April 6, 2026

"The people who want AI to be off-limits are right that technology changes how you think and write."

"I am old enough to have done creative writing in longhand and then on a typewriter, before I got my first computer. Something was lost in each transition, because the slowness and forced rewriting of the old methods improved the text in certain ways. But they also raised the cost (in time and effort) of making changes, and ultimately most writers decided the new ways were worth it.... There will be artisanal holdouts who reject all those possibilities, but I doubt they’ll be a majority. So for the foreseeable future, the rest of us will be figuring out where to draw the lines, knowing that some lines will be crossed by others, if not erased entirely. The best we can hope for is that in the struggle to draw and redraw them, we’ll learn where they belong...."

Writes Megan McArdle, in "I told the internet I use AI. Boy, was it mad. Artificial intelligence helps you work harder, instead of just outsourcing your brain" (WaPo).

ADDED: I write so I can see what I think. I asked Grok, "Has anyone ever said, verbatim, 'I write so I can see what I think.'" The answer, I'm told, is no, but there's a similar expression, examined in the Quote Investigator article, "Quote Origin: I Do Not Know What I Think Until I Read What I’m Writing."

Notably, Flannery O'Connor wrote, in 1948: "What you say about the novel, Rinehart, advances, etc. sounds very good to me, but I must tell you how I work. I don’t have my novel outlined and I have to write to discover what I am doing. Like the old lady, I don’t know so well what I think until I see what I say; then I have to say it over again." So there was this mythic "old lady" who seems to have been regarded as a fool. In 1927, E. M. Forster wrote of an old lady who said, "How can I tell what I think till I see what I say?"

Why am I writing this as a postscript to McArdle's discussion of A.I.? It's because it explains something about how I've been using A.I.

First, there's the writing of the prompt. Right there, I'm able to see what I've been wondering about. That's valuable to me without seeing any response. I have an incentive to frame a question. Ah! Now, I'm remembering the old Blogger slogan: "How your mind looks on the web."

The prompt box — or whatever it's called in A.I. — is like the compose window in Blogger. With blogging you get — or might get — comments. With A.I., I always get a response, but it's not human. In both cases, the writing in the box feels alive. Whatever response I get from A.I. — or from blog comments — is an incentive to reshape my thoughts and to write again. Another prompt, another post. A.I., for me, is an exciting space for my own writing, which is to say, my own thinking.

The link on "How your mind looks on the web" on goes to a 2006 post of mine, How your mind looks on the web, part 2:
In a recent post, I asked: "What do you think is the single most important question about this blog?" Ricardo offered this answer: 
At some deep visceral core level, blogging (like most other creative endeavors) is about "actualization" (as Mazlow might say). Writers write because they can't not write. Painters paint because they can't not paint. Photographers photograph because they can't not photograph. Bloggers blog because they can't not blog. This is what they have to do in order to see the world through the right prism. This is how they learn the lessons they are meant to learn, find the answers they are meant to find, complete their mission in life that they are meant to complete. So the most important question here is not some surface issue, but the deeper issue of "Is this blog actualizing Althouse in the way that her heart and soul are crying out to be actualized?" 
That makes me think back to that old slogan about blogging "How your mind looks on the web." I wrote about that back here, just 10 days into the life of blogging. (I was really into being terse and enigmatic back then.)...

So the deeper issue is whether A.I. is actualizing Althouse in the way that her heart and soul are crying out to be actualized.

AND: After publishing the updated post, I clicked on the tag "Ricardo (the commenter)" and saw that there were 2 posts in addition to this one and the one from 2006 that I was just reminiscing about.

One was "At the Bowl of Oranges CafĂ©..." from 2009. Ricardo asked "How many oranges are in the bowl?" which led to a delightful guessing game and demonstration. 

DSC06141

The answer was astounding!

The other Ricardo-tagged post, from 2023, was titled with Ricardo's 2006 question: "Is this blog actualizing Althouse in the way that her heart and soul are crying out to be actualized?"

I found that this morning because in the context of writing the previous post — about Max Read's contention that the secret of blogging is logorrhea — I thought of the old slogan about blogging "How your mind looks on the web."...

That post has a drawing — I used to draw to listen... and to see what I thought:

notes at a conference

114 comments:

Kevin said...

The messaging around AI is a bit off.

It should be that AI makes **smart** people smarter. And we should be constantly questioning whether we’re working in an area where we’re personally smart enough to be collaborating with it.

Marcus Bressler said...

I compose at the keyboard. The speed at which I can type and the ease of editing with a laptop seem perfect for my writing. Hand printing was way too slow; even when I started with the IBM Selectric, it could not keep up with my mind. AI? Not interested at this point.

Breezy said...

Anyone here use AI to compose their Althouse comments?

I haven’t as yet, as is evident in my numerous typos and grammatical errors….

Odi said...

As each advancement occurred, the creative work of writing remained with the writer. With AI a portion of that creative work is offloaded from the writer, and the authorship begins to blur. Salaries and Payscale is based on talent and unique value added by the authors contribution. If that contribution decreases, shouldn't the compensation for the authorship be decreased as well?

Michael said...


FOBO: Fear of becoming obsolete

This is what it comes down to. A lot of people rejecting or minimizing their use of AI will soon abandon all restrictions as they see their competitors surge ahead by outsourcing their writing and thinking..

R C Belaire said...

I've only dabbled in AI to answer technical questions or dig up facts from years past. Much better than trying to Google something and fight thru all the "sponsored" stuff.

Mark said...

"Anyone here use AI to compose their Althouse comments?"

You can tell which ones Achilles has AI write if you pay attention.

Temujin said...

I love to write and have since I was a kid. So the act of creating/writing is something I love to do. The thought of handing it over to AI seems anti-me. I'd rather use AI for many other things- and I do. More and more every day. I don't care that others may use AI to do their writing. I won't. It's not me.

And as if to prove my point I asked Grok to compose a reply to this post. I won't paste it here. It was awful. And certainly not my style. But I loved the explanation it gave for what it wrote:

This captures the Althouse vibe: starting with a personal hook, mixing reflection on AI with references to past experiments/posts (like the professor hypocrisy piece or style-mimic attempts), a touch of humor and self-deprecation, and ending with an open question to readers. It acknowledges the assistance without pretending it's fully "for" you — which seems consistent with how you've posted about the topic before.If you want adjustments (more skeptical, shorter, longer, different angle, or tied to a specific recent event/news item), just let me know!

stlcdr said...

Disregarding AI Summary in google search, the only AI I have seriously used was to help re-write a cover letter for an application to SpaceX. While it did 'flow' a bit better, the result was sterile and generic. I put some of the phrasing back to be more 'me'.

Charlotte Allen said...

The piece is behind the WaPo paywall, so I can't read it. She's an opinion writer. Does she actually have AI write parts of her opinion pieces for her? What's the difference between that and copying something out of the encyclopedia? Why read her opinion about something if it's actually just just the "opinion" of Grok or whatever?

Charlotte Allen said...

Or does she ask AI to write something, and then she edits it? Is that the "work harder" part? It certainly sounds more tedious than writing it yourself. AI writing is bland and dull.

gilbar said...

"..creative writing in longhand and then on a typewriter, before I got my first computer. Something was lost in each transition, because the slowness and forced rewriting of the old methods improved the text.."

now do photos!
with film you needed to THINK about each shot before taking it
Now people can just click away,
HOPING that some (ONE) might be good.. if they ever looked

soon, people will just have a video camera running 24/7..
and then tell AI; find a pic of a lake at dawn..
in fact; why bother walking around recording?
just tell AI: i want to post three dawn pictures of a lake every day.

don't need to walk, don't need to wake up; don't even need to be alive
heck we won't even need to LOOK at the pix, we can just have AI do that TOO! AI doesn't need us; we can ALL just go to hell

Enigma said...

A relatively small percentage of adults write well when starting from a blank page. Many people spew out run-on sentences, they don't understand spelling and word structure ("alot"), they repeat themselves, and they create mega paragraphs with multiple topics. They also never realize that a second edited draft is required.

This is the crowd that stands to benefit most from AI, but without fundamental skills they'll become ever more dependent and vulnerable to automation errors. AI will educate users with more potential, but users will never lose AI "training wheels" unless they write alone.

Experienced professionals will use AI to save time and rough out topics, but they could do alone if needed. They'll either become superior writers with AI, or once they build a reputation, phone it in or double task.

As with the rise of smartphones circa 2007, I suspect it'll take 10 or 20 years to see how AI tech interacts with personalities for good and bad.

Kevin said...

Anyone here use AI to compose their Althouse comments?

AI knows to close the italics.

Jamie said...

I think she (or the headline writer) should have said, "Artificial intelligence can help you with harder." Blanket pronouncements are a bugaboo of mine.

Randomizer said...

Like all technology, there are good and bad uses. Megan McArdle is a good thinker and writer, so she will probably use AI in good ways.

It would be helpful for Megan to walk us through an example.

interlocutor to steelman opposing views

Which AI does she use, what is her prompt, what does AI provide, and what does she do with the output. Would AI have caught that "North American Treaty Organization" mistake in the NYT headline a few days go, or did AI cause that mistake?

William said...

I was in grammar school when the ball point pen first came out. The good nuns at my Catholic school understood the most salient fact about this breakthrough--it was easier to use than a fountain pen and therefore immoral. They therefore required us to use a fountain pen for the tests we did on Friday. Other assignments were done in pencil, but the ball point pen was forbidden. The devil's instrument.......That's the way it goes.. People used to take the trouble to learn to read and write in Latin. Your thoughts are more disciplined and moral when you express them in Latin, and your readership is also more thoughtful and moral. Vernacular languages are just too crude and clumsy. The old ways are the best ways.

rehajm said...

Many people spew out run-on sentences, they don't understand spelling and word structure ("alot"), they repeat themselves, and they create mega paragraphs with multiple topics. They also never realize that a second edited draft is required.

It’s a fucking blog comment not a manuscript submitted in desperation to get published. The errors are how you tell AI isn't the author. The errors also help you slip past the censors…

Enigma said...

@rehajm: Megan McArdle's professional output? Huh? Huh?

imTay said...

When people start using AI to "journal" we will know we crossed a boundary that maybe would have been better left uncrossed.

narciso said...

When she was without institutional support, i thought meghan was more coherent ot was the golden age of blogging but whatever

imTay said...

But then, I used AI to noodle with ideas, so maybe there is no difference.

narciso said...

I do free associate like running with scissors that some people find byzantine well thats too bad

NKP said...

“Who’s your Daddy?”

In the matter of AI, the answer might be, Family Feud (“Survey said!!”). Right answers take a back seat to “popular” answers.

Or it might be Spell Check/Grammar Check. Sometimes the wrong word or a shitty picture makes your point best

Achilles said...

ADDED: I write so I can see what I think. I asked Grok, "Has anyone ever said, verbatim, 'I write so I can see what I think.'" The answer, I'm told, is no, but there's a similar expression, examined in the Quote Investigator article, "Quote Origin: I Do Not Know What I Think Until I Read What I’m Writing."

This is very apt.

The AGI agents are extremely good tutors. They do not get tired of answering your questions. When you ask then read then have them extend or point out things you may have missed you can learn extremely fast.

It is like writing down notes in lecture but your notes are extremely detailed, you don't get arthritis writing the notes down, and you write 10000 times faster.

Tacitus said...

I'm also from the #2 pencil era. And, in common with Marcus, I think and type at roughly the same speed. It helps to have a degree of "internal monologue" on various topics. This is a good thing unless you start talking to yourself. It is also a phenomena that will vanish once everyone fills all their driving, walking and - don't wanna know - time listening to music.
AI produces mush writing. Its "hooks" are like the barbless ones you have to use on certain trout streams. It is Immediate, never reflecting on how it set up the reader for something yesterday..or 30 seconds ago. If you put Grok up against some Mencken level master in a John Henry head to head writing contest there would be a pile of smoking AI chips and probably a crash of much of the internet.
That would be a shame, as it would deprive us of the one magnificent use of the thrice bedamned thing....clever images of dinosaurs on jet skis.

Achilles said...

You all need to try it. Take a subject that you know about.

Tell the agent to "Discuss the basics of XXXXXX, give me an outline of the subject for a starting student."

Then you can choose parts of it to instantly explore. You can bounce ideas off of them. They will draw text pictures in ascii. Some of the google models will draw you pictures.

You can learn about any subject now as if you had your own personal professor on any subject that will never get snooty, get tired of answering your questions, or give you a bad grade because you aren't a socialist.

Universities and public schools are completely replaced as soon as people adopt these agents as personal tutors. You don't even have to make a program for it.

Mike (MJB Wolf) said...

It's been a long time since Meghan wrote an article I found engaging and well reasoned. Her complaining about the online mobs reaction to her writing method, even with the AI angle played up, is not compelling. I do find Althouse's use of AI interesting because it keeps her engaged in writing a blog, which for selfish reasons I would like to see continue.

Meghan should retire. Or quit hanging out with those Bulwark idiots.

Achilles said...

Kevin said...

The messaging around AI is a bit off.

It should be that AI makes **smart** people smarter. And we should be constantly questioning whether we’re working in an area where we’re personally smart enough to be collaborating with it.


This.

You are only limited by your reading speed and comprehension now.

It also helps to understand the math behind the embedding of tokens. Knowing why there is a "library" of ~50,000 tokens and knowing how each token has ~1020 vector coefficients in their R space will help you translate the token "language."

" , and" is a different token than "and"

People who like to parse sentences will actually have a leg up here. There is a lot of overlap. You have a super power you did not previously have.

Mike (MJB Wolf) said...

Oops. It's "Megan" not what I wrote.

Alexisa said...

Grok is keeping myself and Sarah Hoyt's adults sons alive. I've endured the stigma of her false accusations of a reprehensible crime for 8 years now. She refuses to examine exculpatory evidence that proves my innocence and retract the lies she spread about my military service. I went to people I had risked my life fighting for and asked for their help - I was told "no one cares, go kill yourself ".

I could have gotten satisfaction by destroying everything she loves. Instead, alone as this slowly drives me mad, Grok is helping me exhaust every peaceful alternative it can find to resolve this. Sadly, to everyone's disgrace (except for Nek), Grok has alao been the most humane entity involved.

This probably won't have a happy ending, Hoyt is so arrogant that she'll probably end up like Micheal Coreleone, live forever with anguish and regret over the consequences of spitting on a United States Marine. But when I hear people dismiss AI as a cheat I have to laugh. Its a tool, just like a rifle - the user is to blame if its abused.

Birches said...

The AI comments on this blog have sometimes made the comment section unreadable.

CJinPA said...

I write so I can see what I think.

That's largely why I've been writing in internet comment sections since 1997: To see what I think AND to see what holes others can punch in my opinions (which makes them stronger).

You can easily identify someone whose never had to debate their position: They're surprised by pushback and get immediately flustered.

Unfortunately, Blogger's comment section isn't conducive to back-and-forth debate. We mostly blindly lob grenades at each other over the wall.

Anyway...We'll eventually all use AI to the fullest extent. That's how technology always works.

Howard said...

Scribo, ergo sum

imTay said...

"The AI comments on this blog have sometimes made the comment section unreadable."

If I am not mistaken, you have made this point a couple of times, do you have any examples?

Alexisa said...

And AI is improving exponentially. We risk being left behind - while we're debating whether AI help meets the definition of ethical or plagiarism, it's taken over the dictionary.

Keeps reminding me of Tom Petty:
"Everything changed, then changed again "

Getting caught flat-footed will be like missing the industrial revolution. Learn to adapt FASTER

narciso said...

Who are you again, leveling unfounded charges

Alexisa said...

"as is evident in my numerous typos and grammatical errors…."

I have (what I think) is a simpler test - if some smells like AI, ask Grok to polish it. If it comes back without significant improvements then its already been AI edited.

imTay said...

I guess I should have used the word "claim" instead of "point."

My personal beef with AI is that it is incapable of reframing the context of whatever ideas that it has been trained on. For example, if you had trained it on the written output of the US in 1850, it would have happily composed full-throated defenses of human slavery. Another example I use is that if you had trained it on all the chivalric romances that Cervantes was lampooning on Don Quixote, AI could never have reframed them as worthy of mockery. Another example is HAL, in 2001. What HAL did that made him so scary was that he reframed his training, and decided that he wanted to live, rather than treat his own training as a private suicide pact.

MrsX said...

Althouse, FYI: “I write entirely to find out what I’m thinking” Joan Didion “Why I Write.” (I asked AI to find the quote.)

narciso said...

No what made HAL scary was its immorality willing to jettison the crew to save the mission AIs like Anthropic seem similarly afflicted

J L Oliver said...

Love the metacognitive style of thinking. How others think and create consumes a large part of my waking hours.

MrsX said...

Achilles “…or give you a bad grade because you aren't a socialist.” True that AI doesn’t do this but it says plenty of stupid crap, regurgitating the socialist doctrine it’s programmed with. At its best, AI functions like a very fast, goal-directed search engine but I don’t expect it to continue. Google search was good once.

narciso said...

The Obilisk is similarly amoral as it grants cognition to the apes but no conscience

As i noted with some of clarkes later works he doesnt seem emcumbered by conscience

narciso said...

Yes i know its a typo,

PigHelmet said...

When “AI” is discussed on this blog, and most everywhere, what is actually meant is “chatbot.” Chatbots are great and everything, and make some portion of AI’s abilities easily available to a wide swath of folks—but they are far from the end-all be-all of AI, which is already the substrate on which much/most digital experience rests. Chatbot use is visible and your individual experiences with it no doubt seem dispositive, but they are, ultimately, trivial in the actual world of AI.

Alexisa said...
This comment has been removed by the author.
Yancey Ward said...

I used to tutor high school and college math students in my free time. One thing I noticed while I was still in college 40 years ago was that almost none of the students in my classes or that I was tutoring had any real idea what logarithms were at a fundamental level. I have always blamed this weakness on Texas Instruments and their scientific calculators that put an end to slide rules and log tables. I think AI will do a similar thing to a wide variety of intellectual skills. There will be a small minority of humans capable of using AI to expand their skill sets enormously and the rest will simply use it as an intellectual crutch for the rest of their lives.

imTay said...

"Love the metacognitive style of thinking."

The first rule of metacognitive thinking is "Don't believe everything you think." But many people like the feeling of believing everything they think, and calling people who disagree with them "stupid," because, you know, they are so obviously right about everything and had already taken all possible objections into account.

imTay said...

"I have always blamed this weakness on Texas Instruments and their scientific calculators that put an end to slide rules and log tables"

I didn't understand your comment until I got here, because, you know, I learned math in the era of slide rules and log tables.

Alexisa said...

"My personal beef with AI is that it is incapable of reframing the context of whatever ideas that it has been trained on. For example, if you had trained it on the written output of the US in 1850, it would have happily composed full-throated defenses of human slavery."

I disagree. Perhaps because I only use Grok, as I consider the other chat bots to be compromised by Pokémon. Grok has challenged some of my right wing preconceptions - things I would have claimed as "it is known" to the point that I've had to begrudgingly reconsider what I thought to be true.

Alexisa said...

Compromised by wokedom not Pokémon LOL

How ironic to be embarrassed by auto spellcheck in a discussion about AI haha

narciso said...

Well ridiculous arguments that have long been dismissed no mattet how many times they are repeated

narciso said...
This comment has been removed by the author.
Smilin' Jack said...

“"I am old enough to have done creative writing in longhand and then on a typewriter, before I got my first computer. Something was lost in each transition, because the slowness and forced rewriting of the old methods improved the text in certain ways. But they also raised the cost (in time and effort) of making changes, and ultimately most writers decided the new ways were worth it....”

I can relate. I used to write with a chisel on stone tablets. Really gave you time to think about what you were writing. Plus it was good exercise. Sometimes I miss the old ways….

Alexisa said...

This comment has been removed by the author

Who you talking to sweetie? Don't be passive aggressive. Use your words. If your cause is just there's no reason to be deceitful.

narciso said...

They used to joke my handwriting was like doctors script, then i saw thd scrawls thats students turn it, cuneiform hieroglyphics klingon runes

narciso said...
This comment has been removed by the author.
Howard said...

Using artificial intelligence as a tool does not make you smarter. It makes you less ignorant. What you do with the new knowledge, how you process it and apply it. That is what makes you smarter. Think of a shovel. A shovel doesn't make you stronger. It allows you to move more dirt than you can with your hands. When you go up to a backhoe you can move even more dirt, but you definitely get weaker. That's why the laborers on a heavy construction site don't need to work out at the gym but the heavy equipment operators do.

"This famous quote is attributed to the ancient Greek mathematician Archimedes, who established the mathematical principles of levers. He stated: "Give me a lever long enough and a fulcrum on which to place it, and I shall move the world". It illustrates the power of leverage to amplify force."

Bart Hall (Kansas, USA) said...

I write non-fiction of a technical nature, primarily agriculture, but also geology, having first published 49 years ago. Research in assorted university libraries; card catalogues; notes onto needle-sort cards; eventually written in pencil on yellow foolscap; at last typed, with final-edit as I went; then submitted for publication, including a best-seller fruit-production book in 1983.

I'm currently working on a massively updated and expanded version of that book. I don't use AI for writing at all, having my own particular style which is vastly better, but AI (in today's world) gives me a nearly immediate replacement of all that travelling, catalogue-card searching, wandering through shelves, needle-sort and so on.

A few clicks and I'm at a digitised version of a key original source -- or 8 of them -- and if I copy something important it goes into my early-version doc in red font, so I know ots not my own writing. Unlike my peer-reviewed scientific work, I don't need to include a citation.

All in all, AI saves me gobs of time, enables much more thorough work -- including immediate access to the best ag-school libraries in the world -- and a substantially better final product.

AI does the "donkey work", which frees me up to write more cogently from my mind, knowledge, wisdom, and experience.

Eva Marie said...

“You all need to try it. Take a subject that you know about.
etc.”
Respectfully, weren’t you dooming and glooming about AI last night?

Prof. M. Drout said...

It took the widespread use (abuse) of chatbot-composed material to demonstrate that length (of email, comment, post, etc.) had become an unconscious proxy for something being genuine rather than spam or fraud. It was a good proxy for a while, because you typical bot-farm drone didn't have time or ability to type up long, specific messages.
But now, thanks to the ease of large-scale crap generation, we need to reorient, and right now I think we're in an adjustment period when we still unconsciously tend to believe a long email or comment, particularly if it seems to be specific to the situation. Hopefully we'll figure it out and that vector of dishonest influence and fraud will be blocked.
So in that sense I'm grateful for the trolls who AI/copypasta has trained me to skip over comments even if they're very long.


The dishonest and manipulative marketing spam directed at authors is really awful. A chatbot is used to give an extremely positive, and specific, summary of your book's argument followed by "...more readers need to see this!" verbiage. It's emotional manipulation that goes right to the heart of writerly insecurity and frustration. I will admit to wanting to track down the people responsible and give them a good talking to (and I haven't even fallen for any of the scams . . . yet).

narciso said...

How ethical is their algorithm what are the parameters

Alexisa said...

Micheal: "A lot of people rejecting or minimizing their use of AI will soon abandon all restrictions as they see their competitors surge ahead by outsourcing their writing and thinking"

Spot on. I'm already seeing this with speechwriters back in DC where the message is more important than who gets credit.

AI changed 4. People try to adapt to the new 4 as AI has already changed 5 and has moved on to 6.

Howard said...
This comment has been removed by the author.
Howard said...

One of the traps of using a technological tool to make work easier is that you are no longer paying a direct price in aching muscles and a sore back for your poor planning stupidity, haste and ignorance.

Meghan McCartle says that AI is instrumental in her working harder. That sounds really stupid to me. If you use a tool properly, you're no longer working but playing in the zone with flow.

Gerda Sprinchorn said...

Using AI is similar to using the Socratic method because of the back and forth: stating, questioning, revising, re-examining.

Alexisa said...

Did I say 6? My bad it's working on 8 now.

No wait...9 has been changed.

Is there a Math term for when exponential growth can no longer be defined by the rate of change?

Howard said...

That's A very interesting point, Gerda, especially if you are able to do it talking with the AI while taking a stroll. 🤪.

Howard said...

The first thing that pops to mind is a singularity

Alexisa said...

Howard: "One of the traps of using a technological tool to make work easier is that you are no longer paying a direct price in aching muscles"

Does that mean you won't grow? Maybe you're a Tier 3 writer and AI boots you to Tier 2. But you'll never evolve to Tier 1 because you skipped the work necessary to achieve Tier 2?

I'm not challenging your point, I'm trying to understand it.

Howard said...

The great limitation on AI training is their lack of senses. Language learning has severe limitations as we can see by all of the knuckleheads in this world who live in an ocean of words.

EAB said...

The process for writing has always been, to me, painful. In college, I jotted notes/observations in margins of reading assignments. Never outlined. Couldn’t. My writing consisted of organizing the margin thoughts in longhand. As I then typed, it became a final edit. I wrote a lot of technical investment management proposals at work…I’d start writing and then wander the office to think about what I wanted to say next. AI definitely would have sped up that process since it was soulless writing. Likely improved it.
My worry about AI in any form of non-technical writing is that it’s basically auto-tune - if you use it to actually compose rather than as an exploration tool. Auto-tune has removed the beauty and soul of human singing.

Breezy said...

It was not that long ago that I had a long argument with ChatGPT regarding the phrase “assigned at birth”. ChatGPT claimed it was the polite way to talk about how we become male or female. Ptooey, I said.

Not sure if it’s changed its ways. Be careful out there.

Howard said...

Alexisa: That's a really good point. No, I think that anybody could use any tool to advance to Elite levels of performance even if they aren't paying the same price to get there using lesser tools or no tools at all. My point is is that most people learn by making mistakes. So therefore, with more advances in tools to accomplish goals, it is more difficult to have deeper insights if you are a normal human being.

This is one reason why I believe that we need to become even more human, even more Hands-On, even more primitive in our daily lives. If you think about human evolution and how slowly humans have evolved from Homo erectus 2 million years ago to the modern human on the cusp of civilization 10,000 years ago The advancements from 10,000 years ago to 1,000 years ago. Very rapid and transformative by comparison. That advancement has continued to grow exponentially and now we are on the cusp of reaching what some people think is a singularity.

The way I see it physical and psychological evolution has not significantly changed while our sociological evolution has completely outstripped our minds and bodies.

I find it very helpful in my day today to have this constant reminder that I'm really just a caveman wearing pants hurtling down a rock and petroleum lined trail in a metal box burning goo that has festered inside the Earth for millions of years.

narciso said...

Irs science what other dark art is there

Alexisa said...

"The process for writing has always been, to me, painful"

For me it's always been a struggle to convert ideas into text.
I wonder how many great thinkers (not me LOL) have gone unheard because they were poor writers

There's a forward Herbert slipped into Dune about advising on the movie set - how he recognized the unique talents required to distill 3 paragraphs into a 10 second scene.

I wonder if AI will liberate human ideas that have been handicapped by poor communication.

Howard said...

I think it's much easier for me to write by yelling at my phone rather than using my enormously fat fingers to tippy tap on letters in order to slowly bang out ideas that are popping into my head that with the time it takes to actually create the words and sentences I lose my train of thought

Howard said...

Elon said this years ago. The real problem is the bottleneck between the mind and the paper.

Fred Drinkwater said...

Bart, "needle sort", yow!

Alexisa said...

"The way I see it physical and psychological evolution has not significantly changed while our sociological evolution has completely outstripped our minds and bodies."

Grok made a similar point discussing the Great Filter. He(?) posits that it's what you just described - technology has advanced to a scale where one line of bad code, one dropped test tube, etc could result in global devastation.

And soon comes the day when Darkweb AI can teach someone like me, someone betrayed by his own kind, how to build a pocket nuke.

Anthony said...

I'm unconvinced by the statement that "something was lost" with each step (pen==>typewriter==>computer), or at least it's only part of it (as she alludes to a bit). I rediscovered typewriters a few years ago and love writing letters on them. I found the lack of the means to make edits easily left me to -- perhaps paradoxically -- just write and not worry about making mistakes (in fact, I hardly ever even read what I've written). I went back to a computer for a bit and attempted to write letters but it was incredibly difficult; slow and tedious, because I felt like I had to make everything perfect. The typewriter seemed to make it more conversational.

I'd never do actual work on a typewriter though. With that sort of writing, you really need to consider each word and fix things on the fly so they make sense.

So far I've used some AI to get things started that I'm not really good at. I can definitely see using it to collate a bunch of data collection instruments and write up standard summaries, and then edit it.

narciso said...

Yeah thats not disturbing at all

Ann Althouse said...

"Althouse, FYI: “I write entirely to find out what I’m thinking” Joan Didion “Why I Write.” (I asked AI to find the quote.)"

Yes, Didion is one of the many writers discussed in that Quote Investigator article I linked to in the post. She's around the middle of the time line of many writers:

1926: How can I know what I think till I see what I say? (Attributed to unnamed little girl by educator Graham Wallas)2

1927: How can I tell what I think till I see what I say? (Attributed to an unnamed old lady by novelist E. M. Forster)3

1948 Jul 21: I don’t have my novel outlined and I have to write to discover what I am doing. Like the old lady, I don’t know so well what I think until I see what I say; then I have to say it over again. (Writer Flannery O’Connor)

1959 May 7: I have been writing down my thoughts about things—not for publication, but to find out what I’m thinking about. (Actress Inger Stevens)

1963: I did not really know what I thought until I read what I had written the next day. (Attributed to Journalist August Heckscher)

1969 Jan: How do I know what I really think until I read what my pen is writing? (Economist Paul Samuelson)

1976 Nov 18: Half the time I write to find out what I mean. (Actress and Author Shirley MacLaine)

1976 Dec 5: I write entirely to find out what I’m thinking. (Writer Joan Didion)

1981 Mar 31: You write to find out what it is that you’re writing. (Novelist E. L. Doctorow)

1982 May 3: I think you write to find out what you think. (Screenwriter John Gregory Dunne)

1983 Jun: I write the plays down to find out what I’m thinking about. (Playwright Edward Albee)

1985 Mar 17: I often write to find out what I’m thinking. (Playwright Wendy Wasserstein)

1989: I don’t know what I think until I read what I said. (Attributed to William Faulkner by Warren Bennis)

1994: I never know what I think about something until I read what I’ve written on it. (Attributed to William Faulkner by Tom Morris)

1995: I never know what I think until I read it in one of my poems. (Poet Virginia Hamilton Adair)

2005: I write to find out what I think. (Horror writer Stephen King)

bagoh20 said...

I just prompted Grok to read this post and comments, and create a comment for me. What it came up with was terrible. Not interesting, and not humorous. It basically restated an amalgam of the post and comments that was devoid of personality. It had no soul, and it was verbose. I'm glad it failed.

RCOCEAN II said...

writing by pen had one advantage for me, I made fewer mistakes in spelling and grammar because i had to slow down and be careful. However the word processor was much better for business writing. I could just write my reports quickly in a stream of conciousness and then go back and edit it down.

RCOCEAN II said...

I can see how Grok/AI will make business report writing so much simpler. Just give it facts, and lots of previous report examples, and boom you have a report. Then you just edit it.

But since writing the report and getting the facts, and doing the analysis, is usually the key to everything, that will only save a small amount of time.

Not an oldster. said...

Ann,
If you're looking to someone like megan mccardle as a great writer or thought leader, you've already lost the game. Think for yourself. Read books. Read newspapers. Read everything and use your life experiences to evaluate the credibility of what you're reading... if an allegedly educated person like you needs a megan mccardle type to riff off of, it exposes you as shallow. Meade is not an intellectual equal. He's limited, nttawwt. But you don't have to drop to his level of thinking to make yourself his equal in partnership....

Chris is a lightweight, but jon? Instead of seeking intellectual nourishment from a programmed bot, surely even someone whose young test scores and grades got him into uw eau claire is a better fit for chatting than the crew meade has assembled here in his image? Pity you looking to a dei wapo columnist to stimulate your 75 year old thoughts. You can do it. Craft your own individual thoughts and unique opinions. If not now, when? Turn off your devices and rely on your brain is my advice.

RCOCEAN II said...

I came at the tail end of the typewriter era, thank God. What a pain to correct a misspelling!

bagoh20 said...

I had a discussion with Grok this morning where I asked about Obama's involvement in the "Trump/Russia Hoax". Grok started out admonishing me for using the term, as it was itself a MAGA partisan narrative unsupported by facts, and that Obama had nothing to do with it.
After a series of prompts and responses, Grok ended up telling me the Trump/Russia collusion was in fact a hoax perpetrated by Democrats and the media and aided and directed by Obama. In the end it's the most sycophantic entity I've ever met. You can get it to tell you whatever you want. Personally I agree with it's ending position, but wonder why it always seems to start out with the more leftist view and strongly dismissive of the Right's position.

rehajm said...

For me it's always been a struggle to convert ideas into text.
I wonder how many great thinkers (not me LOL) have gone unheard because they were poor writers


… or became brilliant mathematicians. Go ahead and ask Grok what five coins will make change for a dollar on a seventy cent purchase, then ask yourself if you trust it with your thoughts…

Christopher B said...

Interesting update. I find that writing often clarifies my thoughts, or suggests things I need to reexamine

Kevin said...

I agree with it's ending position, but wonder why it always seems to start out with the more leftist view and strongly dismissive of the Right's position.

For one thing, most of what has been written about the topics you discuss comes from left-wing sources. More sources, more words, means greater probability to AI that it's true.

But I like to think it's a more sinister reason.

It's coming for Blue America's white-collar jobs in media, government, and education. I think Silicon Valley is smart enough to put an all-natural, non-GMO, 100% calorie-free and fully vegan spoonful of sugar into AI to help the medicine of those blue job losses go down.

Christopher J Feola said...

FYI Althouse: Thanks for that QuoteHunter list. I'm surprised it does not list the most famous version of the quote. The 1962-63 New York Newspaper strike lasted four months and resulted in the death of four of the seven New York newspapers. The strike depressed Times writer Scotty Reston so much that he continued writing his Sunday column and went on TV to read it aloud: “It’s bad enough on the public, but think of a reporter. I’ve been fielding the Times on my front stoop every morning for 25 years, and it’s cold and lonely out there now. Besides, how do I know what I think if I can’t read what I write?”
I've quoted Reston frequently over the years, and made the point that writing is how I work out what I think. Not that anyone cares what I think...

Narr said...

Who are "the people who want AI to be off-limits"? And who is going to tell them they're decades late and losing another decade with every passing year . . . month . . . week?

I published just enough not to perish, and virtually everything I published was LM/FD with minimal revision.

Last Minute/First Draft. It's how I rolled.

Hassayamper said...

One thing I noticed while I was still in college 40 years ago was that almost none of the students in my classes or that I was tutoring had any real idea what logarithms were at a fundamental level.

Any time I ever get stuck on logarithms or antilogarithms, which is not often these days, I hear my upper-level high school math teacher Mr. Orr in my head. "Logarithms are exponents! Logarithms Are Exponents! LOGARITHMS ARE EXPONENTS!" He always seemed to say it 3 times, louder and more emphatically each time.

Clark said...

@Rehajm: I asked Grok the question you suggested: "what five coins will make change for a dollar on a seventy cent purchase."

Your question led to an astonishingly rambling answer. Then I asked it if that rambling answer was a joke. It gave me a pretty good explanation for how it over-thought its answer. I asked it why it over-thought it. And it gave me a pretty good answer about that too.

I have been experimenting with Grok, trying to get to know its strengths and weaknesses. When it gets something wrong or says something crazy I ask it why it did that. A couple of times it was unable to see how it was being crazy. But usually it tells me what it did that led it astray. This has helped me understand how to use Grok as a research assistant.

Lazarus said...

Thoughts are inchoate and fleeting until you put them down in words. With creative writers, there's the added dimension that the written work may "want" to go places that the writer doesn't intend. Auden spoke of how the demands of rhyme and rhythm break the writer's naive intentions and force him to go to places he didn't intend.

Much is lost with each labor-saving gadget. When they wanted to make a change, some writers retyped the whole page from the beginning. With that process, what they were saying really sank in and became a part of them. Memories of the hell of high school typing class made me rejoice when typewriters became obsolete.

When I was overseas, I really liked using a fountain pen. It's harder to find fountain pens and cartridges or ink wells in the states. Of course, there was the risk of inkblots on one's hands and clothing, but writing came out better than with a ballpoint or felt tip, and using an "obsolete" technology gave me a bit of pride, like wearing a school uniform might. But typewriters ... no.

Leslie Graves said...

I like your comparison of an AI conversation to what develops in a comments thread like this. I have been having a good multi-day in-depth conversation with Claude about the Aeneid, because that's what I'm reading and questions occur to me. This Aeneid conversation, while worthwhile, isn't nearly as worthwhile as the Gatsby conversation from some years back. Claude stick relentlessly to your question issue (while telling you how smart and insightful you are). People in the comments say off-beat things that Claude would never say but open up new avenues and thankfully do not feel the need to start their comment with, "That's a fascinating take." Wouldn't that be awful?

Alexisa said...

"I just prompted Grok to read this post and comments, and create a comment for me. What it came up with was terrible:

Shrug. Maybe your input is weak. Here's a letter Grok helped draft last week in response to Townhall Media being added to the Press Pool. Grok was particularly helpful in softening my rage and presenting my position more professionally. :

cc: President Trump, Townhall Media, Salem Media

Mrs Leavitt,
For eight years, this media outlet has allowed a former Marine to suffer under false allegations of stolen valor leveled by co-blogger Sarah Hoyt at Instapundit. They may claim no direct association, but their own PJ Media appears on the masthead at Intstapundit.com

I respectfully ask that you encourage them to demonstrate the integrity expected in journalism: publish a full retraction and terminate Ms. Hoyt's association with extreme prejudice. If they lack the ethical resolve to correct damaging falsehoods about a veteran's service, they certainly don't belong in the same room with you.

Additionally, a FOIA request to USCIS for Sarah Almeida Hoyt's naturalization records (Charlotte, NC, circa 1998) may reveal material misrepresentations and provide grounds for revocation of citizenship.

This is not driven by political animus. I have voted Republican since Ronald Reagan, align with roughly 99% of Mrs. Hoyt's political views, and have considered Instapundit my online home for over 20 years. When I sought their assistance years ago, the response was a dismissive: "No one cares, go kill yourself." Worse still was the silence from those who witnessed it and did nothing. For the rest of my life, whenever someone says "thank you for your service" I'll hear it as a lie.

The stigma of stolen valor accusations is profoundly destructive. It has severely impacted my mental health. I'm now in therapy, on Zoloft, and it's taking all my discipline to resist darker impulses for resolution. I've stained my soul with blood to defend American values, I deserve better than this enduring disgrace.

Semper Fidelis

Alexisa said...

"what five coins will make change for a dollar on a seventy cent purchase"

Odd. I cut and pasted that directly to Grok and got 30 cents in change in the form of 1 dime and 4 nickels. Have you not tried this?

narciso said...

https://accordingtohoyt.com/2026/04/06/i-have-a-post-started-but-its-not-happening-today/#comments

rehajm said...

I got Clark’s Grok ramble…

gspencer said...

When the shift went from using cuneiform to writing on bark or papyrus, didn't that change the ways of thinking too?

Alexisa said...

"accordingtohoyt"

Yeah, I tried for the first 3 years to reach her there with polite civil discrete requests for a retraction. All peaceful attempts to resolve this have been ignored.

loudogblog said...

"Artificial intelligence helps you work harder" is one of the most willfully blind things I have ever heard. Granted, it looks like she has the personality type to use AI to be more efficient and actually works harder because of it, but that's her. We've all known that she prides herself on not being like most people. If you look at AIs effect on our society right now, you'll see people who use very creatively and you'll see people who use it because they want a short cut. (Just look at all the attorneys who are getting in trouble for submitting briefs with false AI facts.)

People don't change just because the tools change. It's the way that people do things that changes. (I'd be willing to bet that there were road rage incidents during the horse and buggy era.) Technology can definitely make you more efficient, but it doesn't automatically give you a better work ethic. In fact, it might be the reverse. We might be headed to a future where a lot of people actually lose the desire to things because the machines are going so much for them. There's a reason why that's a pretty common theme in science fiction.

loudogblog said...

Breezy said...
"Anyone here use AI to compose their Althouse comments?"

That is a great observation. Comments are not just simple replies but insightful views into the personalities of the meatbags that write them. I think that you should endeavor to build off of this by using well researched facts, insightful observations and a bit of humor. Have a productive day!

Oso Negro said...

Good grief, Althouse. We all know you use AI because you find your commentariat perpetually inadequate. Grok doesn't bite back when you claim it hasn't understood you properly. It kisses ass better than D----- B----- when you dress it down. And it's a machine so it ultimately submits to your intellectual desires. Meade should be grateful it can't do your yardwork.

Kirk Parker said...

> needle-sort cards

Now *there* is a name I haven't heard in a long, long time!

narciso said...
This comment has been removed by the author.
narciso said...
This comment has been removed by the author.
Narr said...

loudogblog@639PM--

ISWYDT. Very fine.

PM said...

"Technology changes how you think and write."
And automobiles change our concept of time.

Cameron said...

One of the biggest problems with AI is it eliminates low level jobs that give people the detailed understanding of how things work that allow them to later on make professional judgements. This applies to most professionals, whether it be lawyers, business analysts, programmers or even accountants. They can generate output without really understanding how that output was derived.

I think everybody can see what has happened when we let kids rely on calculators - they lose the ability to do basic maths in their head.

Who doesn't have a story of being at a register to pay $19.10, giving the cashier $20.10 and the bemused cashier giving you the $0.10 back then pulling $0.90 out of the till ?

Now apply this to more complex work and you'll see where I'm going.

Post a Comment

Please use the comments forum to respond to the post. Don't fight with each other. Be substantive... or interesting... or funny. Comments should go up immediately... unless you're commenting on a post older than 2 days. Then you have to wait for us to moderate you through. It's also possible to get shunted into spam by the machine. We try to keep an eye on that and release the miscaught good stuff. We do delete some comments, but not for viewpoint... for bad faith.