Comment by PaulKeeble
17 hours ago
I just can't seem to stay interested and read LLM text. It just seems to lack going somewhere, its information density is often really low and its full of odd isms that I don't find very interesting to read. I don't have a good radar for detecting the various LLM particular speech patterns but everytime I have an issue with an article there is always people talking about how its clearly AI written.
I find the same thing with written results out of AI "research", it is a slog to read through it and most of it can be skipped even if it can sometimes be useful its an unenjoyable read.
One of my favourite rules: perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away. This is invaluable to writing. So much text from an LLM feels superfluous, like I could've said the same wall of text with just a few sentences. Writing is tightly coupled with thinking: refining thoughts, arguments, ideas, discarding them, mulling over them. LLM text always raises this nagging question in my mind: am I reading what you think, or what your agent parroted back and you accepted?
There is no way that solely talking to an agent induces the same depth or amount of thinking than when you are actually writing for yourself. It just takes too much discipline to not offload your thinking.
And the discipline gets attacked even more when all you read is LLM text. For example in code reviews there is no noticeable value in answering LLM generated review comments. I remember taking great care answering human-written reviews comments with much thought to either further build on their suggestions or to forward knowledge about the system when I had to defend a particular change.
Now, some bot will read my carefully written response and forgets about it. There is no value anymore.
No wonder that our pull request comments feel like the dead internet theory right now.
I am absolutely pro AI assisted coding. But communication between humans HAS to stay human. I am currently experiencing the opposite and it’s bad.
>It just takes too much discipline
While I think your argument has merit, I would be careful to discount any human skill based on people being too lazy to do it.
1 reply →
The second one, for sure. Heavy AI users who claim it's only for expressing their ideas more quickly/effectively are lying to themselves. They are using AI to think.
"I would have written a shorter letter, but I did not have the time."
s/time/effort
[dead]
I can imagine that one of the aspects of being social creatures is that we are drawn to the entropy/surprise factor of natural language produced by a brain that isn't our own. Deriving a shared understanding without being able to precisely model the language output of someone else is like, the fundamental substance of conversation.
Maybe that's the key ingredient that's missing from LLM-generated text. Even when there is actual content there, there's this lack of surprise / uncertainty in its delivery that we require to stay engaged. We may not know what will be communicated, but knowing exactly how it will be communicated just kills our innate interest.
This presumes an LLM has something to say. The reason the text doesnt make much sense, is because it doesnt. It isnt "on topic" because it has no topic to be on. Sentences go on, are joined irrelevantly, and so on because theyre sampled from a distribution of "possibly relevant sentences". They aren't being addressed to a reader (whose mind is being modelled), on a topic the writer wishes to express.
Communication expresses an understanding. LLMs arent communicating.
LLMs generate code well enough. Code has no natural language semantics. Programming's condition of satisfaction is that the machine executes correctly. The condition of satisfaction of communication is that the communication expresses what there is to be communicated.
In the limit where an LLM is trained to repoduce an exmplar of communication, it seems as if it does so. In that limit, LLMs can reporduce anything.
If frontier labs wanted to improve "LLMs for communication" they would pay for reasoning traces to be produced which reproduced the thought process of writers/speakers thinking about their audiences. Rarely do such traces exist in prior datasets .
What I find they do badly in code is building abstraction and breaking the problem down in a way a human does. They don't seem to really build simplifying components unless you expressly tell them too, they are more than happy to just nest 10 deep or embed a 20 line called back function and use lots of shortened names. I find I can improve the code just by organising better and choosing some abstractions that make sense to the humans and then ironically you save tokens too because the AI will reuse it, but it wont create most of it itself. Its bad for domain work unless you are explicit about it.
> This presumes an LLM has something to say. The reason the text doesnt make much sense, is because it doesnt. It isnt "on topic" because it has no topic to be on. Sentences go on, are joined irrelevantly, and so on because theyre sampled from a distribution of "possibly relevant sentences". They aren't being addressed to a reader (whose mind is being modelled), on a topic the writer wishes to express.
While I don't like LLM voice either, I'd strongly disagree with this. LLM writing can be very tight these days -- sentences do not "go on" and are not "joined irrelevantly". I'm not sure when you last used an LLM, but it must have been a very long time ago; this hasn't been the case since late 2024 for frontier models, in my experience.
Obviously there's no underlying consciousness behind it all, but the verisimilitude is good enough for stimulating conversations on the recent research literature in my field, which I appreciate. For me, at least, the communication level is absolutely fine.
(YMMV, of course, and perhaps my bar for this is lower than yours.)
1 reply →
"I just can't seem to stay interested and read LLM text."
For foreign text, sure. But I never was never able to read New York Times articles about some invention that started with Lucy leaving her home at seven o'clock.
If you have a specific question, reading LLM text can be more interesting than anything in the world. Even Wikipedia.
>If you have a specific question, reading LLM text can be more interesting than anything in the world.
I've noticed this as well. When I get a bot to generate something I am happy to read it (although I often skim read it). But god forbid I have to read someone's LLM generated report! Come to think of it, if someone marks the text as LLM generated, especially after they have plonked a hand typed summary or sentence at the top, I am quite happy to read it. Maybe my beef is being expected to read LLM text the same way I would read high effort human prose? I certainly read LLM text differently to human text - you know the first sentence is going to glaze you and the second paragraph will just repeat the first paragraph so you can skim those annoyances.
> I am happy to read it (although I often skim read it)
I came to think of it, LLM output is not worth to be read like human written text. It's best to just skim it and move on. The problem is when this LLM output finds its ways everywhere on the internet. Being able to skim text requires knowing the context beforehand and knowing that you're dealing with LLM generated output. If LLM output that is published was labeled as such it'd probably make us less frustrated.
I know what you mean. The latest models are _better_, but there's still something off about the prose they generate. Obvious tells are mostly gone now, but some internal spider-sense still makes me narrow my eyes.
I also started to just abandon writeups obviously written by LLMs. They feel (and usually are) just cheap marketing/selling posts. Not to mention why would I read stuff nobody wrote and likely never read themselves? There's little value in such content. I don't even point LLMs to such websites to get a TLDR. Part of why reading is interesting is the author, their history, beliefs and style. In the world of LLMs content, the author is some undefined diffuse alien with a made up personality.
Sorry to single you out, but it’s surprising that this is such a consensus opinion around here. If an LLM isn’t giving me exactly what I need in the format I want, I tweak skills or system prompts until it does.
These aren’t deep AI hustle bro hacks, it’s literally just editing some text. If a big part of your day is reading AI-generated words, slap something together to make it do words gooder.
>its information density is often really low
I find that it often isn't just too low, but also too high at the same time. It has a lot of information that is irrelevant, burying far too little relevant information for the work needed to extract it. I wonder if this is a side effect of attention mechanism, where the AI's ability to track information in a document is far less hampered by other information, leading to it organizing information to be both dense per word but any specific concept to be sparsely spread throughout the writing. I also wonder if this is why the order feels off, because the AI doesn't need for the information to be well ordered to the same extent that a human does, so the order of storing information is often overlooked when storing it into a document (and even information that needs to be ordered might have the ordering information separate from how it is ordered in the document).