable to call these things “artificial intelligence” (because, of course, they’re only one of those words & it isn’t the one starting with “I” 😩)

    so in more formal written or spoken contexts I’ll introduce the concept / terms as something like “large language models^ – usually abbreviated to ‘LLMs’ – and similar tools that generate outputs based on training data” on first use and then condense to something like “LLMs and similar tools” or just “these tools” for subsequent uses 💁‍♀️

    I use “tools” (or sometimes “technologies”) because I think it’s important to remind / make clear to people that they are tools, not magic – and certainly not minds or anything resembling the processes that go on in human brains (which we really don’t understand as well as people may assume we do, but we do understand enough to be confident that “this ain’t it”)
     
     
    but in a less formal talk that I’ve given a couple of times to techie / nerdy audiences I’ve described them thusly:

    “To bring us all up to speed on how these* Large Language Models (or LLMs); these image-
    to-text generators
    ; these stochastic parrots; these plausible word-salad generators;
    these rapacious pirates of human-made creativity – sorry, these rapacious
    “copyright infringers of human-made creativity”
    actually work…” 🙃
     
     
     
    ^ or, if I’m talking about different adjacent / related things like [machine | deep] learning, neural nets, GANs, etc., I’ll use the appropriate specific term for that tool or technique

    💬 0🔄 0⭐ 0
  • Apr 23, 2026, 4:59 AM

    @glyph I am still somewhat stuck in the habit of referring to these as "LLMs", but in this context I think "genAI" is the winner. Not because the "AI" part is descriptive (I think we may have lost that battle) but because it puts emphasis on the "generative" aspect.

    I think that framing helps everyone get onboard with the idea that this is just one way that technology can work. We don't have to make a "content generator" machine, regardless of how it runs, but that's what we have done.

    💬 0🔄 0⭐ 0
  • Apr 23, 2026, 4:57 AM

    @glyph at work i started calling it the slop-machine.
    Makes for interesting conversations if ppl did not know the word yet

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Apr 23, 2026, 1:16 AM

    @glyph Um, this needs a neologism.

    English words do not end in two capital letters -- this is my way of saying "genAI" looks like shit and is bad -- so "genai" is the ideal form it will converge on.

    💬 1🔄 0⭐ 0
  • Apr 23, 2026, 1:18 AM

    @prietschka I'm already doing a lot of heavy lifting here and do not want to additionally take on the project of popularizing a neologism.

    💬 1🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Apr 23, 2026, 1:25 AM

    @glyph “generative model”, or just “model” or “generator” if you need brevity.

    Steer clear of *any* implication of intelligence, IMO.

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Apr 23, 2026, 1:44 AM

    @glyph imho, AI then use historical aspects of the term to dilute it further. Expert systems have been AI; decision trees have been AI;...

    💬 0🔄 0⭐ 0
  • Apr 23, 2026, 1:55 AM

    @glyph no-brainer to use “AI” if you are trying to reach a large audience.

    Yes, “AI” can mean many things. So does “oil” but you don’t see any writer on public policy taking care to exclude salad dressing

    💬 0🔄 0⭐ 0
  • Apr 23, 2026, 2:56 AM

    @glyph applied ML / machine learning? since ai is a marketing term and meaningless?

    💬 0🔄 0⭐ 0
  • Apr 23, 2026, 4:30 AM

    @glyph I use "AI" (with quotes, as in "so-called") for a slightly broader set of technologies that all feed into AI hype.

    I try to avoid "generative AI" because the group of algorithms actually doesn't have that much in common in terms of use, cultural/political effect, etc.

    I find it more meaningful to use more specific terms.

    Classifying by implementation details feels odd. Image generatoon algorithms are a more meaningful class, of which machine learning-based image gen is a subset, e.g.

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Apr 23, 2026, 5:25 AM

    @glyph I think it's ok to use AI without quotes, even though it's inaccurate, because the techbros are using that term in their saturation advertising trying to sell their slop machines. If they don't like it when we use it disparagingly, that's too bad. They did it to themselves.

    💬 0🔄 0⭐ 0
  • Apr 23, 2026, 8:15 AM

    @glyph depending on the point i’m trying to make: language models, probabilistic tools, or "AI" with quotes. specifically never AI without quotes in an effort to push back against the lie that a science-fiction idea is real

    💬 0🔄 0⭐ 0