What a company considers to be "performance" or "insubordination" can certainly still be about office politics. Not that I'm saying this is necessarily the case here, but they're not a hard either/or.

For instance: 996 is a form of office politics. Return to office in general a form of office politics. Pay secrecy is a form of office politics. Some of those things become performance metrics, or evaluations of your "attitude."

"Performance" can mean "the quality of your work." It also can mean "theater." Like when you're measuring performance by token usage, or lines of code, or office attendance... or, in a case I once encountered, manual adherence to a style guide, that required two human beings to check every line of code against a set of conventions, instead of just configuring a formatting tool and being done with it.

Regarding LLMs in particular, I don't necessarily think the absolutist stance is it here, but it is a technology worker's job to know when you can and should automate things, as well as to know when it's not an effective use of resources to throw technology at something just because. I don't know parent's company or use case. I don't know if they were doing something with it that actually enhanced productivity, or just tracking tokens. I don't know if it enhanced their product, or just shoehorned AI into it for investor-driven buzzword utilization. So I'm also not an absolutist on what you're saying.

It could be that parent is too rigid, but (as much as I'm not asking for all this context either) I don't think there's enough information here to be certain about that.