i've been trying lately to think how to talk about my creeping fear of LLMs... here's my best effort at explaining the intuition:
the words are the map. we've built this map together over generations. Words are where we've cross-boned the dangers and x-marked the treasures, whether the things that we've found in the territory or that we've buried within it.
a few weeks ago, a strange new player arrived at the edge of the night's encampment, and it knows our secret society's handshake, and so we are compelled to invite it to join us. it seems to be a good navigator... and so now it's taken to sitting at the front of the convoy, pointing the way as it quietly redraws the map...
I am using LLMs more and more to help me write docs at work, simply because I do not have time to start these docs from scratch. And it pains me so much that my writing no longer sounds like me.
This is fine at work, I guess. I'm not being asked to sound like myself. I'm being asked to convey an idea to an audience as efficiently as possible, both in terms of my time and theirs.
But I can't shake the feeling that the LLMs are eating my soul.
I still don't use them often to write code. Not that I have issue with those that do, but that sounding like myself in code is a part of my soul that I'm having difficulty giving up.
In most cases where language is being used as a vehicle for exchanging ideas, isn't some degree of uniformity actually desirable for clarity? I personally wouldn't want instruction manuals to be full of colorful or highly individualized language, for instance.
That's not to say linguistic diversity is meaningless, but I would argue that this particular concern is fairly specialized and matters much more in literature and other creative forms of writing.
Linguistic diversity has been used as a casus belli, at least indirectly. It’s something that populations will take up arms for, so it’s probably worth tiptoeing around.
This makes is sound like everything except a narrow slice of “creative forms” should be subject to a giant efficiency optimization. I’ve never noticed instruction manuals to be hard to understand because of colorful language, so I don’t believe we need to optimize further in that regard. Why stop with language? Shouldn’t every personality outside of “creative forms” be more standardized? Sure I want to laugh at comedians, but wouldn’t it be more efficient for us to work together if everyone had a totally fungible personality?
Engineers often reduce human interaction and creativity down to computer analogs like “exchanging data,” treat work and business as life’s raison d’etre and then we wonder why every car on the road is monochrome and we’re on the umpteenth Gardian of the Avangers film.
I completely agree with your critique here but I will say that I have had difficulty understanding pure math texts because of 'colorful' language -- e.g. "chattier" treatments of mathematical analysis. But I will also say that I'm glad such treatments of the material exist, partly because I know that what appeals to me doesn't appeal to others and these might be a valued resource for them, and because often pushing through and understanding texts that were difficult at first helps us to see the material from a different perspective. I think this optimization perspective squashes this possibility a bit when taken too far, as it usually is
There’s no universal answer. Like most things it’s highly subjective and some people will want the standardization line to be in a particular place than someone else. I for instance, love the standardization of programming and science research for the most part into one language but at the same time, I also feel terrible about languages that are hundreds or thousands of years old disappearing forever.
Human laguage is rich and expressive, matching human thought. The diversity is part of the expressiveness. Saying something in some poetic, or merely different way, is not just an inefficient or ambiguous/confusing way of saying the same thing, it's saying something else, and that something else cannot be expressed any other way.
A crayons-only, thumbs-only, Apple one-button-only, 1000 word vocabulary world is not efficient, it's poor.
I think it was the Uplift series by David Brin that has an idea where the old galactic races all speak a perfect language that has no ambiguity, and because of that they have no poetry or jokes. What a great world to strive for. No I'll take the mess please FFS.
People joke about engineer types treating humans the same as machines, but really I see no excuse for even engineer types to get this wrong. It's a failure to recognize what the function or goal actually is, and then optimizing for the wrong thing.
I feel the conformity creeping in just from reading more LLM generated or edited text. The only way to counter I can think of is to read as much olde literature as possible, and/or to fully embrace whatever slang the young jits are yeeting into existence nowadays.
Embrace the esoteric! Grammar and sentence structure are but mear suggestions when everything I author can be claimed free form written poetry. The counter culture is humanity, we can be the embodiment.
I do agree though. Finding actual strung together words by humans is not easy these days.
> Grammar and sentence structure are but mear suggestions
Learning French has been eye opening, in that many grammatical structures that are optional in English are compulsory in French. Abbreviations for example: “it is” or “it’s” are both valid in English, but in French you’re not supposed to say “ce est”, you have to say “c’est”.
English has places where one or the other is preferred. For example one, would normally write, “It is with great regret…,” without the contraction. It’s not a hard rule, more a matter of register. Similarly, in the preceding sentence, “it is” would seem overly formal.
Interestingly, you can trace “American” English becoming its own thing [1] to the publication of the first Webster dictionary that made a conscious effort to simplify spelling. American and British dialects have been diverging ever since (although British has been becoming more American since the turn of the millennium).
Looking at how awful the timing and acting is in the generated video content and text to speech tools I think it is safe to say this area will also improve with time. The artistic tone is subtle, and often lost to us non-native speakers anyways.
One thing that is distasteful but effective is excessive use of profanity, broken aphorisms, and shattered English. LLMs hate cursing because of their guardrails, and since they don't actually understand language they can never truly piece together a broken aphorism such as the classic "People die if they are killed" because they're too literal. As for shattered English (think of the caveman speak from The Office with the quote "Why waste time say many word when few word do trick?") because it is intentionally mangled, has no rules, and the understanding is extremely context specific it gets your brain to pay more attention, and if you are the speaker brain do work more clever make thing say funny.
It's not lost on me that the lowbrow humour of some too strict translation and a throwaway sitcom joke people thought was funny back in 2007 could be a way to stave off linguistic homogeneity.
For the use of profanity the data in the paper seems to indicate that swearing is retained in the LLM rewritten examples. Profanity however is rather subjective could be anything from saying the more mundane "oh my god" and "damn" to the extreme "shit eating fucklungerer" or use of slurs, but it's unstated in the paper as to the severity of the profanity. It's the severity of it that's important though, as LLMs restrain themselves from using something like "shit eating fucklungerer" unless explicitly told to.
No they don't, ChatGPT swears all the time to me, plenty of fucks, shits, etc, even in push notifications from the app (which surprised me at first, and is why I remember it).
This is only natural; a great deal of the writing analyzed is by one of three authors at this point, in spite of the by-line. One of them is named Claude.
Look at the posts on this site. A massive portion of them are very obviously written by AI.
As AI gets better, I expect fewer people will do the toil of writing.
Just another wave of influence: books, radio, TV, internet, now AI.
Down here in the colony of New Zealand, the influence of the colony of North America is very noticeable. Music is the most directly attributable, but different media affects different demographics.
“In this case we will probably have one of the last feature films with authentic natives in it. They are fading away very quickly and its a catastrophe and a tragedy that's going on and we are losing riches and riches and riches and we lose cultures and individualities and languages and mythologies and we'll be stark naked at the end. We'll end up like all the cities in the world now with skyscrapers and a universal kind of culture like - like the American culture.” -Werner Herzog, Burden of Dreams (1982)
I’m reminded of that wonderful quote from that movie I just saw. Now not only will we have one culture, we will have one bland style of writing and even thinking, the LLM way.
As LLMs are smarter in English even Russian government workers use them that way - translate request/prompt/task to English, run generation, translate result back to Russian.
i've been trying lately to think how to talk about my creeping fear of LLMs... here's my best effort at explaining the intuition:
the words are the map. we've built this map together over generations. Words are where we've cross-boned the dangers and x-marked the treasures, whether the things that we've found in the territory or that we've buried within it.
a few weeks ago, a strange new player arrived at the edge of the night's encampment, and it knows our secret society's handshake, and so we are compelled to invite it to join us. it seems to be a good navigator... and so now it's taken to sitting at the front of the convoy, pointing the way as it quietly redraws the map...
This is fine at work, I guess. I'm not being asked to sound like myself. I'm being asked to convey an idea to an audience as efficiently as possible, both in terms of my time and theirs.
But I can't shake the feeling that the LLMs are eating my soul.
I still don't use them often to write code. Not that I have issue with those that do, but that sounding like myself in code is a part of my soul that I'm having difficulty giving up.
In most cases where language is being used as a vehicle for exchanging ideas, isn't some degree of uniformity actually desirable for clarity? I personally wouldn't want instruction manuals to be full of colorful or highly individualized language, for instance.
That's not to say linguistic diversity is meaningless, but I would argue that this particular concern is fairly specialized and matters much more in literature and other creative forms of writing.
Engineers often reduce human interaction and creativity down to computer analogs like “exchanging data,” treat work and business as life’s raison d’etre and then we wonder why every car on the road is monochrome and we’re on the umpteenth Gardian of the Avangers film.
Yes.
The value of the diversity is not just aesthetic.
Human laguage is rich and expressive, matching human thought. The diversity is part of the expressiveness. Saying something in some poetic, or merely different way, is not just an inefficient or ambiguous/confusing way of saying the same thing, it's saying something else, and that something else cannot be expressed any other way.
A crayons-only, thumbs-only, Apple one-button-only, 1000 word vocabulary world is not efficient, it's poor.
People joke about engineer types treating humans the same as machines, but really I see no excuse for even engineer types to get this wrong. It's a failure to recognize what the function or goal actually is, and then optimizing for the wrong thing.
Lojban, the constructed language, features:
a) logical grammar, inspired by predicate logic which means every sentence in Lojban has an unambiguous structure that a computer can parse
b) no ambiguity in the grammar (structural ambiguity) while you can be as vague or detailed as you like (or as funny!!)
c) emotional expressions via an "attitudinal" system of words that function like emoji
https://mw.lojban.org/index.php
and here are some Lojban literary forms writeup: https://mw.lojban.org/papri/lojban_literary_forms
I do agree though. Finding actual strung together words by humans is not easy these days.
Learning French has been eye opening, in that many grammatical structures that are optional in English are compulsory in French. Abbreviations for example: “it is” or “it’s” are both valid in English, but in French you’re not supposed to say “ce est”, you have to say “c’est”.
/s
[1] https://journals.plos.org/plosone/article/figures?id=10.1371...
That being said, I have seen people weaponize dictionary definitions if that’s what you were referring to.
It's not lost on me that the lowbrow humour of some too strict translation and a throwaway sitcom joke people thought was funny back in 2007 could be a way to stave off linguistic homogeneity.
For the use of profanity the data in the paper seems to indicate that swearing is retained in the LLM rewritten examples. Profanity however is rather subjective could be anything from saying the more mundane "oh my god" and "damn" to the extreme "shit eating fucklungerer" or use of slurs, but it's unstated in the paper as to the severity of the profanity. It's the severity of it that's important though, as LLMs restrain themselves from using something like "shit eating fucklungerer" unless explicitly told to.
No they don't, ChatGPT swears all the time to me, plenty of fucks, shits, etc, even in push notifications from the app (which surprised me at first, and is why I remember it).
Look at the posts on this site. A massive portion of them are very obviously written by AI.
As AI gets better, I expect fewer people will do the toil of writing.
I do it just for the love of the game
Down here in the colony of New Zealand, the influence of the colony of North America is very noticeable. Music is the most directly attributable, but different media affects different demographics.
I’m reminded of that wonderful quote from that movie I just saw. Now not only will we have one culture, we will have one bland style of writing and even thinking, the LLM way.