I love that O’Connor quote. It’s also useful for distinguishing between things you really need to read (like her short stories) and the huge number of books (not just user manuals and the like) where you really can just extract some information without any loss.
Kling's argument is a nonstarter: the opportunity cost of reading a book for 20 minutes and reading an online essay (or Claude) for 20 minutes is identical, regardless of how much the pool of information is increasing.
But how are you increasing epistemic dependence by replacing the original with a copy? That seems like a 1:1 trade. More broadly, how does relying on AI place you in an echo chamber if it has wide coverage? Of course you should read creative work directly, but in philosophy what matters is the ideas not the guru's words.
I don't vibe-read, but if it means actively engaging with the ideas in a text, asking questions, dialogue, etc., instead of just passively reading, maybe I should.
I think Kling's implicit premise is that most non-fiction books are bloated essays, where all the interesting information could have been conveyed in 20-40 pages. Outside of academic philosophy, quantitative science, and history, I think this is pretty close to correct. Nobody needs to read the whole thing of an Adam Grant book to get the important takeaway.
“Kling's argument is a nonstarter: the opportunity cost of reading a book for 20 minutes and reading an online essay (or Claude) for 20 minutes is identical, regardless of how much the pool of information is increasing.”
Yes! I thought this argument was so bad that I stopped reading the rest of Kling’s piece.
Also, you didn’t suggest this specifically, but your comment made me think about the time cost of vibe-reading versus of “real” reading, and maybe this was part of Kling’s point that I never got to because I bailed on his essay so early (😅): if I can vibe-read three books in the time it takes to real-read one, maybe the trade off is worth it? More generally, is there some ratio of vibe-read to real-read books in the same time, where I learn more from choosing to vibe-read?
Even more generally, conceding that real-reading a book is better than vibe-reading it, all else equal, what about situations where not all else is equal? What if vibe-reading is faster, and I want to spend the extra time in other ways?
Years ago I wanted to write something about how to not read books. If you can only talk or think about books you have read in their entirety, you’re going to be very limited. So you have to learn how to read book reviews, see discussions of books, skim a few chapters, etc, to expand your reach beyond the things you can actually dedicate your time to.
I think that people should probably do even more of this now that there are AI tools to converse with about books.
But I don’t necessarily think this should mean one should fully read fewer books - it could be that the best response is to do all the reading you had been doing, and *also* vibe-read a lot *more* books.
I think this is broadly fine if you truly are only using to read more than you would otherwise. It does seem to be relatively similar to reading a long book review rather than the book itself. The primary reason I enjoy book review periodicals (such as the New York Review of Books and London Review of Books) is that they give me an opportunity to get a taste of a book I would never have time to read. Long history books are the perfect example of this. There are tons of history books that I see that look interesting but I am never going to have time to read all of the 500 - 1000 page history books that I want. Especially since I also have plenty of philosophy, fiction, law, and science books that I also want to read. So reading the NYRB and LRB reviews of them is a reasonable substitute.
Back when I was in grad school, Gilbert Harman once tried to tell me how easy it was for him, and by extension for anyone else too, to “read” an entire book on a single walk from his home on Prospect Street to the philosophy department at the end of that same street: skip the introduction and conclusion and skim everything in between. I objected that shouldn’t it be the reverse (skip the middle, skim the ends), and he replied that this way you can actually see the arguments. 😅
I’m not sure Kling’s unimpressive piece deserved that level of engagement. Perhaps related to O’Connor’s quote, and more specifically concerning fiction, another way to object to vibe-reading (what an absolutely horrible phrase) along the lines of your second objection is to come up with analogies to vibe reading across art forms. Kling seems to treats books are mere receptacles of information to be extracted, as you say. But compare this with music or movies or painting. It would be absurd to ‘vibe listen’ to Bach, ‘vibe watch’ The Third Man or ‘vibe look’ at a Rembrandt, just as you can’t learn about art history without ever experiencing actual art. When you zero in on literature and treat it as more than just digestible information, Kling’s case looks pretty pathetic. He might be right about a lot of nonfiction but then he’s addressing a strawman. When people say you should read more books they mean Austen, Tolstoy, Melville, Dostoyevsky, or O’Connor, not the latest releases in the nonfiction corner of your local Barnes and Noble.
These things are absurd as replacements for actual engagement with the work. But no one can actually engage with every great work of art. Having read reviews of and learned a bit about a large penumbra beyond what one has actually engaged with directly can be beneficial for choosing what next to devote one’s attention to.
Oh absolutely, and there’s a tradeoff between depth and breadth of engagement. But I’m worried that ‘read fewer books’ can lead one on a slippery slope of superficial engagement. And one will benefit from breadth only insofar as one knows what the large penumbra is a penumbra of.
Yeah, I think putting it as “read fewer books” is wrong. But I also think putting it as “against vibe-reading” is wrong. Read as many books as you used to, and vibe-read more!
I know of many film scholars (my field) who prefer to read books about films rather than watching the films themselves. This undoubtedly is one reason why so many claims, about specific films, movements, genres, and so on, that are just plain wrong are perpetuated decade after decade. What some prominent people once wrote just gets repeated, ossified into facts, however wrong or absurd they might be. It's not unique to film studies of course. Maybe more common in the humanities and social sciences, especially the part that's enamoured by Theory, than other academic fields.
And when I discuss films and film history with LLM's they're invariably wrong about a lot of things, for the above stated reason. They're not hallucinating much anymore but they're not able to reach beyond the conventional wisdom, however wrong it may be.
Another benefit of reading books is serendipity. For non-fiction, I'm usually drawn to a book because I find a particular topic or question interesting. But because books are long and rich, I often encounter new ideas or information I would have never encountered otherwise, even if I had vibe-read the book, which sparks new ideas and interests.
I think using them as a screening/filtering mechanism is very alluring, but not without its own dangers. First, as others have noted, learning how to skim and where to focus on dense texts is important, and one doesn't develop that skill without being handed giant amounts of material and obviously too little time to get through it exhaustively.
But also, I wonder about the prior that something is poorly written and padded with BS and thus running it through AI before engaging with it seriously (or at all). I could imagine people doing that with...student essays. Or Substack posts by an unfamiliar writer. And maybe that presumption turns out justified much of the time. But I think there's something quite corrosive about having that prior in the first place, never mind what's downstream of tech-mediated indulgence of it. Like, the only people worth the effort of human engagement without checking with the machine first...are those who are already well-known?
This year I set myself the goal of reading 50 books. This was because I'm sceptical that "4% of Americans" read 50 books a year. I am currently on 30. The deadline is 30th of November. Thank you for your attention to this matter.
But how many are you writing?? Isn't that what the bunch of y'all do all day? So we can then vibe-read them and brag about the whole-ass contemporary philosophy books we've read?
Nice points against vibe-reading. It also does not acknowledge the ways form and content are inseparable in many areas of the humanities, including the great works of literature and philosophy.
>This is why I’m worried about a future where everybody trusts a homogenous pool of AIs. Even if the AIs make us all smarter, we’re going to all end up more epistemically dependent, which means potentially trapping ourselves inside a massive echo chamber. For example, suppose I come to think the stock market is going to go up because my AI and neighbor both tell me it will. I take myself to have two independent sources—but in fact my neighbor is basing her take on her own AI, which is just like mine. Here the AI is the guru, and my neighbor is the follower. By aggregating my two sources, I’m double-counting. This problem will only grow worse as more and more people share an AI, as their confidence builds and builds until it unleashes an “information cascade” in which everybody is so impressed by everybody else’s apparent confidence that nobody’s willing to speak against the consensus.
Wow, this gets at the heart of my epistemic reservations with AI for the past few years. I'm by no means an orthodox anti-AI absolutist (or even really anti-AI in any strict sense), but those individuals seem to grasp this issue much more naturally than AI neutral or pro AI individuals do.
These are all good arguments against “vibe-reading”. I had not heard this term before your essay but I admittedly have an aversion to calls to ignore books and read a summary, etc—you get a lot out of doing the reading. (Same goes for reading a frequently cited paper for yourself to see whether the way it’s cited is actually legitimate and what else it has to say.)
Another element, I think, is that it’s just really easy to fool yourself when you’re talking with AI, which means it’s easy to overestimate how much you’re getting out of vibe-reading. That is: I can imagine someone describing a scenario in which they’re having wonderful, illuminating dialogues with their llm, and it’s true that those described scenarios are harder to argue against—but I think we’re prone to overestimating the rate at which they actually happen vs the rate at which an llm is just being sycophantic.
I love that O’Connor quote. It’s also useful for distinguishing between things you really need to read (like her short stories) and the huge number of books (not just user manuals and the like) where you really can just extract some information without any loss.
Great way to put it.
Kling's argument is a nonstarter: the opportunity cost of reading a book for 20 minutes and reading an online essay (or Claude) for 20 minutes is identical, regardless of how much the pool of information is increasing.
But how are you increasing epistemic dependence by replacing the original with a copy? That seems like a 1:1 trade. More broadly, how does relying on AI place you in an echo chamber if it has wide coverage? Of course you should read creative work directly, but in philosophy what matters is the ideas not the guru's words.
I don't vibe-read, but if it means actively engaging with the ideas in a text, asking questions, dialogue, etc., instead of just passively reading, maybe I should.
I think Kling's implicit premise is that most non-fiction books are bloated essays, where all the interesting information could have been conveyed in 20-40 pages. Outside of academic philosophy, quantitative science, and history, I think this is pretty close to correct. Nobody needs to read the whole thing of an Adam Grant book to get the important takeaway.
I am very sympathetic to that argument. He should have made it instead.
“Kling's argument is a nonstarter: the opportunity cost of reading a book for 20 minutes and reading an online essay (or Claude) for 20 minutes is identical, regardless of how much the pool of information is increasing.”
Yes! I thought this argument was so bad that I stopped reading the rest of Kling’s piece.
Also, you didn’t suggest this specifically, but your comment made me think about the time cost of vibe-reading versus of “real” reading, and maybe this was part of Kling’s point that I never got to because I bailed on his essay so early (😅): if I can vibe-read three books in the time it takes to real-read one, maybe the trade off is worth it? More generally, is there some ratio of vibe-read to real-read books in the same time, where I learn more from choosing to vibe-read?
Even more generally, conceding that real-reading a book is better than vibe-reading it, all else equal, what about situations where not all else is equal? What if vibe-reading is faster, and I want to spend the extra time in other ways?
Years ago I wanted to write something about how to not read books. If you can only talk or think about books you have read in their entirety, you’re going to be very limited. So you have to learn how to read book reviews, see discussions of books, skim a few chapters, etc, to expand your reach beyond the things you can actually dedicate your time to.
I think that people should probably do even more of this now that there are AI tools to converse with about books.
But I don’t necessarily think this should mean one should fully read fewer books - it could be that the best response is to do all the reading you had been doing, and *also* vibe-read a lot *more* books.
I think this is broadly fine if you truly are only using to read more than you would otherwise. It does seem to be relatively similar to reading a long book review rather than the book itself. The primary reason I enjoy book review periodicals (such as the New York Review of Books and London Review of Books) is that they give me an opportunity to get a taste of a book I would never have time to read. Long history books are the perfect example of this. There are tons of history books that I see that look interesting but I am never going to have time to read all of the 500 - 1000 page history books that I want. Especially since I also have plenty of philosophy, fiction, law, and science books that I also want to read. So reading the NYRB and LRB reviews of them is a reasonable substitute.
Back when I was in grad school, Gilbert Harman once tried to tell me how easy it was for him, and by extension for anyone else too, to “read” an entire book on a single walk from his home on Prospect Street to the philosophy department at the end of that same street: skip the introduction and conclusion and skim everything in between. I objected that shouldn’t it be the reverse (skip the middle, skim the ends), and he replied that this way you can actually see the arguments. 😅
I’m not sure Kling’s unimpressive piece deserved that level of engagement. Perhaps related to O’Connor’s quote, and more specifically concerning fiction, another way to object to vibe-reading (what an absolutely horrible phrase) along the lines of your second objection is to come up with analogies to vibe reading across art forms. Kling seems to treats books are mere receptacles of information to be extracted, as you say. But compare this with music or movies or painting. It would be absurd to ‘vibe listen’ to Bach, ‘vibe watch’ The Third Man or ‘vibe look’ at a Rembrandt, just as you can’t learn about art history without ever experiencing actual art. When you zero in on literature and treat it as more than just digestible information, Kling’s case looks pretty pathetic. He might be right about a lot of nonfiction but then he’s addressing a strawman. When people say you should read more books they mean Austen, Tolstoy, Melville, Dostoyevsky, or O’Connor, not the latest releases in the nonfiction corner of your local Barnes and Noble.
These things are absurd as replacements for actual engagement with the work. But no one can actually engage with every great work of art. Having read reviews of and learned a bit about a large penumbra beyond what one has actually engaged with directly can be beneficial for choosing what next to devote one’s attention to.
Oh absolutely, and there’s a tradeoff between depth and breadth of engagement. But I’m worried that ‘read fewer books’ can lead one on a slippery slope of superficial engagement. And one will benefit from breadth only insofar as one knows what the large penumbra is a penumbra of.
Yeah, I think putting it as “read fewer books” is wrong. But I also think putting it as “against vibe-reading” is wrong. Read as many books as you used to, and vibe-read more!
Okay I can get behind that.
I know of many film scholars (my field) who prefer to read books about films rather than watching the films themselves. This undoubtedly is one reason why so many claims, about specific films, movements, genres, and so on, that are just plain wrong are perpetuated decade after decade. What some prominent people once wrote just gets repeated, ossified into facts, however wrong or absurd they might be. It's not unique to film studies of course. Maybe more common in the humanities and social sciences, especially the part that's enamoured by Theory, than other academic fields.
And when I discuss films and film history with LLM's they're invariably wrong about a lot of things, for the above stated reason. They're not hallucinating much anymore but they're not able to reach beyond the conventional wisdom, however wrong it may be.
Another benefit of reading books is serendipity. For non-fiction, I'm usually drawn to a book because I find a particular topic or question interesting. But because books are long and rich, I often encounter new ideas or information I would have never encountered otherwise, even if I had vibe-read the book, which sparks new ideas and interests.
Great points about independence of sources.
I think using them as a screening/filtering mechanism is very alluring, but not without its own dangers. First, as others have noted, learning how to skim and where to focus on dense texts is important, and one doesn't develop that skill without being handed giant amounts of material and obviously too little time to get through it exhaustively.
But also, I wonder about the prior that something is poorly written and padded with BS and thus running it through AI before engaging with it seriously (or at all). I could imagine people doing that with...student essays. Or Substack posts by an unfamiliar writer. And maybe that presumption turns out justified much of the time. But I think there's something quite corrosive about having that prior in the first place, never mind what's downstream of tech-mediated indulgence of it. Like, the only people worth the effort of human engagement without checking with the machine first...are those who are already well-known?
This year I set myself the goal of reading 50 books. This was because I'm sceptical that "4% of Americans" read 50 books a year. I am currently on 30. The deadline is 30th of November. Thank you for your attention to this matter.
PHILOSOPHER PAUL R. SAGAR
But how many are you writing?? Isn't that what the bunch of y'all do all day? So we can then vibe-read them and brag about the whole-ass contemporary philosophy books we've read?
0 at the moment. But if the Leverhulme foundation gives me a stupid amount of money, I'll probably have to start one. (Spoiler: they won't.)
Nice points against vibe-reading. It also does not acknowledge the ways form and content are inseparable in many areas of the humanities, including the great works of literature and philosophy.
>This is why I’m worried about a future where everybody trusts a homogenous pool of AIs. Even if the AIs make us all smarter, we’re going to all end up more epistemically dependent, which means potentially trapping ourselves inside a massive echo chamber. For example, suppose I come to think the stock market is going to go up because my AI and neighbor both tell me it will. I take myself to have two independent sources—but in fact my neighbor is basing her take on her own AI, which is just like mine. Here the AI is the guru, and my neighbor is the follower. By aggregating my two sources, I’m double-counting. This problem will only grow worse as more and more people share an AI, as their confidence builds and builds until it unleashes an “information cascade” in which everybody is so impressed by everybody else’s apparent confidence that nobody’s willing to speak against the consensus.
Wow, this gets at the heart of my epistemic reservations with AI for the past few years. I'm by no means an orthodox anti-AI absolutist (or even really anti-AI in any strict sense), but those individuals seem to grasp this issue much more naturally than AI neutral or pro AI individuals do.
These are all good arguments against “vibe-reading”. I had not heard this term before your essay but I admittedly have an aversion to calls to ignore books and read a summary, etc—you get a lot out of doing the reading. (Same goes for reading a frequently cited paper for yourself to see whether the way it’s cited is actually legitimate and what else it has to say.)
Another element, I think, is that it’s just really easy to fool yourself when you’re talking with AI, which means it’s easy to overestimate how much you’re getting out of vibe-reading. That is: I can imagine someone describing a scenario in which they’re having wonderful, illuminating dialogues with their llm, and it’s true that those described scenarios are harder to argue against—but I think we’re prone to overestimating the rate at which they actually happen vs the rate at which an llm is just being sycophantic.