Paul: Two categories of AI usage
A working theory: There are two broad categories of generative AI use: using it to avoid work, and using it to work harder.
I've been setting up this job hunt database and the workflow for using it. One of the things I'm specifically doing is creating a space in my Claude account with my resume, several pieces of sample writing, and a description of a bunch of my personal projects. I'm then pasting job descriptions in and saying "write my cover letter." This is—despite what it's in service of—an example of avoiding work. It would be more honest and more ideal of me to write cover letters individually, but doing that work slows the process down so much that I can live with the tradeoff. Furthermore, I have a kind of fundamental disdain for everything about the job hunt process before the interview, and this reflects that disdain.
In this way, it's not at all different from students using AI to write essays or complete class assignments. They have a fundamental disdain for the work, and I mean that descriptively—they don't deign ("dis-deign?" omg) to do it; they would prefer to avoid it.
Meanwhile, when I use AI to help me write some code or translate a passage or plan a project, it's the opposite. There's something I want to see created badly enough that I'll deploy any and all tools available to me to make it happen faster. If this involves interminable back-and-forths with a frustratingly inconsistent guy-in-the-computer, so be it. I'll do that work too, in addition to all the other work involved in completing whatever the task is.
I think a lot of people don't understand that the work that results from these two categories of usage is vastly divergent in its quality, because the first category is fueled by apathy, and the second by interest.