I would go even further, and say that the idea that an academic paper should be readable by humans in 2026 is strange. The abstract and the introduction should be. The rest of it, as a reader, I can extract by talking to an LLM and asking what goes on in the paper.
I have reviewed tens of papers this way by now, and one thing I'm certain of is that I get much better understanding of the paper this way. This does not save me time usually. If anything, I now spend more time reviewing a paper, but when I'm done I know more about the paper and if I need to provide feedback, it is more substantial.
I would still prefer the abstract and the intro to be written for humans, as a sanity check. But whether the authors pored over every word, every word is "theirs" does not seem to matter much. If the ideas are good they are good, and if they are bad good writing is not a way to compensate.
Two additional observations here:
* That separation of high level and technical content is not new. Conference papers for a long time now are limited to ~9 pages and Appendix, and it was a well known secret that no one ever reads the Appendix, which contained all proofs, for instance (for all practical purposes. and I did, for the record). Up to now that is, now we do read it again with LLM. No point not to do it.
* There may be a finer point of whether one can separate the ideas from the language used to convey them. But I think it can, and again, we did it for a long time wit the appendices.
I don't think "judge writing by its quality instead of its source" is coherent. Certainly, the value of reading is in the thoughts it produces in the reader's mind. So if you want me to ignore the source, you're asking me to value the thoughts I get from the writing without regard for the thoughts that produced it. This seems dubious, if not actually oxymoronic. As a reader, it feels much like you telling me to eat something you made, which you enthusiastically assure me is high quality, but which you can't be bothered to eat yourself, and in fact you've never eaten anything like it before. Plus it smells like spray-cheese.
> you're asking me to value the thoughts I get from the writing without regard for the thoughts that produced it. This seems dubious, if not actually oxymoronic
The archaeological record is loaded with texts of dubious or unknown authorship. Many of our oldest classical texts fall into this category [1].
That doesn't make them "oxymoronic" to engage with. Just more difficult. From a time when few were writing and fewer of their works survived, this is worth the effort. When a nickel of electricity can produce text, it's not–we need the filter of authorship to decide how much effort we, as humans, should expend in engaging.
Are you trying to argue that archeological texts weren't written by humans? Because otherwise I don't understand how your comment is related to the parent comment. It has nothing to do with not knowing the identity of the author. It's about the author being a human.
> trying to argue that archeological texts weren't written by humans?
I’m arguing they’re minds of unknowable foreignness to our own. We fill in the gaps because the text per se has meaning. If anything, we give those minds meaning through the text, not the other way around.
We impute “the thoughts that produced” those writings. One can just as easily contemplate the method through which human writing got distilled into a matrix of numbers that produced a particular output in response to a prompt.
(I’ve absolutely read works without a care in the world for the author.)
In Jorge Luis Borges's short story "Pierre Menard, Author of the Quixote" we learn that Menard, the fictional 20th century writer of the eponymous title, immerses himself in the language and culture of 16th century Spain (pre-1602, to be precise), to the point where he writes a chapters of a book which are word-for-word identical parts of Don Quixote.
Yet the literary criticism of Borges's short story points out how the two texts have different messages.
For Cervantes, this is a book for the people, written in the language of the people.
For Menard, this is a book for specialists, who would understand the old Spanish language.
In the 20th century of Bertrand Russell and William James, the philosophical concepts that an educated writer like Menard would know are far different than 16th century Spain, so an expression like "history, the mother of truth" has different meanings.
Since you mention Homer, if I were to write of the wine-dark sea you would know I am referring to Homer. It brings along cultural connections, including its own Wikipedia page.
While when Homer uses it, it doesn't have those connections.
> New Criticism developed as a reaction to the older philological and literary history schools of New England, which focused on the history and meaning of individual words and their relation to foreign and ancient languages, comparative sources, and the biographical circumstances of the authors, taking this approach under the influence of nineteenth-century German scholarship. The New Critics felt that this approach tended to distract from the text and meaning of a poem and entirely neglect its aesthetic qualities in favor of teaching about external factors. ...
> It was frequently alleged that the New Criticism treated literary texts as autonomous and divorced from historical context, and that its practitioners were "uninterested in the human meaning, the social function and effect of literature."
Borges's story is a commentary about this debate.
While the linked-to essay in the Chronicle made no mention of the long discussion of the relevant literary theories, seemingly insisting that analytical writing somehow stands outside that debate, by simply ignoring the debate.
One of my favorite New Criticism corollaries is that an author, after a work has been published, is not in a privileged position from which to speak on its meaning. Tolkien qua author is omnipotent. Tolkien qua post-publication commenter is another speaker, albeit a highly knowledgeable one. (Purposely picking an example where the author re-wrote his works after publication, e.g. retconning the Ring into The Hobbit.)
> One of my favorite New Criticism corollaries is that an author, after a work has been published, is not in a privileged position from which to speak on its meaning.
That's a line which was more appropriate in a era before the majority of novels had sequels. Now we read series where the plot extends across multiple books.
Events in earlier books set up situations to be resolved in later books.
For a recent controversy in this area, ask Google "Is Dumbledore gay?"
...I'm not sure if you're agreeing with me? The identity of the author is not germane to my point, only whether there actually is an author, aka an actual mind. You could argue those old texts are worth engaging with exactly because they're our only links to those minds.
Ok, but when you're reading Homer or Socrates or the Bible (we're relatively sure it has several different authors), you're still engaging with a communal tradition of fellow readers who, in large part, define what it actually means to understand the text. If you read the Odyssey and decide it was actually about Pokemon, you're misreading it.
> you're still engaging with a communal tradition of fellow readers
Inasmuch as you’re reading translations upon translations, sure. But none of that is necessary.
(One could credibly argue that engaging with an LLM is engaging with, in a sense, its training corpus.)
> If you read the Odyssey and decide it was actually about Pokemon, you're misreading it
If you can stand by your reading based on the text, I say go for it. It doesn’t work in your example. But there are plenty of cases where what was once an against-the-grain interpretation dived into and out of the mainstream.
I want to hear directly from the finest minds humanity has to offer. When I want something from AI, I'll ask AI myself. In short, I want to know when something is human-authored vs. AI-authored. It really matters to me and I suspect to others too.
While I broadly agree, I'm sympathetic to making some use of using AI when writers with an intermediate command of English are publishing papers in English.
Because, as a German academic once explained to me (years before the rise of LLMs), if you want your ideas to be influential you need people to read your papers, and you get a heck of a lot more readers when publishing in English than you do publishing in German.
Unfortunately, it's much more difficult to draw a clear line once you start allowing some AI writing tools.
are papers really still something to be consumed by humans though?
Papers themselves become a source, something to check back on.
They dont write themselves, it takes significant effort to make a high quality paper even with the help of AI.
And if you write a review paper that is a needed piece of literature, but this is also something that can be perfectly digested by a well crafted paper harness. and review hundreds of papers in the time you review 3.
what is it going to be? Speed of scientific progress or human crafted prose?
This is such a great topic. On one hand academics writings are by default seen as original and a work of deep research. Unique voice of writers adds to the originality.
On the other hand AI can really be a good editor and give a final polish. But this path is fraught with danger as eventually "Editor" might turn into a "Writer"
This is something like handcrafted leather bag, quality assurance of parts like zips are okay but value lies in the craftsmanship rather than industrial production.
Another point is if writing can be done by AI, so can peer review. Then 100s of agents are just reviewing it and no real exchange of thoughts or ideas are happening. That, I feel would be a sorry state to be in
> This is something like handcrafted leather bag, quality assurance of parts like zips are okay but value lies in the craftsmanship rather than industrial production.
Nah. The value is in the branding.
There was an embarrassment a few years ago when it was discovered that one maker of fake Rolex watches was using the same movements as the branded version.
Agreed and my point is "hand crafted" itself is a "branding" and the marketing around "limited" edition and not mass produced add to the allure.
Regarding fakery I think even "Monalisa" could be faked that is inseparable from the original. We hear so many fake paintings are sold for million dollars after all the due diligence by experts. Nothing is sacrosanct against deliberate fakery but the demand and the allure of the original remain for the aficinados.
Many people here complain that AI writing is bad. And I agree with it, AI writing is boring and its distinctive tone is now associated with unpleasant thoughts for most people including me.
But I'm curious about how people will feel when AI writing is good and impossible to differentiate from a decent journalist-level writer, including with tooling. Then what? Are people going to stop reading?
I'm trying to separate the part of the argument related to "this writing is bad" from "this was written by AI".
I personally don't care if AI is a writing instrument for a human who's signing off on the content, as long as the writing is good and the source is legitimate.
The fundamental problem with most AI "writing" is that writing is a communication medium, and if your text was just generated by a prompt, I would always rather read the prompt than the output, as the output contains zero additional bits of information.
When I write using LLMs, each paragraph may have dozens of iterations. The prompt is a lot longer than the resulting text and probably not really understandable. I'm sure nobody would want to read those prompts.
The comparison to painting before and after photography was invented is the closest to how I think about writing before and after AI.
We remember museum-quality painting, but there was a whole economy built around people getting portraits done. It was time-consuming and expensive. Then the camera came along and completely met that need with a far superior result (an accurate portrait) while being much cheaper and faster. So too has the listicle writer all but gone extinct.
But what about art? Painting went in a wild new direction, exploring the medium in a direct counter-trend from portraiture. In the same way, I expect "literary" and "experimental" writing become more outlandish while those middle of the road beach romance novel authors will shrivel up completely or flatten the market into a endless flood of options, possibly self-guided by readers.
The most interesting is what happened to photography, which quickly became its own art. Just as preposterous as saying prompting LLMs could be a form of artistic expression, photography was totally dismissed at first. But photography requires a much different skill set and type of discipline, while relying on the same visual concepts as painting.
If entire novels are going to be generated by AI, someone needs to write that prompt. What will it say? What concept will lead to a bestseller? This may seem reductive in extreme, but it puts a similar focus on the remaining human choices. If photos or prompts are relatively cheap, the expression of art becomes one of curation, instead of creation. Flower arranging and DJing are both curation and both expressive.
> But I'm curious about how people will feel when AI writing is good
This is the question I keep asking the people who seem to really hate AI. At some point it will produce good writing, both fiction and non-fiction, and then what? Will people stop reading?
Someone commented over the weekend they had an LLM generate a sci-fi short story and their community loved it until they told them it was LLM generated.
I’ve also noticed a new type of ad-hominem attack (especially here on HN) where instead of engaging with an argument, people claim a comment is AI and move on. HN users have found 10 of the last 2 LLM generated posts.
Well, the problem I have with this is that I'm paying academics. At the point where they're a proxy between my money as allocated by the government and an AI provider, I no longer wish to pay them. I am capable of paying the AI provider myself. I'm not upset about the output being what it is. That's fine. I just don't care so much about the being wishing to be paid to intermediate me and the machine I am fully capable of prompting without the aid of a professional hand-holder.
> That is why the model needs far more instruction to write than to code. I deploy it for its strengths and am constantly on alert for its weaknesses, which is what makes it a tool and not a ghostwriter.
Everyone is convinced of these grand truths - that they alone have contributed something of value to the machine work. But that value is extractable by almost anyone who interrogates the machine.
Like everyone else alive, I too believe this is monocausal. This is all a problem caused primarily by a distinction between the one who sees the supposed benefit of the academic (therefore induced to pay) and the one allocating resources to the academic. It is time we remove this intermediary who does this. Should you desire to read analytical writing by a professional machine speaker there is no doubt he will gladly take your money to talk to the AI for you. Let there be a market for academia - the mathematicians can be paid for the beauty they wish to glimpse between coffees and the philosophers can prompt their machines for money.
For the rest of us, release us from this compulsory taxation so that the plumber - should he so desire - can ask Fable 5.1 "Why should I pay academics to use AI to write?" and receive a satisfactory answer without needing to pay out of his taxes a compulsory fee for an answer to a question he may never desire to ask.
If that was true then slack, linear, … would be out of business.
Yes, anyone can pay the provider, but not everyone can ask the right questions, followup, generally give the question the attention it deserves.
There are heaps of problems with academia, and poor writing is just one of them. It existed pre AI. Whoever produced slop pre-AI now produces more of it. The good ones still stand out.
Maybe they stand out, or maybe the pool of papers just get flooded and nobody new good gets noticed unless they were already preselected by other means.
Artificial intelligence systems should support our reasoning power rather than substituting our critical thinking, independent thought and reasoning power.
Me. Because I used to love reviewing conference papers and AI has utterly ruined it.
If the end result was reliably good people wouldn't be hating on it, but the vast majority of people are using AI not to elevate their output quality, but to churn out slop quantity, because that's where the incentives push them.
Anyone who actually engages with the output can tell that the speedups you get on producing good prose, maintainable codebases or high quality abstractions are much much less because you can't let the LLM go brrr, so AI output tends to be drivel, and banning it is a good way to prefilter.
It'd the same reason journals and conferences desk reject you if you don't follow their formating rules or don't use their template - it's a price of admission that ensures some consistent quality level, and if you couldn't be bothered to learn how to follow the rules, why should we trust you that the content is worth out time?
I don't want to see the peer review process (which can be weak enough as it is) overwhelmed by a flood of slop. I don't want reviewers to have to suffer reading AI's bad writing either.
If AI wrote it, it's not your voice. If you can't communicate your thoughts then it reflects a lack of clarity on you as the author. If you lack the clarity, AI assisted anything is probably part of the problem.
If connecting with the audience and guiding them to understand your thoughts is irrelevant, then you are missing the point of writing and being an academic.
There is a reason AI slop is revolting to read. It doesn't care.
"Because I was first trained as a mathematician, I probably write more like an AI than most people. But that particular article had been through 10 drafts. I worked on it with Claude Fable 5.1 and other large language models, as I now work on nearly everything, but I reviewed every line and rewrote most of them. By draft 10, I could not have said where the machine’s contribution stopped and mine began.?"
...
"The author drafted this essay in extended collaboration with Claude Fable 5.1."
-----------------
If you read a lot of journal articles, you're probably accustomed to occasionally running into articles a ways outside your bailiwick that are written poorly, you don't really understand, which you suspect made it past reviewers who didn't really understand, and indeed, you suspect the authors didn't really understand. Everyone involved just gave them the benefit of the doubt because they were too busy to figure out what the tortured language really meant. However, you don't know for sure, so you decide to leave it up to people in that particular field to deal with.
I can't speak for others, but I find AI writing to be frequently similar in its effect on me to bad writing. Yes, the grammar is correct, but it beats around the bush and puts emphasis on the wrong things. Even if you filter Claude through a human, those poor choices tend to shine through. This subtly inhuman voice switches your brain off. I suspect a lot of bad science is going to slip past reviewers because of LLM writing, and a lot of good science is going to slip past those who might need to read it, because the LLM co-written article switched their brains off and failed to make itself understood.
I understand the notion that, if the content is good enough, it shouldn't matter if the article is written poorly or co-written with AI. However, if the content is good enough, isn't it worth communicating it well?
Good point about emphasis. Correctness in emphasis, and I do mean correctness, was already underrated. It's a critical measure of whether you actually understand a topic, and an indicator of values in general.
I am on two program committees for major CS conferences this year.
LLM writing is horrible, full stop. The vast majority of papers that I see with obvious LLM writing ticks are also frankly incomprehesible. Claude's biggest sin at the moment is inventing new terminology (sometimes multiple times in a single sentence!) without stopping to explain what it means.
In previous years, I would see a substantial fraction (maybe 30-50%) of papers that were obviously written by non-native English speakers, and full of grammatical mistakes and typos. I have seen zero papers like that this year. Honestly, I preferred those: reading LLM writing is painful by comparison.
So far I have seen exactly one paper that had obvious LLM writing ticks, and yet had gone through enough passes (either of the LLM or the human authors) that the paper itself was comprehensible. So I'll admit that it's possible to do. But based on what I've seen this is vastly in the minority.
I honestly used to enjoy the review process; even when the papers were bad at least I could usually see the effort that went into it and give some constructive guidance on how to improve. The process is dramatically less pleasant now, and it's not obvious if I'm accomplishing anything: if authors who write with LLMs just stick the reviews back into those same LLMs, will they learn anything? I suspect not.
Maybe some day LLMs will be good enough to write, but as of today: please, please, just don't do it. The typos and grammatical mistakes are easier to read than the slop.
> The vast majority of papers that I see with obvious LLM writing ticks are also frankly incomprehensible.
That's a classic problem with academic writing. There are too many papers where, after you finally figure out what they're trying to say, you realize they didn't have much to say. Relevant XKCD: [1]
What we seem to be getting from LLMs is more words per unit of useful knowledge.
I wonder if academic papers are now bloated to allow for more ad slots. Many web sites with semi-academic content throw in an ad in the middle of the text every screen height or so. It's beneficial to the publisher if the author is wordy.
The same problem now plagues AI-generated videos, especially the pseudo-historical ones bloated to allow for more YouTube ad slots. These videos have a consistent structure. About 15% into the video, there will be something like "we have to go back to", followed by a large insert of filler material. Around the 50% mark, the relevant content starts. Around the 80% mark, the topic has been covered and there's some general blithering to conclude.
It's enough to make one long for the days of the neckbeard talking head with the big microphone.
I left academia before the AI craze, but I think the answer is the same as for any other type of author/reader relationship: many human readers care! Because writing human prose is almost always an endeavor in communication from a human to a human.
Many readers want to read the thoughts of the author. Filtered only by the author.
I'm fine with using AI for code, where the quality will likely be better than those of academics.
But writing.. FFS everything I see is just AI slop. And I only realize when I start wondering why the paper is going and going and saying nothing, and then pangram tells me that everything past the intro is just AI. :/
The glass half full take is that we are very lucky that there is some much literature that pre-dates LLMs. Our brains are tuned to enjoy that more. People who grow up with LLM slop from the start won't necessarily see it the same way.
Yup. I can read novels from the 1st century [1] or from the 19th century [2] or from the 1970s [3] for the rest of my life (well, until they get banned).
I care more if work is reproducible. Most academics articles are "trust me bro, I am scientist", no source data, no scripts used to process them. With AI I would expect publication of prompts and harness settings.
To be honest, people overvalue human writing too much. Even in academic research, there were quite a lot of cases where the writing was strange. I actually think AI writing is more readable, to that extent. People sometimes compare AI writing with the very best of human writing, but considering that AI writing is above average, I think it’s readable enough.
I would go even further, and say that the idea that an academic paper should be readable by humans in 2026 is strange. The abstract and the introduction should be. The rest of it, as a reader, I can extract by talking to an LLM and asking what goes on in the paper.
I have reviewed tens of papers this way by now, and one thing I'm certain of is that I get much better understanding of the paper this way. This does not save me time usually. If anything, I now spend more time reviewing a paper, but when I'm done I know more about the paper and if I need to provide feedback, it is more substantial.
I would still prefer the abstract and the intro to be written for humans, as a sanity check. But whether the authors pored over every word, every word is "theirs" does not seem to matter much. If the ideas are good they are good, and if they are bad good writing is not a way to compensate.
Two additional observations here:
* That separation of high level and technical content is not new. Conference papers for a long time now are limited to ~9 pages and Appendix, and it was a well known secret that no one ever reads the Appendix, which contained all proofs, for instance (for all practical purposes. and I did, for the record). Up to now that is, now we do read it again with LLM. No point not to do it.
* There may be a finer point of whether one can separate the ideas from the language used to convey them. But I think it can, and again, we did it for a long time wit the appendices.
I don't think "judge writing by its quality instead of its source" is coherent. Certainly, the value of reading is in the thoughts it produces in the reader's mind. So if you want me to ignore the source, you're asking me to value the thoughts I get from the writing without regard for the thoughts that produced it. This seems dubious, if not actually oxymoronic. As a reader, it feels much like you telling me to eat something you made, which you enthusiastically assure me is high quality, but which you can't be bothered to eat yourself, and in fact you've never eaten anything like it before. Plus it smells like spray-cheese.
> you're asking me to value the thoughts I get from the writing without regard for the thoughts that produced it. This seems dubious, if not actually oxymoronic
The archaeological record is loaded with texts of dubious or unknown authorship. Many of our oldest classical texts fall into this category [1].
That doesn't make them "oxymoronic" to engage with. Just more difficult. From a time when few were writing and fewer of their works survived, this is worth the effort. When a nickel of electricity can produce text, it's not–we need the filter of authorship to decide how much effort we, as humans, should expend in engaging.
[1] https://en.wikipedia.org/wiki/Homer#Biography,_identity,_and...
Are you trying to argue that archeological texts weren't written by humans? Because otherwise I don't understand how your comment is related to the parent comment. It has nothing to do with not knowing the identity of the author. It's about the author being a human.
> trying to argue that archeological texts weren't written by humans?
I’m arguing they’re minds of unknowable foreignness to our own. We fill in the gaps because the text per se has meaning. If anything, we give those minds meaning through the text, not the other way around.
We impute “the thoughts that produced” those writings. One can just as easily contemplate the method through which human writing got distilled into a matrix of numbers that produced a particular output in response to a prompt.
(I’ve absolutely read works without a care in the world for the author.)
In Jorge Luis Borges's short story "Pierre Menard, Author of the Quixote" we learn that Menard, the fictional 20th century writer of the eponymous title, immerses himself in the language and culture of 16th century Spain (pre-1602, to be precise), to the point where he writes a chapters of a book which are word-for-word identical parts of Don Quixote.
Yet the literary criticism of Borges's short story points out how the two texts have different messages.
For Cervantes, this is a book for the people, written in the language of the people.
For Menard, this is a book for specialists, who would understand the old Spanish language.
In the 20th century of Bertrand Russell and William James, the philosophical concepts that an educated writer like Menard would know are far different than 16th century Spain, so an expression like "history, the mother of truth" has different meanings.
Since you mention Homer, if I were to write of the wine-dark sea you would know I am referring to Homer. It brings along cultural connections, including its own Wikipedia page.
While when Homer uses it, it doesn't have those connections.
The idea that a text stands alone is a formalist style known as New Criticism, https://en.wikipedia.org/wiki/New_Criticism .
> New Criticism developed as a reaction to the older philological and literary history schools of New England, which focused on the history and meaning of individual words and their relation to foreign and ancient languages, comparative sources, and the biographical circumstances of the authors, taking this approach under the influence of nineteenth-century German scholarship. The New Critics felt that this approach tended to distract from the text and meaning of a poem and entirely neglect its aesthetic qualities in favor of teaching about external factors. ...
> It was frequently alleged that the New Criticism treated literary texts as autonomous and divorced from historical context, and that its practitioners were "uninterested in the human meaning, the social function and effect of literature."
Borges's story is a commentary about this debate.
While the linked-to essay in the Chronicle made no mention of the long discussion of the relevant literary theories, seemingly insisting that analytical writing somehow stands outside that debate, by simply ignoring the debate.
One of my favorite New Criticism corollaries is that an author, after a work has been published, is not in a privileged position from which to speak on its meaning. Tolkien qua author is omnipotent. Tolkien qua post-publication commenter is another speaker, albeit a highly knowledgeable one. (Purposely picking an example where the author re-wrote his works after publication, e.g. retconning the Ring into The Hobbit.)
> One of my favorite New Criticism corollaries is that an author, after a work has been published, is not in a privileged position from which to speak on its meaning.
That's a line which was more appropriate in a era before the majority of novels had sequels. Now we read series where the plot extends across multiple books. Events in earlier books set up situations to be resolved in later books.
For a recent controversy in this area, ask Google "Is Dumbledore gay?"
Such a rich response, much appreciated.
...I'm not sure if you're agreeing with me? The identity of the author is not germane to my point, only whether there actually is an author, aka an actual mind. You could argue those old texts are worth engaging with exactly because they're our only links to those minds.
> whether there actually is an author, aka an actual mind
The mind is long dead. In true Cartesian form, we assume it existed. But it may not be, as with Homer, a single mind.
One can engage with text on its own. Even find meaning in it. Being unable and refusing to are separate categories.
Ok, but when you're reading Homer or Socrates or the Bible (we're relatively sure it has several different authors), you're still engaging with a communal tradition of fellow readers who, in large part, define what it actually means to understand the text. If you read the Odyssey and decide it was actually about Pokemon, you're misreading it.
> you're still engaging with a communal tradition of fellow readers
Inasmuch as you’re reading translations upon translations, sure. But none of that is necessary.
(One could credibly argue that engaging with an LLM is engaging with, in a sense, its training corpus.)
> If you read the Odyssey and decide it was actually about Pokemon, you're misreading it
If you can stand by your reading based on the text, I say go for it. It doesn’t work in your example. But there are plenty of cases where what was once an against-the-grain interpretation dived into and out of the mainstream.
> Inasmuch as you’re reading translations upon translations, sure. But none of that is necessary.
No, even if you read in the original language. Languages are living, communal things.
> Languages are living, communal things
Dead languages [1] absolutely exist.
[1] https://en.wikipedia.org/wiki/Extinct_language
I want to hear directly from the finest minds humanity has to offer. When I want something from AI, I'll ask AI myself. In short, I want to know when something is human-authored vs. AI-authored. It really matters to me and I suspect to others too.
While I broadly agree, I'm sympathetic to making some use of using AI when writers with an intermediate command of English are publishing papers in English.
Because, as a German academic once explained to me (years before the rise of LLMs), if you want your ideas to be influential you need people to read your papers, and you get a heck of a lot more readers when publishing in English than you do publishing in German.
Unfortunately, it's much more difficult to draw a clear line once you start allowing some AI writing tools.
are papers really still something to be consumed by humans though?
Papers themselves become a source, something to check back on. They dont write themselves, it takes significant effort to make a high quality paper even with the help of AI.
And if you write a review paper that is a needed piece of literature, but this is also something that can be perfectly digested by a well crafted paper harness. and review hundreds of papers in the time you review 3.
what is it going to be? Speed of scientific progress or human crafted prose?
This is such a great topic. On one hand academics writings are by default seen as original and a work of deep research. Unique voice of writers adds to the originality.
On the other hand AI can really be a good editor and give a final polish. But this path is fraught with danger as eventually "Editor" might turn into a "Writer"
This is something like handcrafted leather bag, quality assurance of parts like zips are okay but value lies in the craftsmanship rather than industrial production.
Another point is if writing can be done by AI, so can peer review. Then 100s of agents are just reviewing it and no real exchange of thoughts or ideas are happening. That, I feel would be a sorry state to be in
> This is something like handcrafted leather bag, quality assurance of parts like zips are okay but value lies in the craftsmanship rather than industrial production.
Nah. The value is in the branding.
There was an embarrassment a few years ago when it was discovered that one maker of fake Rolex watches was using the same movements as the branded version.
Agreed and my point is "hand crafted" itself is a "branding" and the marketing around "limited" edition and not mass produced add to the allure.
Regarding fakery I think even "Monalisa" could be faked that is inseparable from the original. We hear so many fake paintings are sold for million dollars after all the due diligence by experts. Nothing is sacrosanct against deliberate fakery but the demand and the allure of the original remain for the aficinados.
Many people here complain that AI writing is bad. And I agree with it, AI writing is boring and its distinctive tone is now associated with unpleasant thoughts for most people including me.
But I'm curious about how people will feel when AI writing is good and impossible to differentiate from a decent journalist-level writer, including with tooling. Then what? Are people going to stop reading?
I'm trying to separate the part of the argument related to "this writing is bad" from "this was written by AI".
I personally don't care if AI is a writing instrument for a human who's signing off on the content, as long as the writing is good and the source is legitimate.
The fundamental problem with most AI "writing" is that writing is a communication medium, and if your text was just generated by a prompt, I would always rather read the prompt than the output, as the output contains zero additional bits of information.
When I write using LLMs, each paragraph may have dozens of iterations. The prompt is a lot longer than the resulting text and probably not really understandable. I'm sure nobody would want to read those prompts.
The comparison to painting before and after photography was invented is the closest to how I think about writing before and after AI.
We remember museum-quality painting, but there was a whole economy built around people getting portraits done. It was time-consuming and expensive. Then the camera came along and completely met that need with a far superior result (an accurate portrait) while being much cheaper and faster. So too has the listicle writer all but gone extinct.
But what about art? Painting went in a wild new direction, exploring the medium in a direct counter-trend from portraiture. In the same way, I expect "literary" and "experimental" writing become more outlandish while those middle of the road beach romance novel authors will shrivel up completely or flatten the market into a endless flood of options, possibly self-guided by readers.
The most interesting is what happened to photography, which quickly became its own art. Just as preposterous as saying prompting LLMs could be a form of artistic expression, photography was totally dismissed at first. But photography requires a much different skill set and type of discipline, while relying on the same visual concepts as painting.
If entire novels are going to be generated by AI, someone needs to write that prompt. What will it say? What concept will lead to a bestseller? This may seem reductive in extreme, but it puts a similar focus on the remaining human choices. If photos or prompts are relatively cheap, the expression of art becomes one of curation, instead of creation. Flower arranging and DJing are both curation and both expressive.
> But I'm curious about how people will feel when AI writing is good
This is the question I keep asking the people who seem to really hate AI. At some point it will produce good writing, both fiction and non-fiction, and then what? Will people stop reading?
Someone commented over the weekend they had an LLM generate a sci-fi short story and their community loved it until they told them it was LLM generated.
I’ve also noticed a new type of ad-hominem attack (especially here on HN) where instead of engaging with an argument, people claim a comment is AI and move on. HN users have found 10 of the last 2 LLM generated posts.
Well, the problem I have with this is that I'm paying academics. At the point where they're a proxy between my money as allocated by the government and an AI provider, I no longer wish to pay them. I am capable of paying the AI provider myself. I'm not upset about the output being what it is. That's fine. I just don't care so much about the being wishing to be paid to intermediate me and the machine I am fully capable of prompting without the aid of a professional hand-holder.
> That is why the model needs far more instruction to write than to code. I deploy it for its strengths and am constantly on alert for its weaknesses, which is what makes it a tool and not a ghostwriter.
Everyone is convinced of these grand truths - that they alone have contributed something of value to the machine work. But that value is extractable by almost anyone who interrogates the machine.
Like everyone else alive, I too believe this is monocausal. This is all a problem caused primarily by a distinction between the one who sees the supposed benefit of the academic (therefore induced to pay) and the one allocating resources to the academic. It is time we remove this intermediary who does this. Should you desire to read analytical writing by a professional machine speaker there is no doubt he will gladly take your money to talk to the AI for you. Let there be a market for academia - the mathematicians can be paid for the beauty they wish to glimpse between coffees and the philosophers can prompt their machines for money.
For the rest of us, release us from this compulsory taxation so that the plumber - should he so desire - can ask Fable 5.1 "Why should I pay academics to use AI to write?" and receive a satisfactory answer without needing to pay out of his taxes a compulsory fee for an answer to a question he may never desire to ask.
> I am capable of paying the AI provider myself.
If that was true then slack, linear, … would be out of business.
Yes, anyone can pay the provider, but not everyone can ask the right questions, followup, generally give the question the attention it deserves.
There are heaps of problems with academia, and poor writing is just one of them. It existed pre AI. Whoever produced slop pre-AI now produces more of it. The good ones still stand out.
Maybe they stand out, or maybe the pool of papers just get flooded and nobody new good gets noticed unless they were already preselected by other means.
Artificial intelligence systems should support our reasoning power rather than substituting our critical thinking, independent thought and reasoning power.
Why do you need academics then? Just ask the question to the LLM directly.
Me. Because I used to love reviewing conference papers and AI has utterly ruined it.
If the end result was reliably good people wouldn't be hating on it, but the vast majority of people are using AI not to elevate their output quality, but to churn out slop quantity, because that's where the incentives push them.
Anyone who actually engages with the output can tell that the speedups you get on producing good prose, maintainable codebases or high quality abstractions are much much less because you can't let the LLM go brrr, so AI output tends to be drivel, and banning it is a good way to prefilter.
It'd the same reason journals and conferences desk reject you if you don't follow their formating rules or don't use their template - it's a price of admission that ensures some consistent quality level, and if you couldn't be bothered to learn how to follow the rules, why should we trust you that the content is worth out time?
I care. I don't want to see the brute forcing of academic publication
I don't want to see the peer review process (which can be weak enough as it is) overwhelmed by a flood of slop. I don't want reviewers to have to suffer reading AI's bad writing either.
If AI wrote it, it's not your voice. If you can't communicate your thoughts then it reflects a lack of clarity on you as the author. If you lack the clarity, AI assisted anything is probably part of the problem.
If connecting with the audience and guiding them to understand your thoughts is irrelevant, then you are missing the point of writing and being an academic.
There is a reason AI slop is revolting to read. It doesn't care.
[dead]
https://archive.is/BrQB7
Lets go further: Who cares wheter biological humanoids use ai to think?
Aside from the people who have to live with the results of outsourced thinking, most notably that would be the people selling the thinking.
"Because I was first trained as a mathematician, I probably write more like an AI than most people. But that particular article had been through 10 drafts. I worked on it with Claude Fable 5.1 and other large language models, as I now work on nearly everything, but I reviewed every line and rewrote most of them. By draft 10, I could not have said where the machine’s contribution stopped and mine began.?"
...
"The author drafted this essay in extended collaboration with Claude Fable 5.1."
-----------------
If you read a lot of journal articles, you're probably accustomed to occasionally running into articles a ways outside your bailiwick that are written poorly, you don't really understand, which you suspect made it past reviewers who didn't really understand, and indeed, you suspect the authors didn't really understand. Everyone involved just gave them the benefit of the doubt because they were too busy to figure out what the tortured language really meant. However, you don't know for sure, so you decide to leave it up to people in that particular field to deal with.
I can't speak for others, but I find AI writing to be frequently similar in its effect on me to bad writing. Yes, the grammar is correct, but it beats around the bush and puts emphasis on the wrong things. Even if you filter Claude through a human, those poor choices tend to shine through. This subtly inhuman voice switches your brain off. I suspect a lot of bad science is going to slip past reviewers because of LLM writing, and a lot of good science is going to slip past those who might need to read it, because the LLM co-written article switched their brains off and failed to make itself understood.
I understand the notion that, if the content is good enough, it shouldn't matter if the article is written poorly or co-written with AI. However, if the content is good enough, isn't it worth communicating it well?
Good point about emphasis. Correctness in emphasis, and I do mean correctness, was already underrated. It's a critical measure of whether you actually understand a topic, and an indicator of values in general.
I am on two program committees for major CS conferences this year.
LLM writing is horrible, full stop. The vast majority of papers that I see with obvious LLM writing ticks are also frankly incomprehesible. Claude's biggest sin at the moment is inventing new terminology (sometimes multiple times in a single sentence!) without stopping to explain what it means.
In previous years, I would see a substantial fraction (maybe 30-50%) of papers that were obviously written by non-native English speakers, and full of grammatical mistakes and typos. I have seen zero papers like that this year. Honestly, I preferred those: reading LLM writing is painful by comparison.
So far I have seen exactly one paper that had obvious LLM writing ticks, and yet had gone through enough passes (either of the LLM or the human authors) that the paper itself was comprehensible. So I'll admit that it's possible to do. But based on what I've seen this is vastly in the minority.
I honestly used to enjoy the review process; even when the papers were bad at least I could usually see the effort that went into it and give some constructive guidance on how to improve. The process is dramatically less pleasant now, and it's not obvious if I'm accomplishing anything: if authors who write with LLMs just stick the reviews back into those same LLMs, will they learn anything? I suspect not.
Maybe some day LLMs will be good enough to write, but as of today: please, please, just don't do it. The typos and grammatical mistakes are easier to read than the slop.
> The vast majority of papers that I see with obvious LLM writing ticks are also frankly incomprehensible.
That's a classic problem with academic writing. There are too many papers where, after you finally figure out what they're trying to say, you realize they didn't have much to say. Relevant XKCD: [1]
What we seem to be getting from LLMs is more words per unit of useful knowledge. I wonder if academic papers are now bloated to allow for more ad slots. Many web sites with semi-academic content throw in an ad in the middle of the text every screen height or so. It's beneficial to the publisher if the author is wordy.
The same problem now plagues AI-generated videos, especially the pseudo-historical ones bloated to allow for more YouTube ad slots. These videos have a consistent structure. About 15% into the video, there will be something like "we have to go back to", followed by a large insert of filler material. Around the 50% mark, the relevant content starts. Around the 80% mark, the topic has been covered and there's some general blithering to conclude.
It's enough to make one long for the days of the neckbeard talking head with the big microphone.
[1] https://xkcd.com/451/
> Allison Stanger is a professor of political science at Middlebury College.
Is that so? And that's about it?
I left academia before the AI craze, but I think the answer is the same as for any other type of author/reader relationship: many human readers care! Because writing human prose is almost always an endeavor in communication from a human to a human.
Many readers want to read the thoughts of the author. Filtered only by the author.
I'm fine with using AI for code, where the quality will likely be better than those of academics.
But writing.. FFS everything I see is just AI slop. And I only realize when I start wondering why the paper is going and going and saying nothing, and then pangram tells me that everything past the intro is just AI. :/
The glass half full take is that we are very lucky that there is some much literature that pre-dates LLMs. Our brains are tuned to enjoy that more. People who grow up with LLM slop from the start won't necessarily see it the same way.
Yup. I can read novels from the 1st century [1] or from the 19th century [2] or from the 1970s [3] for the rest of my life (well, until they get banned).
[1] https://en.wikipedia.org/wiki/Callirhoe_(novel)
[2] https://en.wikipedia.org/wiki/Jane_Eyre
[3] https://en.wikipedia.org/wiki/Swords_and_Deviltry
I care more if work is reproducible. Most academics articles are "trust me bro, I am scientist", no source data, no scripts used to process them. With AI I would expect publication of prompts and harness settings.
In many fields, open data is becoming the norm.
[flagged]
To be honest, people overvalue human writing too much. Even in academic research, there were quite a lot of cases where the writing was strange. I actually think AI writing is more readable, to that extent. People sometimes compare AI writing with the very best of human writing, but considering that AI writing is above average, I think it’s readable enough.