7 ms·

Am I really the only one who doesn't care whether or not prose was written by AI? Assuming that a text is signed off on by a person, then surely what counts is

Am I really the only one who doesn't care whether or not prose was written by AI? Assuming that a text is signed off on by a person, then surely what counts is its substance, not who or what actually strung the words together. I'm constantly surprised by this obsessive need to falsify what is unfalsifiable. And among nerds of all people. It feels religious, like a modern form of heresy-purging.

inknight 2mo ago
people who generate entire blogposts with claude probably do not care about substance, tho. before llms, people would have to put effort into what they write and be mindful about the things that they are saying. now it just generates the whole thing, the people who "signs it" barely reviews it fully most of the time, IF they review it at all, proceed to say "eh, good enough" and publishes it.
- bluebarbet 2mo ago
  Ironic anecdote. Although I don't care about AI, I'm triggered by your own carefree attitude to orthography (no capitalization, unpronounceable words like "llms"). At least AI is easy to parse!
  - inknight 2mo ago
    [flagged]
  - unleaded 2mo ago
    Did you actually find that comment hard to read or are you just choosing to be offended by it?
    - bluebarbet 2mo ago
      The former. As did everyone else who read it, including you. Capitalization and punctuation are not decorative extras, they were invented for a reason.
erfgh 2mo ago
The problem is that AI writing uses a pretty distinctive style and when you read a whole newspaper that was written or edited by AI it is very monotonous. Also, I can't really trust it because I know that the author probably phoned in his fact-checking.
user43928 2mo ago
I would be fine with it conceptually, if the result was not as grating as Claude's ramblings about load-bearing seams and the real shape of the problem. It's just exhausting to read, and it displays a lack of effort put into communication.
Timwi 2mo ago
In addition to the sibling comment, which I agree with [edit: the one starting with “people who generate entire blogposts with claude probably do not care about substance, tho”], I want to add another point. Even when the substance is key, the delivery has a quality to it that encodes a useful signal. To illustrate what I mean, imagine two blog entries on the same subject, but one is written by an expert on the subject and one by a layman. As humans, we can generally (not perfectly) tell which is which. The expert will write in a certain quality that comes with expertise and is genuinely difficult to fake without it. LLMs disrupt this pattern. LLMs are able to write with the expert’s quality even while writing bullshit. I don’t mean quality as “measure of goodness” here but merely a set of traits or properties. Humans reading LLM text are far more likely to misjudge the author’s level of expertise. The next step is unscrupulous humans exploiting this to trick unsuspecting readers into misjudging a text, and now you have an internet where you can’t trust anything anymore, at least not at first glance. Even relatively discerning readers now have to waste time reading more of the text before being able to dismiss it as having substance of little value. So, as I see it, the upset is not simply from AI provenance of a text, but from this level of dishonesty and subterfuge, coupled with the helplessness with which we’re exposed to it.
backlava12 2mo ago
> Assuming that a text is signed off on by a person, then surely what counts is its substance Unfortunately that's a big assumption and for many their "signing off" process will amount to "skimmed it and seems ok?". When I get an LLM-generated doc or runbook, my first thought is that its very possible that I'm the first person who has ever read this. It used to be that writing something took a big time and energy investment up front by one person so that many others could comprehend the ideas with less time and energy needed. LLMs flip this equation around which is a very bad thing.
nunez 2mo ago
I care because in most (not all) cases, (a) AI UGC is one- to few-shotted with no concern for accuracy or readability, and (b) why would I read the results of someone else's prompt if I can achieve more or less the same result for free on ChatGPT? For example, I was searching for a guide for realigning the v-brakes on my wife's hybrid bike and found this: https://volatacycles.com/how-to-adjust-bike-brakes-rubbing/ https://volatacycles.com/how-to-adjust-bike-brakes-rubbing/ Now, this is for disc brakes, so not applicable in my situation. I've never ridden a bike with disc brakes, so I can't vet for the accuracy of this guide. Maybe it's totally right; I'm sure someone reading this will know. However, just look at this article! Zero pictures for a process that _really_ needs it. It just drones on. Imagine following this guide only to discover that the 12-15 Nm rotor torque the article is asking for is actually too high! A similar article from before AI would at least have (stolen) pictures describing the process. It would also very likely be shorter. Which creates the rub: if I know how to write something engaging and correcting AI's copy will take _just as much time_ as writing it myself, why would I use AI to do it? In short, AI content is lazy content. My time is valuable and finite; if I'm reading your stuff, I want to know that you at least tried to give a shit about my time as a reader while producing it.
- bluebarbet 2mo ago
  Sure, I get that argument. I just don't buy it. What counts is whether the article helps you fix your brakes. If it doesn't, it seems immaterial to me whether or not the bylined author sweated over it for hours. Not least because I have no means of knowing.
  - andrewflnr 2mo ago
    GP described multiple reasons it didn't help them fix their brakes.
    - bluebarbet 2mo ago
      So blame whoever signed off on it. AI or not-AI is their problem.
bothers 2mo ago
Why shouldn't people be upset if authors are being replaced by like a handful of the same ghostwriters? An LLM can't convey your ideas and your voice better than you can convey them to the LLM, right? It's a middleman between you and those you want to reach that adds more points of failure, an additional entity that must be understood and made to understand, or else meaning is lost. EDIT: Never mind the motivations behind a majority of LLM-generated articles, which is to be One Unit Whole Content, a colloid for ads.
no_multitudes 2mo ago
Because it sucks! It's badly written, too long, and often pointless. I don't like reading tacky, useless garbage
- kcoul 2mo ago
  Yeah, I keep feeling the same thing once I know I'm reading something or watching something purely AI-generated - that there are good points being made, but the delivery is just so off that it kind of ruins the points themselves. If people used AI to get the scaffolding of their article going and then re-humanized the entire article, it might be really good, still at a fraction of the time it would have taken to write it themselves, but no one seems to want to expend enough effort to go the extra mile. With video it's pretty well hopeless, unless the AI helps accelerate fully human traditional production techniques. With music, there might be a middle ground, I have had some success getting AI to generate good sounding loops to use in electronic music production, but it was just a matter of brute forcing enough outputs to finally get something decent, which isn't fun at all.

inknight 2mo ago

people who generate entire blogposts with claude probably do not care about substance, tho. before llms, people would have to put effort into what they write and be mindful about the things that they are saying. now it just generates the whole thing, the people who "signs it" barely reviews it fully most of the time, IF they review it at all, proceed to say "eh, good enough" and publishes it.

bluebarbet 2mo ago
Ironic anecdote. Although I don't care about AI, I'm triggered by your own carefree attitude to orthography (no capitalization, unpronounceable words like "llms"). At least AI is easy to parse!
- inknight 2mo ago
  [flagged]
- unleaded 2mo ago
  Did you actually find that comment hard to read or are you just choosing to be offended by it?
  - bluebarbet 2mo ago
    The former. As did everyone else who read it, including you. Capitalization and punctuation are not decorative extras, they were invented for a reason.

erfgh 2mo ago

The problem is that AI writing uses a pretty distinctive style and when you read a whole newspaper that was written or edited by AI it is very monotonous. Also, I can't really trust it because I know that the author probably phoned in his fact-checking.

user43928 2mo ago

I would be fine with it conceptually, if the result was not as grating as Claude's ramblings about load-bearing seams and the real shape of the problem. It's just exhausting to read, and it displays a lack of effort put into communication.

Timwi 2mo ago

In addition to the sibling comment, which I agree with [edit: the one starting with “people who generate entire blogposts with claude probably do not care about substance, tho”], I want to add another point. Even when the substance is key, the delivery has a quality to it that encodes a useful signal. To illustrate what I mean, imagine two blog entries on the same subject, but one is written by an expert on the subject and one by a layman. As humans, we can generally (not perfectly) tell which is which. The expert will write in a certain quality that comes with expertise and is genuinely difficult to fake without it. LLMs disrupt this pattern. LLMs are able to write with the expert’s quality even while writing bullshit. I don’t mean quality as “measure of goodness” here but merely a set of traits or properties. Humans reading LLM text are far more likely to misjudge the author’s level of expertise. The next step is unscrupulous humans exploiting this to trick unsuspecting readers into misjudging a text, and now you have an internet where you can’t trust anything anymore, at least not at first glance. Even relatively discerning readers now have to waste time reading more of the text before being able to dismiss it as having substance of little value. So, as I see it, the upset is not simply from AI provenance of a text, but from this level of dishonesty and subterfuge, coupled with the helplessness with which we’re exposed to it.

backlava12 2mo ago

> Assuming that a text is signed off on by a person, then surely what counts is its substance Unfortunately that's a big assumption and for many their "signing off" process will amount to "skimmed it and seems ok?". When I get an LLM-generated doc or runbook, my first thought is that its very possible that I'm the first person who has ever read this. It used to be that writing something took a big time and energy investment up front by one person so that many others could comprehend the ideas with less time and energy needed. LLMs flip this equation around which is a very bad thing.

nunez 2mo ago

I care because in most (not all) cases, (a) AI UGC is one- to few-shotted with no concern for accuracy or readability, and (b) why would I read the results of someone else's prompt if I can achieve more or less the same result for free on ChatGPT? For example, I was searching for a guide for realigning the v-brakes on my wife's hybrid bike and found this: https://volatacycles.com/how-to-adjust-bike-brakes-rubbing/ https://volatacycles.com/how-to-adjust-bike-brakes-rubbing/ Now, this is for disc brakes, so not applicable in my situation. I've never ridden a bike with disc brakes, so I can't vet for the accuracy of this guide. Maybe it's totally right; I'm sure someone reading this will know. However, just look at this article! Zero pictures for a process that _really_ needs it. It just drones on. Imagine following this guide only to discover that the 12-15 Nm rotor torque the article is asking for is actually too high! A similar article from before AI would at least have (stolen) pictures describing the process. It would also very likely be shorter. Which creates the rub: if I know how to write something engaging and correcting AI's copy will take _just as much time_ as writing it myself, why would I use AI to do it? In short, AI content is lazy content. My time is valuable and finite; if I'm reading your stuff, I want to know that you at least tried to give a shit about my time as a reader while producing it.

bluebarbet 2mo ago
Sure, I get that argument. I just don't buy it. What counts is whether the article helps you fix your brakes. If it doesn't, it seems immaterial to me whether or not the bylined author sweated over it for hours. Not least because I have no means of knowing.
- andrewflnr 2mo ago
  GP described multiple reasons it didn't help them fix their brakes.
  - bluebarbet 2mo ago
    So blame whoever signed off on it. AI or not-AI is their problem.

bothers 2mo ago

Why shouldn't people be upset if authors are being replaced by like a handful of the same ghostwriters? An LLM can't convey your ideas and your voice better than you can convey them to the LLM, right? It's a middleman between you and those you want to reach that adds more points of failure, an additional entity that must be understood and made to understand, or else meaning is lost. EDIT: Never mind the motivations behind a majority of LLM-generated articles, which is to be One Unit Whole Content, a colloid for ads.

no_multitudes 2mo ago

Because it sucks! It's badly written, too long, and often pointless. I don't like reading tacky, useless garbage

kcoul 2mo ago
Yeah, I keep feeling the same thing once I know I'm reading something or watching something purely AI-generated - that there are good points being made, but the delivery is just so off that it kind of ruins the points themselves. If people used AI to get the scaffolding of their article going and then re-humanized the entire article, it might be really good, still at a fraction of the time it would have taken to write it themselves, but no one seems to want to expend enough effort to go the extra mile. With video it's pretty well hopeless, unless the AI helps accelerate fully human traditional production techniques. With music, there might be a middle ground, I have had some success getting AI to generate good sounding loops to use in electronic music production, but it was just a matter of brute forcing enough outputs to finally get something decent, which isn't fun at all.