I agree with you in principle but I am seeing my colleague who supposedly is an experienced engineer just sends out 40-file PRs without going through the code. When someone asked why the code does a certain thing he used LLM to come up with an answer. Apparently he has no idea what the code is doing and I doubt he reviews any of it or cares.

▲

phoghed 22 minutes ago | parent | next [-]

Yes, welcome to the new world. This is how things are now. Every couple months the models and tools get better, while everyone keeps assuring themselves that any day now they’re going to run into the wall. Whether we like it or not, this is how it is. Let go and let Claude.

▲

bakugo 20 minutes ago | parent | prev [-]

I don't understand what the value proposition of such an employee is. If all you're doing is acting as a meat proxy between your manager and your AI, why does your role even exist? Why shouldn't your superiors just prompt the AI directly?

	▲	cosmic_cheese 4 minutes ago \| parent \| next [-]
		I agree. Even in the era of LLMs, part of one’s job as a software developer is to understand the codebase and how it works and be able to fix bugs and add features without breaking other things in the process or comprising quality or security. If you’re not doing that and just blindly committing whatever code the LLM generates without so much as glancing at it, you’re not adding any value and there’s a good chance that with a little initial guidance, the janitor could perform your prompt-jockeying about as well.
	▲	fg137 5 minutes ago \| parent \| prev [-]
		The management cares more about "this dude ships things" than "this dude ships things that are maintainable". Whether someone else will be left cleaning up the mess (I have been in that position) is a separate question.