This was a really interesting incisive essay, Mark, and I will need to come back to it to follow up the references and lines of thinking. Just a few points I'd like to make:
1. I don't believe in shitty first drafts because, perhaps egotistically, I don't regard my first drafts as shitty. They are initial steps. You wouldn't say to a toddler learning to walk "Wow, that was a shitty first effort!". The issue becomes, does this draft reflect what I want to say, or have the effect I want it to have. And it's the result of a thinking process.
2. So it seems to me that when we use ChatGPT or embedded AI to create the first draft, we become critics rather than writers, at least in the first instance. But we have bypassed the initial part of the process.
3. However, it may be useful to give a head start. I've been experimenting with ChatGPT again and embedded AI recently (Squarespace, which I use for two of my websites, has introduced an embedded AI to write text for you in response to a prompt.) I used Chatgpt to generate personas for a target readership, and the students on my course (about blogging) and I agreed that it would give someone a few useful ideas and starting points.
You must be among the blessed perfect-just-out-of-the-pen writers, and I’m envious. I squeezed out a first draft of my Thursday post and barely had the gumption to remain in the room to begin editing! (It’s on Barbie, by the way. If it makes it to post-able shape.)
I guess the Anne Lamont “shitty draft” moniker could be a bit heavy-handed, especially given the toddler metaphor you use. But I think it also has its uses for writing mortals: You need to get it out … the message, I mean. And you need to clean it up. The colorful metaphor implies, even demands, a writerly process. And, as I mentioned in the post, the focus of the labor shifts. It aims at something beyond utterance — the gruntings of getting the ideas together — and seeks greater clarity. Perfection even, though I’ve never gotten there!
As you said, we take on a bit of the critic as we move through the processes of writing. I’d say the process itself is a kind of criticism, maybe even more formalized and structured that your typical “critic” mode. It’s hard, too, and sometimes a wise editor can nudge us. Or at least help us defend our words and our thought.
I agree with your last statement, too. I think ChatGPT can be useful. But I think the real question is how we make ChatGPT useful. I think that ChatGPT could be useful for certain kinds of writing, especially for writing that values or requires production — think deadline-driven journalism, certain forms of marketing activity. (Of course these are not forms that ChatGPT or any LLM can just do “on their own.”) But there are other forms of writing that place less value on efficient production: fiction, history, forms that rely of creative and insightful argument. Forms that seek qualities that, for now at least, seem to be situated in the human.
David Humphrey’s post that I linked to in the “Catch up links” points out the difference between CS students using AI assists when they’re noobs and when they’re more experienced. The same resource means something different depending on where the coder is in his or her development. (Humphrey, David. “CheatGPT.” Bread & Circuits, February 19, 2023. blog.humphd.org/cheatgpt.)
I’d say the same dynamic works for writers.
Thanks for reading and thinking about this post, Terry. It’s always good to see you comment!
Thanks for this Mark. I particularly liked your point about the importance of the first draft. This is going to be the first thing to go for folks using AI to “help them write” and I have my concerns. It’s obvious in the design of Google’s tools that they’re targeting the blank page because of how they integrated the tool as a button right on the page when you create a new doc. The blank page is often the most challenging part of the process but I’m not sure that we are fully appreciating the value of those early, tough steps.
Interested in following the experiments in your classes and seeing what you learn!
I really enjoyed this, Mark. I’ve been trying to put to words what bothers me, as folks try to convince me of the “time-savings”. But it’s the process of writing, and the role in extending and working through our thoughts, that is lost with AI. Creative thought isn’t a process to be Kaizen’d to death.
The "productivity" crowd does miss a point, but I think that part of the issue is the larger framework. Writers today need to up the output in order to stay afloat. I do wonder if that's always been the case, but I think it's particularly so today. AI and GPT and such add to the competition, but the race has been inappropriately cast as a matter of word-counts.
If writing would be a mode of thought more generally used, we'd have a fuller appreciation of its powers and its necessity for us humans. Reduced to a cog in a professional publishing machine, writing gets a rep as a part of economic engine of publishing. At worst, a "cost."
Easy for me to say, I guess, because I'm not trying to feed a family with literary output. But we all need to push for the greater good of thinking and writing!
To have come to a stirring conclusion and/or a definitive path would have really been audacious ... I don’t see how you can do anything but recognize we’re all still floating in the figuring-it-out stage.
And I bet we'll be floating there for a long, long time!
Using GPT feels different from, say, discussing knotty topics in class. The knotty topics are difficult and also finally indefinite. GPT is like a drug in some ways. At least it feels like that to me. Using it could rewire the "pleasures" of writing enough that you wouldn't want to write without it, even though you know it's "not good for you." Drug addicts don't quit using because someone convinces them that the drug effects are unpleasant.
Found a couple of interesting articles that I'll be posting in my next "catch-up" which I think will surface in a couple weeks.
This was a really interesting incisive essay, Mark, and I will need to come back to it to follow up the references and lines of thinking. Just a few points I'd like to make:
1. I don't believe in shitty first drafts because, perhaps egotistically, I don't regard my first drafts as shitty. They are initial steps. You wouldn't say to a toddler learning to walk "Wow, that was a shitty first effort!". The issue becomes, does this draft reflect what I want to say, or have the effect I want it to have. And it's the result of a thinking process.
2. So it seems to me that when we use ChatGPT or embedded AI to create the first draft, we become critics rather than writers, at least in the first instance. But we have bypassed the initial part of the process.
3. However, it may be useful to give a head start. I've been experimenting with ChatGPT again and embedded AI recently (Squarespace, which I use for two of my websites, has introduced an embedded AI to write text for you in response to a prompt.) I used Chatgpt to generate personas for a target readership, and the students on my course (about blogging) and I agreed that it would give someone a few useful ideas and starting points.
You must be among the blessed perfect-just-out-of-the-pen writers, and I’m envious. I squeezed out a first draft of my Thursday post and barely had the gumption to remain in the room to begin editing! (It’s on Barbie, by the way. If it makes it to post-able shape.)
I guess the Anne Lamont “shitty draft” moniker could be a bit heavy-handed, especially given the toddler metaphor you use. But I think it also has its uses for writing mortals: You need to get it out … the message, I mean. And you need to clean it up. The colorful metaphor implies, even demands, a writerly process. And, as I mentioned in the post, the focus of the labor shifts. It aims at something beyond utterance — the gruntings of getting the ideas together — and seeks greater clarity. Perfection even, though I’ve never gotten there!
As you said, we take on a bit of the critic as we move through the processes of writing. I’d say the process itself is a kind of criticism, maybe even more formalized and structured that your typical “critic” mode. It’s hard, too, and sometimes a wise editor can nudge us. Or at least help us defend our words and our thought.
I agree with your last statement, too. I think ChatGPT can be useful. But I think the real question is how we make ChatGPT useful. I think that ChatGPT could be useful for certain kinds of writing, especially for writing that values or requires production — think deadline-driven journalism, certain forms of marketing activity. (Of course these are not forms that ChatGPT or any LLM can just do “on their own.”) But there are other forms of writing that place less value on efficient production: fiction, history, forms that rely of creative and insightful argument. Forms that seek qualities that, for now at least, seem to be situated in the human.
David Humphrey’s post that I linked to in the “Catch up links” points out the difference between CS students using AI assists when they’re noobs and when they’re more experienced. The same resource means something different depending on where the coder is in his or her development. (Humphrey, David. “CheatGPT.” Bread & Circuits, February 19, 2023. blog.humphd.org/cheatgpt.)
I’d say the same dynamic works for writers.
Thanks for reading and thinking about this post, Terry. It’s always good to see you comment!
Thanks for this Mark. I particularly liked your point about the importance of the first draft. This is going to be the first thing to go for folks using AI to “help them write” and I have my concerns. It’s obvious in the design of Google’s tools that they’re targeting the blank page because of how they integrated the tool as a button right on the page when you create a new doc. The blank page is often the most challenging part of the process but I’m not sure that we are fully appreciating the value of those early, tough steps.
Interested in following the experiments in your classes and seeing what you learn!
There is something to that terror of the blank page, isn't there! Maybe the terror is something deeper than mere discomfort and annoyance.
I really enjoyed this, Mark. I’ve been trying to put to words what bothers me, as folks try to convince me of the “time-savings”. But it’s the process of writing, and the role in extending and working through our thoughts, that is lost with AI. Creative thought isn’t a process to be Kaizen’d to death.
The "productivity" crowd does miss a point, but I think that part of the issue is the larger framework. Writers today need to up the output in order to stay afloat. I do wonder if that's always been the case, but I think it's particularly so today. AI and GPT and such add to the competition, but the race has been inappropriately cast as a matter of word-counts.
If writing would be a mode of thought more generally used, we'd have a fuller appreciation of its powers and its necessity for us humans. Reduced to a cog in a professional publishing machine, writing gets a rep as a part of economic engine of publishing. At worst, a "cost."
Easy for me to say, I guess, because I'm not trying to feed a family with literary output. But we all need to push for the greater good of thinking and writing!
To have come to a stirring conclusion and/or a definitive path would have really been audacious ... I don’t see how you can do anything but recognize we’re all still floating in the figuring-it-out stage.
And I bet we'll be floating there for a long, long time!
Using GPT feels different from, say, discussing knotty topics in class. The knotty topics are difficult and also finally indefinite. GPT is like a drug in some ways. At least it feels like that to me. Using it could rewire the "pleasures" of writing enough that you wouldn't want to write without it, even though you know it's "not good for you." Drug addicts don't quit using because someone convinces them that the drug effects are unpleasant.
Found a couple of interesting articles that I'll be posting in my next "catch-up" which I think will surface in a couple weeks.