My first response to this bit of moral panic is, "Yeah, good luck with that." Readers may recall that I automated the LinkedIn version of my newsletter using a simple Python script. There's no difference between automated Stephen and real Stephen. I'm actually using Chrome to make the requests. If instead of reading from an RSS file my script got its input from a large language model (LLM) using, say, the ChatGPT API the way I do in CList, I would have an automated agentic AI browser. I could do things like "move through an LMS to locate assignments, complete quizzes, and submit results" or, with access credentials, "impersonate instructors by grading student work and posting feedback." These are simple off-the-shelf technologies anybody can mis and match to automate whatever they want. Good luck banning them.
Today: Total: [Direct link] [Share]
