It seems to me that those who can't see the fundamentally destructive nature of LLMs never fully grasped the miracle that is language in the first place. It is the delicate thread in which we weave meaning, and therefore human connection and purpose. Appropriating and poisoning it for...well whatever they think they're after, is a bonfire on a ship made of its own timbers.
18
Replies
4
Boosts
Concrete example: “Claudish”.
That’s what some developers who use Claude Code and related tools call the weird text blobs it spits out when “explaining” what it does.
In my opinion, it’s even worse than the well-known ChatGPT verbal tics like “It’s not X; it’s Y” and overusing Rule of Three. Imagine taking the worst trends in both software developer jargon and corporate marketing language, inducing some model collapse so the vocabulary and grammar feel artificially restrictive and reductive, and finally wrapping it all up in a punchy, terse style that is neither clear nor pleasant to read.
Claudish has been around for a while (I’ve seen it creep into various documents at work over the last 6 months or so), but has finally reached a point where even hardcore Claude fans are noticing and getting irritated. https://slhck.info/software/2026/06/22/claudish.
@mttaggart not just language but writing and editing specifically. there is a comprehensive misunderstanding of what the process of writing *is*. “this is uncomfortable and I am not quickly getting to the result (e.g.: a word count); I must be doing this wrong”. no! the discomfort is the point. and not like in a “suffering is valorous” way but in a “resolving cognitive dissonance is painful” way. you’re ordering your own disordered thoughts! that is hard and it feels bad
@glyph @mttaggart It can feel bad and it’s hard and sometimes what you end up with doesn’t effectively communicate our thoughts to others whose language, experience and cultural context is just enough off from our own that our intent is not accurately (enough) reflected into their mind. It’s quite amazing what machine learning can do with human language, but it also is missing so much of what is going on because we often don’t know what is missing with our words until they fail.
@mttaggart @glyph Yeah. Our ability to communicate is so much tied into our ability to imagine a mind that works like ours creating the words. When I’m reading LLM generated stuff and I realize that wait this is nonsense because this wasn’t written (or even revised) by a person it’s so painful. Because I spent so much energy assuming it is.
@mttaggart @r343l even the imperfections can be extremely revealing! but with an LLM there is no revealing because there is nothing *to* reveal
@mttaggart “I must be moving these weights wrong, every time I go to the gym my arms hurt”
not a perfect metaphor since not all of what hurts here is just self-improvement, the ordering of the ideas also improves the ideas themselves, but, still, same vibes
@glyph Yeah for me "language" is broadly "the ordering of human thought into speech or writing" in all forms.
@mttaggart it is hard to talk about because the origin point for all their usage is still “language”. they write so many prompts. so many agents dot em dee. but it is … debased, somehow. the kind of thinking that goes into a prompt is distorted and bizarre. flaccid, careless, the perverse approximation of language that you get out of a chatbot is a strangely fitting reward for the perverse approximation of language you put in
@glyph With whom does the user converse in a prompt? What do they receive back? How can it be anything nourishing? And over time, what happens to the mind that speaks mostly to the thoughtless machine?
@mttaggart @astraluma Part (the smallest part, TBH) of the damage that LLMs do to society is devalue the word. Words matter. And just because 2 things are semantically equivalent does not mean that they are equal - but LLMs act as they are.
@brass75 @astraluma I actually don't think it is the smallest part. By severing constructed language that appears human from human intent, and flooding our world with that intentless language, erodes the fabric of meaning that binds us
@mttaggart I'm going to slightly butcher this because it's late: IIRC Socrates (or Plato?) was against writing because it dulled the mind or something. It kinda looks like we could arrive at what they may well have wanted (less writing) with the full destruction of the value of writing because nobody will be able to trust written text.
@mttaggart Reading this is an absolute trip in the modern context
@tehfishman I used to use it as a counter to parents who were terrified of the internet. Joke's on me.
@mttaggart This is the closest I've heard someone say that if nerds would learn some humanities nuance then things would be better as far as LLM adoption is concerned. I don't know if that's the intent but it's where my brain took it. Possibly because I would agree with it.
@cR0w I actually don't think any amount of semiotics or epistemology would prevent or undo the craven nature of these people. But yes, perhaps the metaphysical danger would be better understood.
