Almost everything I’ve seen about using “ai” to help you write essays straight up ignores the aspect of invention with writing. Simply put, writing helps you clarify what you think and leads to new insights. I remain unconvinced that generative tools are making that process “better.”
“A bottleneck of writing is sentence generation—getting ideas into sentences…That is a big task. That part is really costly in terms of cognitive load.”
For day-to-day stuff, the way *I* say it isn't the *only* way to say it and may not even be the *best* way to say it, but tell you what, when my predictive text starts guessing correctly what I'm going to text on the regular, then let's talk.
so you can see the numbers you get over on threads. while not being on threads. if you search threads for your name, posts come up from people expressing they miss your tweets. SWAK!
Absolutely this. I just finished my final assignment in my second degree and going through the actual process of writing is always where my ‘aha! That’s my argument’ angle springs from.
Exactly! Writing is coagulated thinking. When you can't think, you won't be innovative, creative, anything else than recreating what was already there. I gave a talk some months ago on writing in and as sociological methodology in the advent of AI.
Which is the selling point and major flaw. It is not just a pencil. It is replacing, imho, significant parts of every type of writing process. Even the doodling while in the meeting or on the phone is something productive and creative, if you really want to look at it. AI is not easing a burden here
Yes like, when I write using a pencil I do it slower than typing but also not so careful like when using a pen, which I can’t erase, so I instinctively write slower, which makes my brain work differently… it’s had to explain but it’s why journaling on paper works for me and on a phone it doesn’t
Writing as content versus writing as communication / contemplation (or indeed almost anything else). Content exists only to be further replicated and mined for (training) data... forever. That's all that matters to platform capitalism.
To add to this Ive discovered that when doing creative work I need to write by hand. Its both faster and I find myself being more deliberate or asking better questions about what Im trying to say. I wouldnt mind an AI that would type it for me tho.
same. Do you write on paper or on screens? Just curious. I found both useful but for some things I prefer paper, e.g. journals and memos during research and field work. Not for speed only or for privacy and security, but also for the embodied experience of work there
I agree. But drawing is also something different with a tool in your hand and the imperfections of the lines. I get told I would draw good, even though I never thought so. But I still like the experience of doing it. Drawing, painting or doodling
Yeah it’s totally different. I have been drawing on paper more lately and I really enjoy it. But Im comfortable enough with digital drawing so that I can get what I need. I imagine if I forced myself to type more Id get used to it but idk
New from me, first of a series:
Life on the avoidant web: How our desire for a frictionless life online may be making us miserable.
open.substack.com/pub/artifici...
You improve your writing by writing a lot, not by having someone else do the hard part. If you’re 50 and can’t write, sure, do what you gotta, but if you’re a student, learn to write a good sentence, that’s what you’re there for.
I think this could vary based on genre of text, the form of tool employed, and how thoughtful the use of the tool is. I have concerns about tools like the one you re-posted too, and especially their emphasis on "efficiency." There could be some gains for *clarity* for some writers, though.
I don’t disagree on the clarity part. I can’t say I’ve done a comprehensive audit of these kinds of tools, but I’ve yet to see one that even engages with the connection btwn writing and thinking. W/o that foregrounded, I have severe doubts about what good it’s doing.
Yeah, current tools, to paraphrase Shania Twain, don't impress me much. Still, as a teacher who is aware students find these tools attractive, one thing I'm thinking about is how to encourage critical engagement with the tool that might minimize the damage to critical thought...
And part of that, frankly, has been more explicitly teaching "writing as thinking" which students seem only vaguely aware of when they reach college writing, even at the more advanced levels!
do you think there's any worth in straight up inviting students to get the tool to 'write' some text and then critique it as a device to get them to think critically about their own writing... 🤔
I think it's a place to start! People have had success with it. I would recommend thinking carefully about the writing task you have them engage with for such an exercise. Make it specific and concrete. Guide their critique. They may be impressed by the apparent "smartness." Get them past that.
A lack of clarity in writing often reflects a lack of clarity in thought. Having LLMs "improve" your writing doesn't help with this, and that makes it seem like a very poor teaching tool.
It seems like just one more shortcut that fails students (and society at large). It provides the illusion...
of clarity, but it doesn't actually teach students anything. You can't give feedback and require re-writes.
Teaching students how to write better prompts isn't a useful skill either - we might be stuck with LLMs in our life, but they won't be the precise things we have now.
Yes. I’ve been thinking a lot about the parallels between “ai” boosters hostility towards art and their hostility towards writing—or anything creative for that matter.
I wonder if this all originated in the believe we ought to master every basic human skill there is. You can't just draw, you must head for masterpieces. You can't just dance, you must dance perfectly. You can't just make music, you must master it. You can't just write, it has to be worth a Pulitzer
Yes. It’s a weird conceit. By definition you become a writer, dancer, artist when you do those things. Doing them better requires effort and time. The tech promise to disrupt them is an illusion.
I am thinking of researching that. Maybe tech bros did not allow or did not want to have themselves spending time on these things, that they consider them to be replaced by automation. I'm thinking of looking at CVs and the hobbies section or their high school classes. Surely it's more complex
“Improved writing” is an important outcome in itself, but it’s subordinate to the goal “improved thinking,” which writing produces/is produced by. (My HS English teacher had a large sign on her whiteboard that said WRITING IS THINKING and she motioned to it often).
How do I know what I think, until I see what I say?
— E. M. Forster
We write to give form and substance to our nebulous, emergent, and ill-formed thoughts. The process of writing is requires thinking about what we are thinking. The core problem is most people aren't thinking to start with.
Absolutely. I even feel this way about writing code. It’s actually a good thing to have my own ideas and not be constantly interrupted by fancy, bad autocorrect. If I want copypasta, plenty of that on stack overflow.
True as well. Coding is hand craft.
I once met someone at London Imperial that had a habit of coding late at night while drinking a bottle of wine. It was a way to bring down ideas that work but are not helpful at first sight but interesting and creative to some extent that you wouldn't have thought
Coding is like being involved in a giant group novel where you must keep continuity 100%. You’ll get asked to edit a specific part of chapter 85, without having any idea what that chapter is supposed to be about, and have to keep continuity with what the characters did back in chapters 5, 28, and 3.
Basic thinking skills can lead to critical thinking skills. Critical thinking skills can lead to questioning the system. Questioning the system can lead to reduction in productivity. Can't have that.
I was curious if the researchers had grant money (it looks like Ishizaki was originally in the Design department, then moved over to English). The tool they previously developed is being tested/used by . . the RAND corporation, and ETS: www.cmu.edu/dietrich/eng...