Hey Nick - good stuff and I'm going to read it in more detail. How much was generated by AI? The hierarchical structure and the word "delve" from the introduction is a dead giveaway. I have no problem with reading AI generated work but when it is so obvious it takes me out of the piece and all I can think of is how much of this was generated by the prompt. Just curious.
I developed all the slide content, decision tree, and roadmap. I used AI to translate from slide format to a more expository format. I hope that eases you mind some. I just find AI's ability to infuse a little storytelling into this slide material helpful to lift the content a bit for readers.
I get it. I do stuff like this all the time. I'm new to the newsletter thing and wondered about the protocol for using AI generated stuff in posts. This is what we are all figuring out.
You set your own parameters and feel it out with your audience. For a more programmatic document like this, AI serves that purpose of a good assistant. For more personal criticism and analysis, my voice and writing are primary. We really are figuring this out together right now.
Hi Steve—this is a tough problem. No matter how careful we are to “disguise” uses of AI, something always seems to give it away. Like you, I’m attuned to “delving” and over-adjectiving and broad pronoun references. The problem is we share language with the bot. My solution? Look at the words as the authors words, chosen by the author. If the piece as a whole is credible, interesting, and speaks to me, it doesn’t matter that the word “delve” shows up. The author selected it from a range of options and fearlessly used it because it works for the purpose. It’s going to take a while but I think the solution will be natural. A truly serious problem is passing along hallucinations whether bot words are swapped out or not.
When it comes to students, this is the crux of the issue. It's just so interesting to me that we are willing to pass off AI generated text without attribution (though once you start editing AI stuff and incorporating your own sentences, interleaving various configurations, the whole thing becomes muddled and then, as you say, as long is the piece as a whole is useful and communicates clearly what it is supposed to, I have no issue). It's unclear to me how much of this guide in this post is pure AI but it felt like most of it which takes me out of the piece the same way when I read an article and it's clear after one paragraph the entire thing was generated by AI. This is absolutely what we don't want kids to be doing at this particular moment in time and most do not have the skills or ability to shape and rework an AI piece into something more personal that retains their voice. I am all for using AI to help revise and edit something I've written and generate stuff for my own consumption and jumpstart organizing an assignment. There are so many different kinds of writing for different audiences. It's a slippery slope.
I personally think we need to be teaching students to write productively with AI and to think about the rhetorical complexities of that writing situation. This is precisely the work Terry and I are doing in our AI Theory and Writing class. This new writing situation presents an opportunities to re-present many of our core writing skills and competencies in a context that students naturally find engaging and useful.
Thanks, Nick. If I read your bio correctly, you teach at the university level. The issues are a little different in secondary schools but I tend to agree there needs to be some sort of recognition where AI fits into the writing process. I work at a school where the Department Heads of the Humanities (History and English) both are both adamant that AI has no place in the writing process which I do not think is sustainable long term. I am very interested in your work as I am grappling with this issue right now with respect to our yearly research paper which is coming up. The release of the Deep Research models present both opportunities and challenges with respect to the learning process. Would love to see the syllabus for the AI Theory and Writing class.
Good conversation. I straddle upper-level secondary and introductory university writing instruction. Yes, the no AI or the highway policy only pushes the use underground. In my secondary classes, I prefer a toggle on / toggle off model that aligns with the developmental readiness of students. 9th graders can only handle very curated experience with AI. 12th graders can be given a little more room to experiment. Ideally, this new curricular focus of ethical AI will lean down into middle and lower school where students will work in more simulated environments with pre-printed prompting cycles---evaluating AI generated texts for the things you are already attuned to in your own reading practice. I will be happy to share our syllabus: https://docs.google.com/document/d/10ZfBZhmYgow-iPY1D15xpt9VMS4PDm6biHokfFPHeaQ/edit?usp=sharing
The syllabus is very barebones. The experience lies in student interactions with mentor prompts. These cultivate amazing extended writing experiences for students. Inside these workflows, the premium is on human-originated text as AI primarily plays the role of interlocutor, not text generated. This is going to the be the way of the future (perhaps), as student learn to create immersive experiences like this to shepherd themselves through specific writing challenges. But a lot of new skills and competencies need to be taught along the way. In Terry's words, most of our current writing assignments offer students a pre-processed semantic vision that collapses the complexity of the writing situation, and does much of the thinking for the students. In an AI-infused writing classroom, students will have to assess each writing situation closely, decide where AI can help and where it can hurt, and generate inquiry-based immersive work cycles that don't steal away their thinking and experience.
This is a lofty goal, and it is no wonder so many writing teachers just want to ban AI out of existence, rather than wade into these complexities.
Hey Nick - good stuff and I'm going to read it in more detail. How much was generated by AI? The hierarchical structure and the word "delve" from the introduction is a dead giveaway. I have no problem with reading AI generated work but when it is so obvious it takes me out of the piece and all I can think of is how much of this was generated by the prompt. Just curious.
I developed all the slide content, decision tree, and roadmap. I used AI to translate from slide format to a more expository format. I hope that eases you mind some. I just find AI's ability to infuse a little storytelling into this slide material helpful to lift the content a bit for readers.
I get it. I do stuff like this all the time. I'm new to the newsletter thing and wondered about the protocol for using AI generated stuff in posts. This is what we are all figuring out.
You set your own parameters and feel it out with your audience. For a more programmatic document like this, AI serves that purpose of a good assistant. For more personal criticism and analysis, my voice and writing are primary. We really are figuring this out together right now.
Hi Steve—this is a tough problem. No matter how careful we are to “disguise” uses of AI, something always seems to give it away. Like you, I’m attuned to “delving” and over-adjectiving and broad pronoun references. The problem is we share language with the bot. My solution? Look at the words as the authors words, chosen by the author. If the piece as a whole is credible, interesting, and speaks to me, it doesn’t matter that the word “delve” shows up. The author selected it from a range of options and fearlessly used it because it works for the purpose. It’s going to take a while but I think the solution will be natural. A truly serious problem is passing along hallucinations whether bot words are swapped out or not.
When it comes to students, this is the crux of the issue. It's just so interesting to me that we are willing to pass off AI generated text without attribution (though once you start editing AI stuff and incorporating your own sentences, interleaving various configurations, the whole thing becomes muddled and then, as you say, as long is the piece as a whole is useful and communicates clearly what it is supposed to, I have no issue). It's unclear to me how much of this guide in this post is pure AI but it felt like most of it which takes me out of the piece the same way when I read an article and it's clear after one paragraph the entire thing was generated by AI. This is absolutely what we don't want kids to be doing at this particular moment in time and most do not have the skills or ability to shape and rework an AI piece into something more personal that retains their voice. I am all for using AI to help revise and edit something I've written and generate stuff for my own consumption and jumpstart organizing an assignment. There are so many different kinds of writing for different audiences. It's a slippery slope.
I personally think we need to be teaching students to write productively with AI and to think about the rhetorical complexities of that writing situation. This is precisely the work Terry and I are doing in our AI Theory and Writing class. This new writing situation presents an opportunities to re-present many of our core writing skills and competencies in a context that students naturally find engaging and useful.
Thanks, Nick. If I read your bio correctly, you teach at the university level. The issues are a little different in secondary schools but I tend to agree there needs to be some sort of recognition where AI fits into the writing process. I work at a school where the Department Heads of the Humanities (History and English) both are both adamant that AI has no place in the writing process which I do not think is sustainable long term. I am very interested in your work as I am grappling with this issue right now with respect to our yearly research paper which is coming up. The release of the Deep Research models present both opportunities and challenges with respect to the learning process. Would love to see the syllabus for the AI Theory and Writing class.
Good conversation. I straddle upper-level secondary and introductory university writing instruction. Yes, the no AI or the highway policy only pushes the use underground. In my secondary classes, I prefer a toggle on / toggle off model that aligns with the developmental readiness of students. 9th graders can only handle very curated experience with AI. 12th graders can be given a little more room to experiment. Ideally, this new curricular focus of ethical AI will lean down into middle and lower school where students will work in more simulated environments with pre-printed prompting cycles---evaluating AI generated texts for the things you are already attuned to in your own reading practice. I will be happy to share our syllabus: https://docs.google.com/document/d/10ZfBZhmYgow-iPY1D15xpt9VMS4PDm6biHokfFPHeaQ/edit?usp=sharing
The syllabus is very barebones. The experience lies in student interactions with mentor prompts. These cultivate amazing extended writing experiences for students. Inside these workflows, the premium is on human-originated text as AI primarily plays the role of interlocutor, not text generated. This is going to the be the way of the future (perhaps), as student learn to create immersive experiences like this to shepherd themselves through specific writing challenges. But a lot of new skills and competencies need to be taught along the way. In Terry's words, most of our current writing assignments offer students a pre-processed semantic vision that collapses the complexity of the writing situation, and does much of the thinking for the students. In an AI-infused writing classroom, students will have to assess each writing situation closely, decide where AI can help and where it can hurt, and generate inquiry-based immersive work cycles that don't steal away their thinking and experience.
This is a lofty goal, and it is no wonder so many writing teachers just want to ban AI out of existence, rather than wade into these complexities.
Amazing piece of work Nick