I don't entirely mind the idea of a LLM programming tool running while I work, so long as it is completely powerless to do anything at all, which includes forbidding it from making suggestions. I am more productive when there is a captive entity watching in horror as I slowly and carefully place every block into the square hole. My rubber duck also works fine for this, however.
@aeva See, but one burns fossil fuels and VC dollars, while the other is a friend. So I think I know which one I want 😛
@devmcclu burns fossil fuel, wastes VC dollars, AND if we're lucky has the potential to take down the west coast tech industry once and for all when it finally pops.

@aeva if I'm ever forced to use an LLM in my work I suspect I'd do something like this.

Have a script throw prompts at it in the background while I actively discard its outputs.

@beeoproblem there's all sorts of normal, prosocial, productive things you can do with LLMs. for example, you can use revision control history to find the names of the people who have wronged you, and then tell the LLM to pretend to be them
@aeva I want to imagine a sad robot screaming into the void, its only contact being another, dumber, robot.
@aeva If not go square hole, why square hole fit?
@aeva asserting dominance by forcing the AI to compliment my ungodly shitcode