This is my particular hobby horse, but we've been explicitly training student writers using templated rhetoric (e.g. the five-paragraph essay) for a couple of generations, to it makes sense that the out-of-school versions seem just peachy. What frustrated me about my students starting 20 years or so ago was not their writing skills, so much as their attitudes about writing, that they were there to follow a set of rules in the service of satisfying a rubric for a grade.
That distortion of writing into a writing-related simulation was very bad for their ability to think rhetorically, which is the central problem I started to attack as I evolved my approaches to teaching writing. When I would go on my rants about the pernicious effects of the 5PE I had legions of people tell me that it was necessary to give students "training wheels," but we weren't training them to write. We were conditioning them to accept bad AI writing all along.
I am struck by how rapidly awareness is growing about how easy it is to spot AI generated text (Mollick’s latest) and by the fact that AI slop writing is only becoming more pervasive by the day.
Does this mean we're heading toward AI “templating” writing the way platforms eventually templated web design? Maybe. Another possibility is that many of the people who are now generating this slop but who can write reasonably well will soon notice what’s happening and stop leaning on AI for so much. We may just be in a period of mass experimentation.
Excellent post, as usual. So, it made me think about the popular writing textbook They Say/I Say, which I have used many times in my own comp classes. I find that students like it and that we can smoothly move from its template-based approach to learning to write better. I think that's because the book ONLY provides empty templates for parts of sentences. The writer still has to come up with the content, and with what you're saying in your post is that connection a human makes between a word and actual reality. In other words, we can teach writing using a template approach that demystifies how academic language is structured, but what LLMs do is generate the template AND the purported content without rhetorical awareness or the option for the user to understand genre and form. And again, an LLM has a "user," someone who's consuming the output....
This is my particular hobby horse, but we've been explicitly training student writers using templated rhetoric (e.g. the five-paragraph essay) for a couple of generations, to it makes sense that the out-of-school versions seem just peachy. What frustrated me about my students starting 20 years or so ago was not their writing skills, so much as their attitudes about writing, that they were there to follow a set of rules in the service of satisfying a rubric for a grade.
That distortion of writing into a writing-related simulation was very bad for their ability to think rhetorically, which is the central problem I started to attack as I evolved my approaches to teaching writing. When I would go on my rants about the pernicious effects of the 5PE I had legions of people tell me that it was necessary to give students "training wheels," but we weren't training them to write. We were conditioning them to accept bad AI writing all along.
I am struck by how rapidly awareness is growing about how easy it is to spot AI generated text (Mollick’s latest) and by the fact that AI slop writing is only becoming more pervasive by the day.
Does this mean we're heading toward AI “templating” writing the way platforms eventually templated web design? Maybe. Another possibility is that many of the people who are now generating this slop but who can write reasonably well will soon notice what’s happening and stop leaning on AI for so much. We may just be in a period of mass experimentation.
Excellent post, as usual. So, it made me think about the popular writing textbook They Say/I Say, which I have used many times in my own comp classes. I find that students like it and that we can smoothly move from its template-based approach to learning to write better. I think that's because the book ONLY provides empty templates for parts of sentences. The writer still has to come up with the content, and with what you're saying in your post is that connection a human makes between a word and actual reality. In other words, we can teach writing using a template approach that demystifies how academic language is structured, but what LLMs do is generate the template AND the purported content without rhetorical awareness or the option for the user to understand genre and form. And again, an LLM has a "user," someone who's consuming the output....