I've come to the conclusion that AI presents us with an updated form of rhetorical practice and a new way to teach composition. We need to learn how to communicate effectively with "robots" so that we can accelerate our writing output while making it more digitally sophisticated -- automated and enhanced by programming languages or at least by markup and formatting languages (HTML / CSS).
AI assistants are only as effective as the prompts they are given. You need to write (or say) precisely what you want. You have to eliminate ambiguity, unnecessary complexity, unintended meanings, and irrelevant details. You have to articulate your assumptions and realize that the machine has no implicit knowledge. It does not and cannot know what you are thinking. You need to provide a context, an intended audience, and you need to specify how you want the response formatted. And even then you may need to iterate.
Technical writers are accustomed to these requirements because effective communication has always been based on them. The only difference now is that the initial (not the intended) audience is a robot. You get instantaneous feedback on how well you crafted your prompt and therefore it's an efficient place to practice technical writing. It's also an effective way to clarify your thinking. If you don't quite know what you want, what you get won't be what you are looking for: revise and resubmit. Rethinking and refining instructions is a great rhetorical exercise. The better you get at it, the more you will be able to produce and the more digitally sophisticated your output will become. You will also become a better writer in the traditional sense because you can practice with instantaneous feedback.
I have some of my own AI interactions embedded in this website. If you aren't already experimenting with AI-assisted composition, I encourage you to.
Here is a concrete example of an AI-assist on this website. When you go to the Lectures section, the current week's work is open. That happens by design. From week 1 to week 9, the code is very simple: if week = 1, show 1 (essentially). But after week 11 I need, if week = 11 or week = 12 or week = 13 and so on. Verbose and inefficient. So I asked ChatGPT: "I need help with some php code. How do I write this if statement: if variable is > or = 11 but < 15, then ... " As I was writing that request I started remembering the syntax, but I let ChatGPT write it for me: "To write an if statement in PHP that checks if a variable is greater than or equal to 10 and less than or equal to 15, you can use the following syntax: if ($variable >= 11 && $variable < 15) { // Your code here }." Someone more accustomed to programming wouldn't have needed that assist. And someone who knows less than I do wouldn't likely have known where to start. But I knew what I wanted and how to articulate it precisely. My memory was activated by articulating exactly what I needed. But since it was already writing the code for me, I didn't have to write it myself. Memory enhancement is another way AI can assist us rhetorically.