> I don’t write a daily blog to crank out a post every day. If that was the point, I’d have switched to AI long ago already. I write a daily blog to make sure I remember how to think.
I'm always surprised when people say they use LLMs to do stuff in their Journal/Obsidian/Notion. The whole point of those systems is to make you think better, and then you just offload all of that to a computer.
reminds me of descartes. I was mentioning him the other day to my schizo mother in law, who hears voices, to offer the comfort of "I think, therefore I am". The idea being that during her worst episodes she might latch onto the thought that she is thinking, so while the voices _are_ scary; she still exists, she's still in control, she is thinking, therefore she is.
Anyway, I casually mentioned he did a lot of his thinking in an oven and her curiousity was really piqued by that idea. Which is funny because every time I mention it to someone, that's the bit that is most interesting to them. I'm not convinced that an AI would necessarily pick up on that detail being of note as much as a human would.
> The path behind easy only leads to the lowest common denominator. The real artists, fighters, makers—they stick with a truth as old as time itself: The suck is why we’re here, and only those who overcome it themselves will reap all the rewards of their hard labor.
The other day, my wife needed to divide something, and rather than get up and walk to the next room to grab her phone, she did it on pen and paper longhand.
At first I was amazed that she bothered instead of grabbing her phone to do it.
Then it occurred to me that, while more people than I expect probably remember how to divide by hand correctly, I don't think I've actually seen someone do it in years, perhaps since my school days.
I do agree with the author that art is a human endeavor and mastery requires practice... But I'm less optimistic that mass adoption of the easy way will let masters stand out. More likely, they'll just be buried under the deluge of slop the public craves.
> Because I don’t write a daily blog to crank out a post every day. If that was the point, I’d have switched to AI long ago already. I write a daily blog to make sure I remember how to think.
I feel like this will get missed by the general public. What’s the point in generating writing or generating art if it gives next to zero feelings of accomplishment?
I could generate some weird 70s sci fi art, make an Instagram profile around that, barrage the algorithm with my posts and rack up likes. The likes will give that instant dopamine but it will never fill that need of accomplishing something.
I like LLMs to get me to reword something, since I struggle with that. But just like in programming I focus it on a specific sentence or two. Otherwise why am I doing it?
> “Having AI summarize a book or a paper for me is a disaster. It has no idea what I really wanted to know. It would not have made the connections I would have made.”
I don't disagree, but on the other hand, searches are not useless. They're limited because you do need to create a query capturing what it is that you're looking for, in advance. But we do that all the time.
> When you’re stuck and sit there, thinking, trying to come up with what’s next, that’s the valuable part of writing.
Not just what’s next, but the question of what to write in the first place.
I’ve pointed it out before, but this idea of quiet contemplation is exactly where LLMs completely pratfall. The fewer details or instructions you give them, the less novel the output.
I can’t speak for everyone, but when I want to write a new blog post on my site, it’s precisely the opposite. I dim the lights, sit quietly, and let the neurological brownian motion machine do its thing.
Some folks might enjoy writing for the sake of writing. But I'd wager they don't enjoy, say, plumbing for the sake of plumbing? When their toilet is clogged, they call a pro and don't treat it as a journey of personal growth.
I think this works both ways. Your average plumber doesn't enjoy writing. It's something they might need to do from now and then, but if you give them a magic box that solves the problem, they're gonna be overjoyed. One less chore.
Plumbing or writing, I don't think you can convince people not to take shortcuts by telling them "but the fact it's hard is what makes it worthwhile for you!"
What happens when the AI perfects the art of writing? From one turing test goalpost to the next, fooling the human utterly each time until it's forever. Is there a ceiling to this?
> The more I think about it, the happier I am that AI is transforming the world of writing. In a way, I think it’ll make it even easier to stand out—because the more people take shortcuts, the less quality will remain for readers to flock to, even if the overall quantity of options is much larger.
It's easier than ever to be a p99.999% oil painter, but compared with p99.999% film directors basically no one cares at all. Because painting is not in high demand, and film still is, for now.
If the demand for your certain kind of writing vastly diminishes, it is your detriment. AI's supply effect is changing the demand.
George Eliot already wrote a p99.99999% novel, Middlemarch. It is only thanks to massive population growth that the number of readers of her novel has increased or remained steady. As a proportion of the population, Middlemarch has no readership, and is a side show of a side show. It has almost completely lost its once hallowed place in society and culture.
>The more I think about it, the happier I am that AI is transforming the world of writing. In a way, I think it’ll make it even easier to stand out—because the more people take shortcuts, the less quality will remain for readers to flock to, even if the overall quantity of options is much larger.
I had some musings about this with respect to blogging. Especially because search engines are now placing their own summaries above SEO-optimized junk posts. Those posts become disincentivized. Hopefully, it leaves us with more people writing blogs for the sake of writing rather than trying to sell clicks.
"...reading actual books in full might now be more valuable than it ever has been..."
Call me old fashioned, but when has this been ever not true? Like yeah, does someone read cliffs notes and go, "that was really edifying and I gleaned incredible insights into myself and the world!!!".
Well, if the demand for people who can and do read full books stays steady and the supply decreases, then the value (or at least the price) will increase. Seems plausible to me.
I wonder if the premium on consistent writing quality is different than the premium we place on consistent novelty.
I'd hazard a guess that from the writer's perspective, novelty scales with volume of thought / connections, which is (at present) a fragmented process and not that well-assisted by AI. OTOH, can "writing quality" be better approximated by LLMs?
> In a way, I think it’ll make it even easier to stand out—because the more people take shortcuts, the less quality will remain for readers to flock to, even if the overall quantity of options is much larger.
I really want this to be the case, but what I've observed so far is that slop networks with thousands of domains and millions generated articles simply drown out everything else. It's becoming increasingly difficult to tell apart pages written by humans from those written by conmen, especially if I'm not an expert on the subject matter.
A cautionary warning about AI I'm starting to use is, "make sure you aren't taking a forklift to the gym". Getting heavy objects of the ground is vastly easier with mechanical assistance, but doing so completely misses the point of lifting weights.
A work related example I have is using AI to generate project plans. LLMs can probably generate an ok project plan for straightforward projects with plenty of examples to be trained on. But perhaps the most important value of generating a plan is the thinking that goes into it. Considering alternatives, likely failures, unlikely failures, etc. In generating the plan you are starting to practice dealing with problem that would come up while implementing it. The knowledge in your head is more valuable than the document produced. The document is just a summary of all the thinking you have done. Essentially a collection of mnemonics. Many details in your head will never make it into the formal plan, but will be needed during implementation.
I strongly recommend "How to Take Smart Notes" by Sonle Ahrens, it gets into how important writing is as a part of the process of thinking and learning.
This is conflating two things: The stuck, and the suck.
As the author says, the time you spend stuck is the time you're actually thinking. The friction is where the work happens.
But being stuck doesn't have to suck. It does suck, most of the time, for most people; but most people have also experienced flow, where you are still thinking hard, but in a way that does not suck.
Current psychotechnology for reducing or removing the suck is very limited. The best you can do is like... meditate a lot. Or take stimulants, maybe. I am optimistic that within the next few decades we will develop much more sophisticated means of un-suckifying these experiences, so that we can dispense with cope like "it's supposed to be unpleasant" once and for all.
The point about recognizing what’s valuable and making sure you don’t outsource that resonates.
The other day I was on LinkedIn and a Chief Design Officer at a notable company posted her reflections on leadership for the year. There were some potentially interesting insights, but they never got past a surface level. The AI-ness of the writing was as clear as day (and GPTZero tagged it as 100% likely to be AI).
It’s disappointing when you see leaders and so-called stewards of taste farming out that part of their voice.
> The more I think about it, the happier I am that AI is transforming the world of writing. In a way, I think it’ll make it even easier to stand out
I totally disagree with this point. It's a combination of wishful thinking and denial. LLMs do a very fine job at writing if you give them the right base of information/insights. I think it will totally obliterate 'writing' as a differentiable skill.
What will happen IMO is that people who have interesting ideas and experiences but suck at writing will have the upper hand. The market for content will be flooded by articles from people who would normally not write. They will feed the LLMs bullet points of interesting facts and observations and let the LLM fill in the gaps and actually make the article engaging. What matters is that the core points have to be interesting. The AI cannot come up with brilliant insights but it can convey brilliant insights really well.
I think even if, hypothetically, some people could tell apart AI-generated content from manually written content, some AI-generated content may actually be more interesting and valuable to read than the manually written one...
At the end of the day, writing by itself doesn't matter; it's just a communication medium. What matters are insights, ideas, concepts, perspectives... It was always about substance, not form. It's a flaw of the human mind that some people used form as a proxy for substance.
There are a lot of people who know a lot and have a lot to say but they were so busy experiencing and learning that they never had time to write before... And even if they did, they could not convey their ideas effectively before.
Now given that LLMs have mastered the superficial aspects of communication, those aspects are no longer valuable and substance is more valuable. But IMO nobody will care whether articles or books were written by AI in the future. It won't have much effect on quality or value of the book/article.
I think what will matter in the future are:
- Insights, ideas, perspectives.
- Media (the most important still); who intermediates content distribution gets to decide what people consume and can shape their perception of quality to a significant extent.
I'm hoping that as more people get involved in writing using LLMs, that it will force more people to confront the second point... People will be forced to pay more attention to substance as it will be the only real differentiator. I'm hoping people will begin to feel disgusted by the low level of substance that current media platforms purvey... It's already kind of happening; people invented the term "AI slop" but really it's not just AI which produces slop. The media has been guilty of spreading slop for quite some time and it kept getting worse. Now AI is just a convenient strawman to bash.
If you don't have the writing skills to express your thoughts in your own language, how can you have the reading skills to be certain the LLM is expressing your thoughts correctly?
If someone, without my permission, used my content to create a replacement of me, for me -- however shitty -- I would probably commit a crime against that person. How are people so brazen? Or rude? Or stupid? Or psychopathic?
> The more I think about it, the happier I am that AI is transforming the world of writing. In a way, I think it’ll make it even easier to stand out
I think this may be a form of denial. The reality is likely the opposite: AI will commoditize the act of writing entirely, shifting the value solely to insight.
For too long, we’ve confused "good writing" with "good thinking." We assumed that if someone wrote beautifully, they had something smart to say. Conversely, we ignored brilliant people simply because they couldn't articulate their complex ideas effectively.
AI fixes this market inefficiency. It allows experts who are too busy actually doing things to finally compete with professional writers. They provide the raw brilliance (the substance), and the AI provides the polish (the form).
When the camera was invented, no one was claiming a photo was a painting.
The camera replaced painters for the "I want to capture a static image" market but not the "I want art as expressed by a painter" market. While tragic for a lot of painters at the time, it does seem like the cost of progress.
In writing, there is no "objective capture" like with cameras; there's no tech that can take a picture of your conscious thoughts and translate that to words on a page in a way that's reproducible. So there is only a single arena of "written expression" that LLMs and traditional writers are competing in. And while there is a strong desired market for "art as expressed by the [human] writer," the product itself is much more difficult to distinguish from the new tech (LLM writing) than a photo from a painting. And the low effort of entry into this desired market with LLM writing is driving its dilution.
The analogy would work if instead they invented a magical camera that could turn any scene into a painting in any art style reliably, and hide the painting's origins.
39 comments
[ 6.0 ms ] story [ 58.6 ms ] threadI'm always surprised when people say they use LLMs to do stuff in their Journal/Obsidian/Notion. The whole point of those systems is to make you think better, and then you just offload all of that to a computer.
Anyway, I casually mentioned he did a lot of his thinking in an oven and her curiousity was really piqued by that idea. Which is funny because every time I mention it to someone, that's the bit that is most interesting to them. I'm not convinced that an AI would necessarily pick up on that detail being of note as much as a human would.
The other day, my wife needed to divide something, and rather than get up and walk to the next room to grab her phone, she did it on pen and paper longhand.
At first I was amazed that she bothered instead of grabbing her phone to do it.
Then it occurred to me that, while more people than I expect probably remember how to divide by hand correctly, I don't think I've actually seen someone do it in years, perhaps since my school days.
I do agree with the author that art is a human endeavor and mastery requires practice... But I'm less optimistic that mass adoption of the easy way will let masters stand out. More likely, they'll just be buried under the deluge of slop the public craves.
I feel like this will get missed by the general public. What’s the point in generating writing or generating art if it gives next to zero feelings of accomplishment?
I could generate some weird 70s sci fi art, make an Instagram profile around that, barrage the algorithm with my posts and rack up likes. The likes will give that instant dopamine but it will never fill that need of accomplishing something.
I like LLMs to get me to reword something, since I struggle with that. But just like in programming I focus it on a specific sentence or two. Otherwise why am I doing it?
It seems to me that it still takes a significant amount of luck to end up actually racking up likes that way.
I don't disagree, but on the other hand, searches are not useless. They're limited because you do need to create a query capturing what it is that you're looking for, in advance. But we do that all the time.
Surely, an AI generated text would have been pedantically correct and used the subjunctive mood there, "If that were..."
> When you’re stuck and sit there, thinking, trying to come up with what’s next, that’s the valuable part of writing.
Not just what’s next, but the question of what to write in the first place.
I’ve pointed it out before, but this idea of quiet contemplation is exactly where LLMs completely pratfall. The fewer details or instructions you give them, the less novel the output.
I can’t speak for everyone, but when I want to write a new blog post on my site, it’s precisely the opposite. I dim the lights, sit quietly, and let the neurological brownian motion machine do its thing.
I think this works both ways. Your average plumber doesn't enjoy writing. It's something they might need to do from now and then, but if you give them a magic box that solves the problem, they're gonna be overjoyed. One less chore.
Plumbing or writing, I don't think you can convince people not to take shortcuts by telling them "but the fact it's hard is what makes it worthwhile for you!"
It's easier than ever to be a p99.999% oil painter, but compared with p99.999% film directors basically no one cares at all. Because painting is not in high demand, and film still is, for now.
If the demand for your certain kind of writing vastly diminishes, it is your detriment. AI's supply effect is changing the demand.
George Eliot already wrote a p99.99999% novel, Middlemarch. It is only thanks to massive population growth that the number of readers of her novel has increased or remained steady. As a proportion of the population, Middlemarch has no readership, and is a side show of a side show. It has almost completely lost its once hallowed place in society and culture.
I had some musings about this with respect to blogging. Especially because search engines are now placing their own summaries above SEO-optimized junk posts. Those posts become disincentivized. Hopefully, it leaves us with more people writing blogs for the sake of writing rather than trying to sell clicks.
Call me old fashioned, but when has this been ever not true? Like yeah, does someone read cliffs notes and go, "that was really edifying and I gleaned incredible insights into myself and the world!!!".
I'd hazard a guess that from the writer's perspective, novelty scales with volume of thought / connections, which is (at present) a fragmented process and not that well-assisted by AI. OTOH, can "writing quality" be better approximated by LLMs?
I really want this to be the case, but what I've observed so far is that slop networks with thousands of domains and millions generated articles simply drown out everything else. It's becoming increasingly difficult to tell apart pages written by humans from those written by conmen, especially if I'm not an expert on the subject matter.
As an incredibly egregious example, here's one of the top results (#1/#2 on duckduckgo) for "wireguard mesh": https://www.ltwireworks.com/blog/how-to-configure-wireguard-.... Yes, it's a grill mesh manufacturer.
A work related example I have is using AI to generate project plans. LLMs can probably generate an ok project plan for straightforward projects with plenty of examples to be trained on. But perhaps the most important value of generating a plan is the thinking that goes into it. Considering alternatives, likely failures, unlikely failures, etc. In generating the plan you are starting to practice dealing with problem that would come up while implementing it. The knowledge in your head is more valuable than the document produced. The document is just a summary of all the thinking you have done. Essentially a collection of mnemonics. Many details in your head will never make it into the formal plan, but will be needed during implementation.
As the author says, the time you spend stuck is the time you're actually thinking. The friction is where the work happens.
But being stuck doesn't have to suck. It does suck, most of the time, for most people; but most people have also experienced flow, where you are still thinking hard, but in a way that does not suck.
Current psychotechnology for reducing or removing the suck is very limited. The best you can do is like... meditate a lot. Or take stimulants, maybe. I am optimistic that within the next few decades we will develop much more sophisticated means of un-suckifying these experiences, so that we can dispense with cope like "it's supposed to be unpleasant" once and for all.
The other day I was on LinkedIn and a Chief Design Officer at a notable company posted her reflections on leadership for the year. There were some potentially interesting insights, but they never got past a surface level. The AI-ness of the writing was as clear as day (and GPTZero tagged it as 100% likely to be AI).
It’s disappointing when you see leaders and so-called stewards of taste farming out that part of their voice.
I totally disagree with this point. It's a combination of wishful thinking and denial. LLMs do a very fine job at writing if you give them the right base of information/insights. I think it will totally obliterate 'writing' as a differentiable skill.
What will happen IMO is that people who have interesting ideas and experiences but suck at writing will have the upper hand. The market for content will be flooded by articles from people who would normally not write. They will feed the LLMs bullet points of interesting facts and observations and let the LLM fill in the gaps and actually make the article engaging. What matters is that the core points have to be interesting. The AI cannot come up with brilliant insights but it can convey brilliant insights really well.
I think even if, hypothetically, some people could tell apart AI-generated content from manually written content, some AI-generated content may actually be more interesting and valuable to read than the manually written one...
At the end of the day, writing by itself doesn't matter; it's just a communication medium. What matters are insights, ideas, concepts, perspectives... It was always about substance, not form. It's a flaw of the human mind that some people used form as a proxy for substance.
There are a lot of people who know a lot and have a lot to say but they were so busy experiencing and learning that they never had time to write before... And even if they did, they could not convey their ideas effectively before.
Now given that LLMs have mastered the superficial aspects of communication, those aspects are no longer valuable and substance is more valuable. But IMO nobody will care whether articles or books were written by AI in the future. It won't have much effect on quality or value of the book/article.
I think what will matter in the future are:
- Insights, ideas, perspectives.
- Media (the most important still); who intermediates content distribution gets to decide what people consume and can shape their perception of quality to a significant extent.
I'm hoping that as more people get involved in writing using LLMs, that it will force more people to confront the second point... People will be forced to pay more attention to substance as it will be the only real differentiator. I'm hoping people will begin to feel disgusted by the low level of substance that current media platforms purvey... It's already kind of happening; people invented the term "AI slop" but really it's not just AI which produces slop. The media has been guilty of spreading slop for quite some time and it kept getting worse. Now AI is just a convenient strawman to bash.
I think this may be a form of denial. The reality is likely the opposite: AI will commoditize the act of writing entirely, shifting the value solely to insight.
For too long, we’ve confused "good writing" with "good thinking." We assumed that if someone wrote beautifully, they had something smart to say. Conversely, we ignored brilliant people simply because they couldn't articulate their complex ideas effectively.
AI fixes this market inefficiency. It allows experts who are too busy actually doing things to finally compete with professional writers. They provide the raw brilliance (the substance), and the AI provides the polish (the form).
casual painting also "makes you remember how to see" and stuff - that doesn't mean that taking photos stop you. It's just different
The camera replaced painters for the "I want to capture a static image" market but not the "I want art as expressed by a painter" market. While tragic for a lot of painters at the time, it does seem like the cost of progress.
In writing, there is no "objective capture" like with cameras; there's no tech that can take a picture of your conscious thoughts and translate that to words on a page in a way that's reproducible. So there is only a single arena of "written expression" that LLMs and traditional writers are competing in. And while there is a strong desired market for "art as expressed by the [human] writer," the product itself is much more difficult to distinguish from the new tech (LLM writing) than a photo from a painting. And the low effort of entry into this desired market with LLM writing is driving its dilution.
The analogy would work if instead they invented a magical camera that could turn any scene into a painting in any art style reliably, and hide the painting's origins.